summaryrefslogtreecommitdiffstats
path: root/src/plugins/nat/nat64
diff options
context:
space:
mode:
Diffstat (limited to 'src/plugins/nat/nat64')
-rw-r--r--src/plugins/nat/nat64/nat64.api316
-rw-r--r--src/plugins/nat/nat64/nat64.c1646
-rw-r--r--src/plugins/nat/nat64/nat64.h531
-rw-r--r--src/plugins/nat/nat64/nat64_api.c458
-rw-r--r--src/plugins/nat/nat64/nat64_cli.c992
-rw-r--r--src/plugins/nat/nat64/nat64_db.c742
-rw-r--r--src/plugins/nat/nat64/nat64_db.h380
-rw-r--r--src/plugins/nat/nat64/nat64_doc.md73
-rw-r--r--src/plugins/nat/nat64/nat64_in2out.c1411
-rw-r--r--src/plugins/nat/nat64/nat64_out2in.c796
10 files changed, 7345 insertions, 0 deletions
diff --git a/src/plugins/nat/nat64/nat64.api b/src/plugins/nat/nat64/nat64.api
new file mode 100644
index 00000000000..5fc4129d041
--- /dev/null
+++ b/src/plugins/nat/nat64/nat64.api
@@ -0,0 +1,316 @@
+/*
+ * Copyright (c) 2020 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+option version = "1.0.0";
+import "vnet/ip/ip_types.api";
+import "vnet/interface_types.api";
+import "plugins/nat/nat_types.api";
+
+/**
+ * @file nat64.api
+ * @brief VPP control-plane API messages.
+ *
+ * This file defines VPP control-plane API messages which are generally
+ * called through a shared memory interface.
+ */
+
+/** \brief Enable/disable NAT64 plugin
+ @param client_index - opaque cookie to identify the sender
+ @param context - sender context, to match reply w/ request
+ @param bib_buckets - Number of BIB hash buckets
+ @param bib_memory_size - Memory size of BIB hash
+ @param st_buckets - Number of session table hash buckets
+ @param st_memory_size - Memory size of session table hash
+ @param enable - true if enable, false if disable
+*/
+autoreply define nat64_plugin_enable_disable {
+ u32 client_index;
+ u32 context;
+ u32 bib_buckets;
+ u32 bib_memory_size;
+ u32 st_buckets;
+ u32 st_memory_size;
+ bool enable;
+ option status="in_progress";
+};
+
+/** \brief Set values of timeouts for NAT64 sessions (seconds)
+ @param client_index - opaque cookie to identify the sender
+ @param context - sender context, to match reply w/ request
+ @param udp - UDP timeout (default 300sec)
+ @param tcp_established - TCP established timeout (default 7440sec)
+ @param tcp_transitory - TCP transitory timeout (default 240sec)
+ @param icmp - ICMP timeout (default 60sec)
+*/
+autoreply define nat64_set_timeouts {
+ u32 client_index;
+ u32 context;
+ u32 udp;
+ u32 tcp_established;
+ u32 tcp_transitory;
+ u32 icmp;
+};
+
+/** \brief Get values of timeouts for NAT64 sessions (seconds)
+ @param client_index - opaque cookie to identify the sender
+ @param context - sender context, to match reply w/ request
+*/
+define nat64_get_timeouts {
+ u32 client_index;
+ u32 context;
+};
+
+/** \brief Get values of timeouts for NAT64 sessions reply
+ @param context - sender context, to match reply w/ request
+ @param retval - return code
+ @param udp - UDP timeout
+ @param tcp_established - TCP established timeout
+ @param tcp_transitory - TCP transitory timeout
+ @param icmp - ICMP timeout
+*/
+define nat64_get_timeouts_reply {
+ u32 context;
+ i32 retval;
+ u32 udp;
+ u32 tcp_established;
+ u32 tcp_transitory;
+ u32 icmp;
+};
+
+/** \brief Add/delete address range to NAT64 pool
+ @param client_index - opaque cookie to identify the sender
+ @param context - sender context, to match reply w/ request
+ @param start_addr - start IPv4 address of the range
+ @param end_addr - end IPv4 address of the range
+ @param vrf_id - VRF id of tenant, ~0 means independent of VRF
+ @param is_add - true if add, false if delete
+*/
+autoreply define nat64_add_del_pool_addr_range {
+ u32 client_index;
+ u32 context;
+ vl_api_ip4_address_t start_addr;
+ vl_api_ip4_address_t end_addr;
+ u32 vrf_id;
+ bool is_add;
+};
+
+/** \brief Dump NAT64 pool addresses
+ @param client_index - opaque cookie to identify the sender
+ @param context - sender context, to match reply w/ request
+*/
+define nat64_pool_addr_dump {
+ u32 client_index;
+ u32 context;
+};
+
+/** \brief NAT64 pool address details response
+ @param context - sender context, to match reply w/ request
+ @param address - IPv4 address
+ @param vfr_id - VRF id of tenant, ~0 means independent of VRF
+*/
+define nat64_pool_addr_details {
+ u32 context;
+ vl_api_ip4_address_t address;
+ u32 vrf_id;
+};
+
+/** \brief Enable/disable NAT64 feature on the interface
+ @param client_index - opaque cookie to identify the sender
+ @param context - sender context, to match reply w/ request
+ @param is_add - true if add, false if delete
+ @param flags - flag NAT_IS_INSIDE if interface is inside else
+ interface is outside
+ @param sw_if_index - index of the interface
+*/
+autoreply define nat64_add_del_interface {
+ u32 client_index;
+ u32 context;
+ bool is_add;
+ vl_api_nat_config_flags_t flags;
+ vl_api_interface_index_t sw_if_index;
+};
+
+/** \brief Dump interfaces with NAT64 feature
+ @param client_index - opaque cookie to identify the sender
+ @param context - sender context, to match reply w/ request
+*/
+define nat64_interface_dump {
+ u32 client_index;
+ u32 context;
+};
+
+/** \brief NAT64 interface details response
+ @param context - sender context, to match reply w/ request
+ @param flags - flag NAT_IS_INSIDE if interface is inside,
+ flag NAT_IS_OUTSIDE if interface is outside
+ and if both flags are set the interface is
+ both inside and outside
+ @param sw_if_index - index of the interface
+*/
+define nat64_interface_details {
+ u32 context;
+ vl_api_nat_config_flags_t flags;
+ vl_api_interface_index_t sw_if_index;
+};
+
+/** \brief Add/delete NAT64 static BIB entry
+ @param client_index - opaque cookie to identify the sender
+ @param context - sender context, to match reply w/ request
+ @param i_addr - inside IPv6 address
+ @param o_addr - outside IPv4 address
+ @param i_port - inside port number
+ @param o_port - outside port number
+ @param vrf_id - VRF id of tenant
+ @param proto - protocol number
+ @param is_add - true if add, false if delete
+*/
+ autoreply define nat64_add_del_static_bib {
+ u32 client_index;
+ u32 context;
+ vl_api_ip6_address_t i_addr;
+ vl_api_ip4_address_t o_addr;
+ u16 i_port;
+ u16 o_port;
+ u32 vrf_id;
+ u8 proto;
+ bool is_add;
+};
+
+/** \brief Dump NAT64 BIB
+ @param client_index - opaque cookie to identify the sender
+ @param context - sender context, to match reply w/ request
+ @param proto - protocol of the BIB: 255 - all BIBs
+ 6 - TCP BIB
+ 17 - UDP BIB
+ 1/58 - ICMP BIB
+ otherwise - "unknown" protocol BIB
+*/
+define nat64_bib_dump {
+ u32 client_index;
+ u32 context;
+ u8 proto;
+};
+
+/** \brief NAT64 BIB details response
+ @param context - sender context, to match reply w/ request
+ @param i_addr - inside IPv6 address
+ @param o_addr - outside IPv4 address
+ @param i_port - inside port number
+ @param o_port - outside port number
+ @param vrf_id - VRF id of tenant
+ @param proto - protocol number
+ @param flags - flag NAT_IS_STATIC if BIB entry is static
+ or BIB entry is dynamic
+ @param ses_num - number of sessions associated with the BIB entry
+*/
+define nat64_bib_details {
+ u32 context;
+ vl_api_ip6_address_t i_addr;
+ vl_api_ip4_address_t o_addr;
+ u16 i_port;
+ u16 o_port;
+ u32 vrf_id;
+ u8 proto;
+ vl_api_nat_config_flags_t flags;
+ u32 ses_num;
+};
+
+/** \brief Dump NAT64 session table
+ @param client_index - opaque cookie to identify the sender
+ @param context - sender context, to match reply w/ request
+ @param proto - protocol of the session table: 255 - all STs
+ 6 - TCP ST
+ 17 - UDP ST
+ 1/58 - ICMP ST
+ otherwise - "unknown" proto ST
+*/
+define nat64_st_dump {
+ u32 client_index;
+ u32 context;
+ u8 proto;
+};
+
+/** \brief NAT64 session table details response
+ @param context - sender context, to match reply w/ request
+ @param il_addr - inside IPv6 address of the local host
+ @param ol_addr - outside IPv4 address of the local host
+ @param il_port - inside port number id of the local host/inside ICMP id
+ @param ol_port - outside port number of the local host/outside ICMP id
+ @param ir_addr - inside IPv6 address of the remote host
+ @param or_addr - outside IPv4 address of the remote host
+ @param r_port - port number of the remote host (not used for ICMP)
+ @param vrf_id - VRF id of tenant
+ @param proto - protocol number
+*/
+define nat64_st_details {
+ u32 context;
+ vl_api_ip6_address_t il_addr;
+ vl_api_ip4_address_t ol_addr;
+ u16 il_port;
+ u16 ol_port;
+ vl_api_ip6_address_t ir_addr;
+ vl_api_ip4_address_t or_addr;
+ u16 r_port;
+ u32 vrf_id;
+ u8 proto;
+};
+
+/** \brief Add/del NAT64 prefix
+ @param client_index - opaque cookie to identify the sender
+ @param context - sender context, to match reply w/ request
+ @param prefix - NAT64 prefix
+ @param vrf_id - VRF id of tenant
+ @param is_add - true if add, false if delete
+*/
+autoreply define nat64_add_del_prefix {
+ u32 client_index;
+ u32 context;
+ vl_api_ip6_prefix_t prefix;
+ u32 vrf_id;
+ bool is_add;
+};
+
+/** \brief Dump NAT64 prefix
+ @param client_index - opaque cookie to identify the sender
+ @param context - sender context, to match reply w/ request
+*/
+define nat64_prefix_dump {
+ u32 client_index;
+ u32 context;
+};
+
+/** \brief Dump NAT64 prefix details response
+ @param context - sender context, to match reply w/ request
+ @param prefix - NAT64 prefix
+ @param vrf_id - VRF id of tenant
+*/
+define nat64_prefix_details {
+ u32 context;
+ vl_api_ip6_prefix_t prefix;
+ u32 vrf_id;
+};
+
+/** \brief Add/delete NAT64 pool address from specific interfce
+ @param client_index - opaque cookie to identify the sender
+ @param context - sender context, to match reply w/ request
+ @param is_add - true if add, false if delete
+ @param sw_if_index - software index of the interface
+*/
+autoreply define nat64_add_del_interface_addr {
+ u32 client_index;
+ u32 context;
+ bool is_add;
+ vl_api_interface_index_t sw_if_index;
+};
diff --git a/src/plugins/nat/nat64/nat64.c b/src/plugins/nat/nat64/nat64.c
new file mode 100644
index 00000000000..5da498670f6
--- /dev/null
+++ b/src/plugins/nat/nat64/nat64.c
@@ -0,0 +1,1646 @@
+/*
+ * Copyright (c) 2020 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vppinfra/crc32.h>
+#include <vnet/fib/ip4_fib.h>
+
+#include <vnet/ip/reass/ip4_sv_reass.h>
+#include <vnet/ip/reass/ip6_sv_reass.h>
+#include <vnet/plugin/plugin.h>
+#include <vpp/app/version.h>
+
+#include <nat/nat64/nat64.h>
+
+nat64_main_t nat64_main;
+
+/* *INDENT-OFF* */
+/* Hook up input features */
+VNET_FEATURE_INIT (nat64_in2out, static) = {
+ .arc_name = "ip6-unicast",
+ .node_name = "nat64-in2out",
+ .runs_before = VNET_FEATURES ("ip6-lookup"),
+ .runs_after = VNET_FEATURES ("ip6-sv-reassembly-feature"),
+};
+VNET_FEATURE_INIT (nat64_out2in, static) = {
+ .arc_name = "ip4-unicast",
+ .node_name = "nat64-out2in",
+ .runs_before = VNET_FEATURES ("ip4-lookup"),
+ .runs_after = VNET_FEATURES ("ip4-sv-reassembly-feature"),
+};
+VNET_FEATURE_INIT (nat64_in2out_handoff, static) = {
+ .arc_name = "ip6-unicast",
+ .node_name = "nat64-in2out-handoff",
+ .runs_before = VNET_FEATURES ("ip6-lookup"),
+ .runs_after = VNET_FEATURES ("ip6-sv-reassembly-feature"),
+};
+VNET_FEATURE_INIT (nat64_out2in_handoff, static) = {
+ .arc_name = "ip4-unicast",
+ .node_name = "nat64-out2in-handoff",
+ .runs_before = VNET_FEATURES ("ip4-lookup"),
+ .runs_after = VNET_FEATURES ("ip4-sv-reassembly-feature"),
+};
+VLIB_PLUGIN_REGISTER () = {
+ .version = VPP_BUILD_VER,
+ .description = "NAT64",
+};
+static u8 well_known_prefix[] = {
+ 0x00, 0x64, 0xff, 0x9b,
+ 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00,
+ 0x00, 0x00, 0x00, 0x00
+};
+/* *INDENT-ON* */
+
+#define nat_elog_str(_str) \
+do \
+ { \
+ ELOG_TYPE_DECLARE (e) = \
+ { \
+ .format = "nat-msg " _str, \
+ .format_args = "", \
+ }; \
+ ELOG_DATA (&vlib_global_main.elog_main, e); \
+ } while (0);
+
+static void
+nat64_ip4_add_del_interface_address_cb (ip4_main_t * im, uword opaque,
+ u32 sw_if_index,
+ ip4_address_t * address,
+ u32 address_length,
+ u32 if_address_index, u32 is_delete)
+{
+ nat64_main_t *nm = &nat64_main;
+ int i, j;
+
+ if (plugin_enabled () == 0)
+ return;
+
+ for (i = 0; i < vec_len (nm->auto_add_sw_if_indices); i++)
+ {
+ if (sw_if_index == nm->auto_add_sw_if_indices[i])
+ {
+ if (!is_delete)
+ {
+ /* Don't trip over lease renewal, static config */
+ for (j = 0; j < vec_len (nm->addr_pool); j++)
+ if (nm->addr_pool[j].addr.as_u32 == address->as_u32)
+ return;
+
+ (void) nat64_add_del_pool_addr (vlib_get_thread_index (),
+ address, ~0, 1);
+ return;
+ }
+ else
+ {
+ (void) nat64_add_del_pool_addr (vlib_get_thread_index (),
+ address, ~0, 0);
+ return;
+ }
+ }
+ }
+}
+
+u32
+nat64_get_worker_in2out (ip6_address_t * addr)
+{
+ nat64_main_t *nm = &nat64_main;
+ u32 next_worker_index = nm->first_worker_index;
+ u32 hash;
+
+#ifdef clib_crc32c_uses_intrinsics
+ hash = clib_crc32c ((u8 *) addr->as_u32, 16);
+#else
+ u64 tmp = addr->as_u64[0] ^ addr->as_u64[1];
+ hash = clib_xxhash (tmp);
+#endif
+
+ if (PREDICT_TRUE (is_pow2 (_vec_len (nm->workers))))
+ next_worker_index += nm->workers[hash & (_vec_len (nm->workers) - 1)];
+ else
+ next_worker_index += nm->workers[hash % _vec_len (nm->workers)];
+
+ return next_worker_index;
+}
+
+u32
+nat64_get_worker_out2in (vlib_buffer_t * b, ip4_header_t * ip)
+{
+ nat64_main_t *nm = &nat64_main;
+ udp_header_t *udp;
+ u16 port;
+ u32 proto;
+
+ proto = ip_proto_to_nat_proto (ip->protocol);
+ udp = ip4_next_header (ip);
+ port = udp->dst_port;
+
+ /* unknown protocol */
+ if (PREDICT_FALSE (proto == NAT_PROTOCOL_OTHER))
+ {
+ nat64_db_t *db;
+ ip46_address_t daddr;
+ nat64_db_bib_entry_t *bibe;
+
+ clib_memset (&daddr, 0, sizeof (daddr));
+ daddr.ip4.as_u32 = ip->dst_address.as_u32;
+
+ /* *INDENT-OFF* */
+ vec_foreach (db, nm->db)
+ {
+ bibe = nat64_db_bib_entry_find (db, &daddr, 0, ip->protocol, 0, 0);
+ if (bibe)
+ return (u32) (db - nm->db);
+ }
+ /* *INDENT-ON* */
+ return vlib_get_thread_index ();
+ }
+
+ /* ICMP */
+ if (PREDICT_FALSE (ip->protocol == IP_PROTOCOL_ICMP))
+ {
+ icmp46_header_t *icmp = (icmp46_header_t *) udp;
+ icmp_echo_header_t *echo = (icmp_echo_header_t *) (icmp + 1);
+ if (!icmp_type_is_error_message
+ (vnet_buffer (b)->ip.reass.icmp_type_or_tcp_flags))
+ port = vnet_buffer (b)->ip.reass.l4_src_port;
+ else
+ {
+ /* if error message, then it's not fragmented and we can access it */
+ ip4_header_t *inner_ip = (ip4_header_t *) (echo + 1);
+ proto = ip_proto_to_nat_proto (inner_ip->protocol);
+ void *l4_header = ip4_next_header (inner_ip);
+ switch (proto)
+ {
+ case NAT_PROTOCOL_ICMP:
+ icmp = (icmp46_header_t *) l4_header;
+ echo = (icmp_echo_header_t *) (icmp + 1);
+ port = echo->identifier;
+ break;
+ case NAT_PROTOCOL_UDP:
+ case NAT_PROTOCOL_TCP:
+ port = ((tcp_udp_header_t *) l4_header)->src_port;
+ break;
+ default:
+ return vlib_get_thread_index ();
+ }
+ }
+ }
+
+ /* worker by outside port (TCP/UDP) */
+ port = clib_net_to_host_u16 (port);
+ if (port > 1024)
+ return nm->first_worker_index + ((port - 1024) / nm->port_per_thread);
+
+ return vlib_get_thread_index ();
+}
+
+clib_error_t *
+nat64_init (vlib_main_t * vm)
+{
+ nat64_main_t *nm = &nat64_main;
+ vlib_thread_main_t *tm = vlib_get_thread_main ();
+ ip4_add_del_interface_address_callback_t cb4;
+ vlib_node_t *node;
+
+ clib_memset (nm, 0, sizeof (*nm));
+
+ nm->ip4_main = &ip4_main;
+ nm->log_class = vlib_log_register_class ("nat64", 0);
+
+ nm->port_per_thread = 0xffff - 1024;
+
+ nm->fq_in2out_index = ~0;
+ nm->fq_out2in_index = ~0;
+
+ node = vlib_get_node_by_name (vm, (u8 *) "error-drop");
+ nm->error_node_index = node->index;
+ node = vlib_get_node_by_name (vm, (u8 *) "nat64-in2out");
+ nm->in2out_node_index = node->index;
+ node = vlib_get_node_by_name (vm, (u8 *) "nat64-in2out-slowpath");
+ nm->in2out_slowpath_node_index = node->index;
+ node = vlib_get_node_by_name (vm, (u8 *) "nat64-out2in");
+ nm->out2in_node_index = node->index;
+
+ node = vlib_get_node_by_name (vm, (u8 *) "nat64-expire-worker-walk");
+ nm->expire_worker_walk_node_index = node->index;
+
+ nm->fib_src_hi = fib_source_allocate ("nat64-hi",
+ FIB_SOURCE_PRIORITY_HI,
+ FIB_SOURCE_BH_SIMPLE);
+ nm->fib_src_low = fib_source_allocate ("nat64-low",
+ FIB_SOURCE_PRIORITY_LOW,
+ FIB_SOURCE_BH_SIMPLE);
+
+ // set protocol timeouts to defaults
+ nat64_reset_timeouts ();
+
+ /* Set up the interface address add/del callback */
+ cb4.function = nat64_ip4_add_del_interface_address_cb;
+ cb4.function_opaque = 0;
+ vec_add1 (nm->ip4_main->add_del_interface_address_callbacks, cb4);
+
+ /* Init counters */
+ nm->total_bibs.name = "total-bibs";
+ nm->total_bibs.stat_segment_name = "/nat64/total-bibs";
+ vlib_validate_simple_counter (&nm->total_bibs, 0);
+ vlib_zero_simple_counter (&nm->total_bibs, 0);
+ nm->total_sessions.name = "total-sessions";
+ nm->total_sessions.stat_segment_name = "/nat64/total-sessions";
+ vlib_validate_simple_counter (&nm->total_sessions, 0);
+ vlib_zero_simple_counter (&nm->total_sessions, 0);
+
+ uword *p = hash_get_mem (tm->thread_registrations_by_name, "workers");
+ if (p)
+ {
+ vlib_thread_registration_t *tr;
+ tr = (vlib_thread_registration_t *) p[0];
+ if (tr)
+ {
+ nm->num_workers = tr->count;
+ nm->first_worker_index = tr->first_index;
+ }
+ }
+
+ if (nm->num_workers > 1)
+ {
+ int i;
+ uword *bitmap = 0;
+
+ for (i = 0; i < nm->num_workers; i++)
+ bitmap = clib_bitmap_set (bitmap, i, 1);
+
+ /* *INDENT-OFF* */
+ clib_bitmap_foreach (i, bitmap,
+ ({
+ vec_add1(nm->workers, i);
+ }));
+ /* *INDENT-ON* */
+
+ clib_bitmap_free (bitmap);
+
+ nm->port_per_thread = (0xffff - 1024) / _vec_len (nm->workers);
+ }
+
+ // TODO: ipfix needs to be separated from NAT base plugin
+ /* Init IPFIX logging */
+ //snat_ipfix_logging_init (vm);
+
+#define _(x) \
+ nm->counters.in2out.x.name = #x; \
+ nm->counters.in2out.x.stat_segment_name = "/nat64/in2out/" #x; \
+ nm->counters.out2in.x.name = #x; \
+ nm->counters.out2in.x.stat_segment_name = "/nat64/out2in/" #x;
+ foreach_nat_counter;
+#undef _
+ return nat64_api_hookup (vm);
+}
+
+VLIB_INIT_FUNCTION (nat64_init);
+
+static void nat64_free_out_addr_and_port (struct nat64_db_s *db,
+ ip4_address_t * addr, u16 port,
+ u8 protocol);
+
+int
+nat64_init_hash (nat64_config_t c)
+{
+ vlib_thread_main_t *tm = vlib_get_thread_main ();
+ nat64_main_t *nm = &nat64_main;
+ nat64_db_t *db;
+ int rv = 0;
+
+ vec_validate (nm->db, tm->n_vlib_mains - 1);
+
+ /* *INDENT-OFF* */
+ vec_foreach (db, nm->db)
+ {
+ if (nat64_db_init (db, c, nat64_free_out_addr_and_port))
+ {
+ nat64_log_err ("NAT64 DB init failed");
+ rv = 1;
+ }
+ }
+ /* *INDENT-ON* */
+
+ return rv;
+}
+
+int
+nat64_free_hash ()
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_db_t *db;
+ int rv = 0;
+
+ /* *INDENT-OFF* */
+ vec_foreach (db, nm->db)
+ {
+ if (nat64_db_free (db))
+ {
+ nat64_log_err ("NAT64 DB free failed");
+ rv = 1;
+ }
+ }
+ /* *INDENT-ON* */
+
+ vec_free (nm->db);
+
+ return rv;
+}
+
+int
+nat64_add_del_pool_addr (u32 thread_index,
+ ip4_address_t * addr, u32 vrf_id, u8 is_add)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_address_t *a = 0;
+ nat64_interface_t *interface;
+ int i;
+ nat64_db_t *db;
+ vlib_thread_main_t *tm = vlib_get_thread_main ();
+
+ /* Check if address already exists */
+ for (i = 0; i < vec_len (nm->addr_pool); i++)
+ {
+ if (nm->addr_pool[i].addr.as_u32 == addr->as_u32)
+ {
+ a = nm->addr_pool + i;
+ break;
+ }
+ }
+
+ if (is_add)
+ {
+ if (a)
+ return VNET_API_ERROR_VALUE_EXIST;
+
+ vec_add2 (nm->addr_pool, a, 1);
+ a->addr = *addr;
+ a->fib_index = ~0;
+ if (vrf_id != ~0)
+ a->fib_index =
+ fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, vrf_id,
+ nm->fib_src_hi);
+#define _(N, id, n, s) \
+ clib_memset (a->busy_##n##_port_refcounts, 0, sizeof(a->busy_##n##_port_refcounts)); \
+ a->busy_##n##_ports = 0; \
+ vec_validate_init_empty (a->busy_##n##_ports_per_thread, tm->n_vlib_mains - 1, 0);
+ foreach_nat_protocol
+#undef _
+ }
+ else
+ {
+ if (!a)
+ return VNET_API_ERROR_NO_SUCH_ENTRY;
+
+ if (a->fib_index != ~0)
+ fib_table_unlock (a->fib_index, FIB_PROTOCOL_IP6, nm->fib_src_hi);
+ /* Delete sessions using address */
+ /* *INDENT-OFF* */
+ vec_foreach (db, nm->db)
+ {
+ nat64_db_free_out_addr (thread_index, db, &a->addr);
+ vlib_set_simple_counter (&nm->total_bibs, db - nm->db, 0,
+ db->bib.bib_entries_num);
+ vlib_set_simple_counter (&nm->total_sessions, db - nm->db, 0,
+ db->st.st_entries_num);
+ }
+ /* *INDENT-ON* */
+ vec_del1 (nm->addr_pool, i);
+ }
+
+ /* Add/del external address to FIB */
+ /* *INDENT-OFF* */
+ pool_foreach (interface, nm->interfaces,
+ ({
+ if (nat64_interface_is_inside(interface))
+ continue;
+
+ nat64_add_del_addr_to_fib (addr, 32, interface->sw_if_index, is_add);
+ break;
+ }));
+ /* *INDENT-ON* */
+
+ return 0;
+}
+
+void
+nat64_pool_addr_walk (nat64_pool_addr_walk_fn_t fn, void *ctx)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_address_t *a = 0;
+
+ /* *INDENT-OFF* */
+ vec_foreach (a, nm->addr_pool)
+ {
+ if (fn (a, ctx))
+ break;
+ };
+ /* *INDENT-ON* */
+}
+
+int
+nat64_add_interface_address (u32 sw_if_index, int is_add)
+{
+ nat64_main_t *nm = &nat64_main;
+ ip4_main_t *ip4_main = nm->ip4_main;
+ ip4_address_t *first_int_addr;
+ int i;
+
+ first_int_addr = ip4_interface_first_address (ip4_main, sw_if_index, 0);
+
+ for (i = 0; i < vec_len (nm->auto_add_sw_if_indices); i++)
+ {
+ if (nm->auto_add_sw_if_indices[i] == sw_if_index)
+ {
+ if (is_add)
+ return VNET_API_ERROR_VALUE_EXIST;
+ else
+ {
+ /* if have address remove it */
+ if (first_int_addr)
+ (void) nat64_add_del_pool_addr (vlib_get_thread_index (),
+ first_int_addr, ~0, 0);
+ vec_del1 (nm->auto_add_sw_if_indices, i);
+ return 0;
+ }
+ }
+ }
+
+ if (!is_add)
+ return VNET_API_ERROR_NO_SUCH_ENTRY;
+
+ /* add to the auto-address list */
+ vec_add1 (nm->auto_add_sw_if_indices, sw_if_index);
+
+ /* If the address is already bound - or static - add it now */
+ if (first_int_addr)
+ (void) nat64_add_del_pool_addr (vlib_get_thread_index (),
+ first_int_addr, ~0, 1);
+
+ return 0;
+}
+
+static void
+nat64_validate_counters (nat64_main_t * nm, u32 sw_if_index)
+{
+#define _(x) \
+ vlib_validate_simple_counter (&nm->counters.in2out.x, sw_if_index); \
+ vlib_zero_simple_counter (&nm->counters.in2out.x, sw_if_index); \
+ vlib_validate_simple_counter (&nm->counters.out2in.x, sw_if_index); \
+ vlib_zero_simple_counter (&nm->counters.out2in.x, sw_if_index);
+ foreach_nat_counter;
+#undef _
+}
+
+void
+nat64_add_del_addr_to_fib (ip4_address_t * addr, u8 p_len, u32 sw_if_index,
+ int is_add)
+{
+ nat64_main_t *nm = &nat64_main;
+ fib_prefix_t prefix = {
+ .fp_len = p_len,
+ .fp_proto = FIB_PROTOCOL_IP4,
+ .fp_addr = {
+ .ip4.as_u32 = addr->as_u32,
+ },
+ };
+ u32 fib_index = ip4_fib_table_get_index_for_sw_if_index (sw_if_index);
+
+ if (is_add)
+ fib_table_entry_update_one_path (fib_index,
+ &prefix,
+ nm->fib_src_low,
+ (FIB_ENTRY_FLAG_CONNECTED |
+ FIB_ENTRY_FLAG_LOCAL |
+ FIB_ENTRY_FLAG_EXCLUSIVE),
+ DPO_PROTO_IP4,
+ NULL,
+ sw_if_index,
+ ~0, 1, NULL, FIB_ROUTE_PATH_FLAG_NONE);
+ else
+ fib_table_entry_delete (fib_index, &prefix, nm->fib_src_low);
+}
+
+int
+nat64_interface_add_del (u32 sw_if_index, u8 is_inside, u8 is_add)
+{
+ vlib_main_t *vm = vlib_get_main ();
+ nat64_main_t *nm = &nat64_main;
+ nat64_interface_t *interface = 0, *i;
+ nat64_address_t *ap;
+ const char *feature_name, *arc_name;
+
+ // TODO: is enabled ? we can't signal if it is not
+
+ /* Check if interface already exists */
+ /* *INDENT-OFF* */
+ pool_foreach (i, nm->interfaces,
+ ({
+ if (i->sw_if_index == sw_if_index)
+ {
+ interface = i;
+ break;
+ }
+ }));
+ /* *INDENT-ON* */
+
+ if (is_add)
+ {
+ if (interface)
+ goto set_flags;
+
+ pool_get (nm->interfaces, interface);
+ interface->sw_if_index = sw_if_index;
+ interface->flags = 0;
+ nat64_validate_counters (nm, sw_if_index);
+ set_flags:
+ if (is_inside)
+ interface->flags |= NAT64_INTERFACE_FLAG_IS_INSIDE;
+ else
+ interface->flags |= NAT64_INTERFACE_FLAG_IS_OUTSIDE;
+
+ nm->total_enabled_count++;
+ vlib_process_signal_event (vm,
+ nm->expire_walk_node_index,
+ NAT64_CLEANER_RESCHEDULE, 0);
+
+ }
+ else
+ {
+ if (!interface)
+ return VNET_API_ERROR_NO_SUCH_ENTRY;
+
+ if ((nat64_interface_is_inside (interface)
+ && nat64_interface_is_outside (interface)))
+ interface->flags &=
+ is_inside ? ~NAT64_INTERFACE_FLAG_IS_INSIDE :
+ ~NAT64_INTERFACE_FLAG_IS_OUTSIDE;
+ else
+ pool_put (nm->interfaces, interface);
+
+ nm->total_enabled_count--;
+ }
+
+ if (!is_inside)
+ {
+ /* *INDENT-OFF* */
+ vec_foreach (ap, nm->addr_pool)
+ nat64_add_del_addr_to_fib (&ap->addr, 32, sw_if_index, is_add);
+ /* *INDENT-ON* */
+ }
+
+ if (nm->num_workers > 1)
+ {
+ feature_name =
+ is_inside ? "nat64-in2out-handoff" : "nat64-out2in-handoff";
+ if (nm->fq_in2out_index == ~0)
+ nm->fq_in2out_index =
+ vlib_frame_queue_main_init (nat64_in2out_node.index, 0);
+ if (nm->fq_out2in_index == ~0)
+ nm->fq_out2in_index =
+ vlib_frame_queue_main_init (nat64_out2in_node.index, 0);
+ }
+ else
+ feature_name = is_inside ? "nat64-in2out" : "nat64-out2in";
+
+ arc_name = is_inside ? "ip6-unicast" : "ip4-unicast";
+
+ if (is_inside)
+ {
+ int rv = ip6_sv_reass_enable_disable_with_refcnt (sw_if_index, is_add);
+ if (rv)
+ return rv;
+ }
+ else
+ {
+ int rv = ip4_sv_reass_enable_disable_with_refcnt (sw_if_index, is_add);
+ if (rv)
+ return rv;
+ }
+
+ return vnet_feature_enable_disable (arc_name, feature_name, sw_if_index,
+ is_add, 0, 0);
+}
+
+void
+nat64_interfaces_walk (nat64_interface_walk_fn_t fn, void *ctx)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_interface_t *i = 0;
+
+ /* *INDENT-OFF* */
+ pool_foreach (i, nm->interfaces,
+ ({
+ if (fn (i, ctx))
+ break;
+ }));
+ /* *INDENT-ON* */
+}
+
+// TODO: plugin independent
+static_always_inline u16
+nat64_random_port (u16 min, u16 max)
+{
+ nat64_main_t *nm = &nat64_main;
+ u32 rwide;
+ u16 r;
+
+ rwide = random_u32 (&nm->random_seed);
+ r = rwide & 0xFFFF;
+ if (r >= min && r <= max)
+ return r;
+
+ return min + (rwide % (max - min + 1));
+}
+
+static_always_inline int
+nat64_alloc_addr_and_port_default (nat64_address_t * addresses,
+ u32 fib_index,
+ u32 thread_index,
+ nat_protocol_t proto,
+ ip4_address_t * addr,
+ u16 * port,
+ u16 port_per_thread, u32 nat_thread_index)
+{
+ int i;
+ nat64_address_t *a, *ga = 0;
+ u32 portnum;
+
+ for (i = 0; i < vec_len (addresses); i++)
+ {
+ a = addresses + i;
+ switch (proto)
+ {
+#define _(N, j, n, s) \
+ case NAT_PROTOCOL_##N: \
+ if (a->busy_##n##_ports_per_thread[thread_index] < port_per_thread) \
+ { \
+ if (a->fib_index == fib_index) \
+ { \
+ while (1) \
+ { \
+ portnum = (port_per_thread * \
+ nat_thread_index) + \
+ nat64_random_port(0, port_per_thread - 1) + 1024; \
+ if (a->busy_##n##_port_refcounts[portnum]) \
+ continue; \
+ --a->busy_##n##_port_refcounts[portnum]; \
+ a->busy_##n##_ports_per_thread[thread_index]++; \
+ a->busy_##n##_ports++; \
+ *addr = a->addr; \
+ *port = clib_host_to_net_u16(portnum); \
+ return 0; \
+ } \
+ } \
+ else if (a->fib_index == ~0) \
+ { \
+ ga = a; \
+ } \
+ } \
+ break;
+ foreach_nat_protocol
+#undef _
+ default:
+ return 1;
+ }
+
+ }
+
+ if (ga)
+ {
+ a = ga;
+ switch (proto)
+ {
+#define _(N, j, n, s) \
+ case NAT_PROTOCOL_##N: \
+ while (1) \
+ { \
+ portnum = (port_per_thread * \
+ nat_thread_index) + \
+ nat64_random_port(0, port_per_thread - 1) + 1024; \
+ if (a->busy_##n##_port_refcounts[portnum]) \
+ continue; \
+ ++a->busy_##n##_port_refcounts[portnum]; \
+ a->busy_##n##_ports_per_thread[thread_index]++; \
+ a->busy_##n##_ports++; \
+ *addr = a->addr; \
+ *port = clib_host_to_net_u16(portnum); \
+ return 0; \
+ }
+ break;
+ foreach_nat_protocol
+#undef _
+ default:
+ return 1;
+ }
+ }
+
+ /* Totally out of translations to use... */
+ //snat_ipfix_logging_addresses_exhausted (thread_index, 0);
+ return 1;
+}
+
+int
+nat64_alloc_out_addr_and_port (u32 fib_index, nat_protocol_t proto,
+ ip4_address_t * addr, u16 * port,
+ u32 thread_index)
+{
+ nat64_main_t *nm = &nat64_main;
+ u32 worker_index = 0;
+ int rv;
+
+ if (nm->num_workers > 1)
+ worker_index = thread_index - nm->first_worker_index;
+
+ rv = nat64_alloc_addr_and_port_default (nm->addr_pool, fib_index,
+ thread_index,
+ proto, addr, port,
+ nm->port_per_thread, worker_index);
+
+ return rv;
+}
+
+static void
+nat64_free_out_addr_and_port (struct nat64_db_s *db, ip4_address_t * addr,
+ u16 port, u8 protocol)
+{
+ nat64_main_t *nm = &nat64_main;
+ u32 thread_index = db - nm->db;
+ nat_protocol_t proto = ip_proto_to_nat_proto (protocol);
+ u16 port_host_byte_order = clib_net_to_host_u16 (port);
+ nat64_address_t *a;
+ int i;
+
+ for (i = 0; i < vec_len (nm->addr_pool); i++)
+ {
+ a = nm->addr_pool + i;
+ if (addr->as_u32 != a->addr.as_u32)
+ continue;
+ switch (proto)
+ {
+#define _(N, j, n, s) \
+ case NAT_PROTOCOL_##N: \
+ ASSERT (a->busy_##n##_port_refcounts[port_host_byte_order] >= 1); \
+ --a->busy_##n##_port_refcounts[port_host_byte_order]; \
+ a->busy_##n##_ports--; \
+ a->busy_##n##_ports_per_thread[thread_index]--; \
+ break;
+ foreach_nat_protocol
+#undef _
+ default:
+ nat_elog_str ("unknown protocol");
+ return;
+ }
+ break;
+ }
+}
+
+/**
+ * @brief Add/delete static BIB entry in worker thread.
+ */
+static uword
+nat64_static_bib_worker_fn (vlib_main_t * vm, vlib_node_runtime_t * rt,
+ vlib_frame_t * f)
+{
+ nat64_main_t *nm = &nat64_main;
+ u32 thread_index = vm->thread_index;
+ nat64_db_t *db = &nm->db[thread_index];
+ nat64_static_bib_to_update_t *static_bib;
+ nat64_db_bib_entry_t *bibe;
+ ip46_address_t addr;
+
+ /* *INDENT-OFF* */
+ pool_foreach (static_bib, nm->static_bibs,
+ ({
+ if ((static_bib->thread_index != thread_index) || (static_bib->done))
+ continue;
+
+ if (static_bib->is_add)
+ {
+ (void) nat64_db_bib_entry_create (thread_index, db,
+ &static_bib->in_addr,
+ &static_bib->out_addr,
+ static_bib->in_port,
+ static_bib->out_port,
+ static_bib->fib_index,
+ static_bib->proto, 1);
+ vlib_set_simple_counter (&nm->total_bibs, thread_index, 0,
+ db->bib.bib_entries_num);
+ }
+ else
+ {
+ addr.as_u64[0] = static_bib->in_addr.as_u64[0];
+ addr.as_u64[1] = static_bib->in_addr.as_u64[1];
+ bibe = nat64_db_bib_entry_find (db, &addr, static_bib->in_port,
+ static_bib->proto,
+ static_bib->fib_index, 1);
+ if (bibe)
+ {
+ nat64_db_bib_entry_free (thread_index, db, bibe);
+ vlib_set_simple_counter (&nm->total_bibs, thread_index, 0,
+ db->bib.bib_entries_num);
+ vlib_set_simple_counter (&nm->total_sessions, thread_index, 0,
+ db->st.st_entries_num);
+ }
+ }
+
+ static_bib->done = 1;
+ }));
+ /* *INDENT-ON* */
+
+ return 0;
+}
+
+static vlib_node_registration_t nat64_static_bib_worker_node;
+
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE (nat64_static_bib_worker_node, static) = {
+ .function = nat64_static_bib_worker_fn,
+ .type = VLIB_NODE_TYPE_INPUT,
+ .state = VLIB_NODE_STATE_INTERRUPT,
+ .name = "nat64-static-bib-worker",
+};
+/* *INDENT-ON* */
+
+int
+nat64_add_del_static_bib_entry (ip6_address_t * in_addr,
+ ip4_address_t * out_addr, u16 in_port,
+ u16 out_port, u8 proto, u32 vrf_id, u8 is_add)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_db_bib_entry_t *bibe;
+ u32 fib_index = fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, vrf_id,
+ nm->fib_src_hi);
+ nat_protocol_t p = ip_proto_to_nat_proto (proto);
+ ip46_address_t addr;
+ int i;
+ nat64_address_t *a;
+ u32 thread_index = 0;
+ nat64_db_t *db;
+ nat64_static_bib_to_update_t *static_bib;
+ vlib_main_t *worker_vm;
+ u32 *to_be_free = 0, *index;
+
+ if (nm->num_workers > 1)
+ {
+ thread_index = nat64_get_worker_in2out (in_addr);
+ db = &nm->db[thread_index];
+ }
+ else
+ db = &nm->db[nm->num_workers];
+
+ addr.as_u64[0] = in_addr->as_u64[0];
+ addr.as_u64[1] = in_addr->as_u64[1];
+ bibe =
+ nat64_db_bib_entry_find (db, &addr, clib_host_to_net_u16 (in_port),
+ proto, fib_index, 1);
+
+ if (is_add)
+ {
+ if (bibe)
+ return VNET_API_ERROR_VALUE_EXIST;
+
+ /* outside port must be assigned to same thread as internall address */
+ if ((out_port > 1024) && (nm->num_workers > 1))
+ {
+ if (thread_index != ((out_port - 1024) / nm->port_per_thread))
+ return VNET_API_ERROR_INVALID_VALUE_2;
+ }
+
+ for (i = 0; i < vec_len (nm->addr_pool); i++)
+ {
+ a = nm->addr_pool + i;
+ if (out_addr->as_u32 != a->addr.as_u32)
+ continue;
+ switch (p)
+ {
+#define _(N, j, n, s) \
+ case NAT_PROTOCOL_##N: \
+ if (a->busy_##n##_port_refcounts[out_port]) \
+ return VNET_API_ERROR_INVALID_VALUE; \
+ ++a->busy_##n##_port_refcounts[out_port]; \
+ if (out_port > 1024) \
+ { \
+ a->busy_##n##_ports++; \
+ a->busy_##n##_ports_per_thread[thread_index]++; \
+ } \
+ break;
+ foreach_nat_protocol
+#undef _
+ default:
+ clib_memset (&addr, 0, sizeof (addr));
+ addr.ip4.as_u32 = out_addr->as_u32;
+ if (nat64_db_bib_entry_find (db, &addr, 0, proto, fib_index, 0))
+ return VNET_API_ERROR_INVALID_VALUE;
+ }
+ break;
+ }
+ if (!nm->num_workers)
+ {
+ bibe =
+ nat64_db_bib_entry_create (thread_index, db, in_addr, out_addr,
+ clib_host_to_net_u16 (in_port),
+ clib_host_to_net_u16 (out_port),
+ fib_index, proto, 1);
+ if (!bibe)
+ return VNET_API_ERROR_UNSPECIFIED;
+
+ vlib_set_simple_counter (&nm->total_bibs, thread_index, 0,
+ db->bib.bib_entries_num);
+ }
+ }
+ else
+ {
+ if (!bibe)
+ return VNET_API_ERROR_NO_SUCH_ENTRY;
+
+ if (!nm->num_workers)
+ {
+ nat64_db_bib_entry_free (thread_index, db, bibe);
+ vlib_set_simple_counter (&nm->total_bibs, thread_index, 0,
+ db->bib.bib_entries_num);
+ }
+ }
+
+ if (nm->num_workers)
+ {
+ /* *INDENT-OFF* */
+ pool_foreach (static_bib, nm->static_bibs,
+ ({
+ if (static_bib->done)
+ vec_add1 (to_be_free, static_bib - nm->static_bibs);
+ }));
+ vec_foreach (index, to_be_free)
+ pool_put_index (nm->static_bibs, index[0]);
+ /* *INDENT-ON* */
+ vec_free (to_be_free);
+ pool_get (nm->static_bibs, static_bib);
+ static_bib->in_addr.as_u64[0] = in_addr->as_u64[0];
+ static_bib->in_addr.as_u64[1] = in_addr->as_u64[1];
+ static_bib->in_port = clib_host_to_net_u16 (in_port);
+ static_bib->out_addr.as_u32 = out_addr->as_u32;
+ static_bib->out_port = clib_host_to_net_u16 (out_port);
+ static_bib->fib_index = fib_index;
+ static_bib->proto = proto;
+ static_bib->is_add = is_add;
+ static_bib->thread_index = thread_index;
+ static_bib->done = 0;
+ worker_vm = vlib_mains[thread_index];
+ if (worker_vm)
+ vlib_node_set_interrupt_pending (worker_vm,
+ nat64_static_bib_worker_node.index);
+ else
+ return VNET_API_ERROR_UNSPECIFIED;
+ }
+
+ return 0;
+}
+
+int
+nat64_set_udp_timeout (u32 timeout)
+{
+ nat64_main_t *nm = &nat64_main;
+
+ if (timeout == 0)
+ nm->udp_timeout = NAT_UDP_TIMEOUT;
+ else
+ nm->udp_timeout = timeout;
+
+ return 0;
+}
+
+u32
+nat64_get_udp_timeout (void)
+{
+ nat64_main_t *nm = &nat64_main;
+
+ return nm->udp_timeout;
+}
+
+int
+nat64_set_icmp_timeout (u32 timeout)
+{
+ nat64_main_t *nm = &nat64_main;
+
+ if (timeout == 0)
+ nm->icmp_timeout = NAT_ICMP_TIMEOUT;
+ else
+ nm->icmp_timeout = timeout;
+
+ return 0;
+}
+
+void
+nat64_reset_timeouts ()
+{
+ nat64_main_t *nm = &nat64_main;
+
+ nm->udp_timeout = NAT_UDP_TIMEOUT;
+ nm->icmp_timeout = NAT_ICMP_TIMEOUT;
+ nm->tcp_est_timeout = NAT_TCP_ESTABLISHED_TIMEOUT;
+ nm->tcp_trans_timeout = NAT_TCP_TRANSITORY_TIMEOUT;
+}
+
+u32
+nat64_get_icmp_timeout (void)
+{
+ nat64_main_t *nm = &nat64_main;
+
+ return nm->icmp_timeout;
+}
+
+int
+nat64_set_tcp_timeouts (u32 trans, u32 est)
+{
+ nat64_main_t *nm = &nat64_main;
+
+ if (trans == 0)
+ nm->tcp_trans_timeout = NAT_TCP_TRANSITORY_TIMEOUT;
+ else
+ nm->tcp_trans_timeout = trans;
+
+ if (est == 0)
+ nm->tcp_est_timeout = NAT_TCP_ESTABLISHED_TIMEOUT;
+ else
+ nm->tcp_est_timeout = est;
+
+ return 0;
+}
+
+u32
+nat64_get_tcp_trans_timeout (void)
+{
+ nat64_main_t *nm = &nat64_main;
+
+ return nm->tcp_trans_timeout;
+}
+
+u32
+nat64_get_tcp_est_timeout (void)
+{
+ nat64_main_t *nm = &nat64_main;
+
+ return nm->tcp_est_timeout;
+}
+
+void
+nat64_session_reset_timeout (nat64_db_st_entry_t * ste, vlib_main_t * vm)
+{
+ nat64_main_t *nm = &nat64_main;
+ u32 now = (u32) vlib_time_now (vm);
+
+ switch (ip_proto_to_nat_proto (ste->proto))
+ {
+ case NAT_PROTOCOL_ICMP:
+ ste->expire = now + nm->icmp_timeout;
+ return;
+ case NAT_PROTOCOL_TCP:
+ {
+ switch (ste->tcp_state)
+ {
+ case NAT64_TCP_STATE_V4_INIT:
+ case NAT64_TCP_STATE_V6_INIT:
+ case NAT64_TCP_STATE_V4_FIN_RCV:
+ case NAT64_TCP_STATE_V6_FIN_RCV:
+ case NAT64_TCP_STATE_V6_FIN_V4_FIN_RCV:
+ case NAT64_TCP_STATE_TRANS:
+ ste->expire = now + nm->tcp_trans_timeout;
+ return;
+ case NAT64_TCP_STATE_ESTABLISHED:
+ ste->expire = now + nm->tcp_est_timeout;
+ return;
+ default:
+ return;
+ }
+ }
+ case NAT_PROTOCOL_UDP:
+ ste->expire = now + nm->udp_timeout;
+ return;
+ default:
+ ste->expire = now + nm->udp_timeout;
+ return;
+ }
+}
+
+void
+nat64_tcp_session_set_state (nat64_db_st_entry_t * ste, tcp_header_t * tcp,
+ u8 is_ip6)
+{
+ switch (ste->tcp_state)
+ {
+ case NAT64_TCP_STATE_CLOSED:
+ {
+ if (tcp->flags & TCP_FLAG_SYN)
+ {
+ if (is_ip6)
+ ste->tcp_state = NAT64_TCP_STATE_V6_INIT;
+ else
+ ste->tcp_state = NAT64_TCP_STATE_V4_INIT;
+ }
+ return;
+ }
+ case NAT64_TCP_STATE_V4_INIT:
+ {
+ if (is_ip6 && (tcp->flags & TCP_FLAG_SYN))
+ ste->tcp_state = NAT64_TCP_STATE_ESTABLISHED;
+ return;
+ }
+ case NAT64_TCP_STATE_V6_INIT:
+ {
+ if (!is_ip6 && (tcp->flags & TCP_FLAG_SYN))
+ ste->tcp_state = NAT64_TCP_STATE_ESTABLISHED;
+ return;
+ }
+ case NAT64_TCP_STATE_ESTABLISHED:
+ {
+ if (tcp->flags & TCP_FLAG_FIN)
+ {
+ if (is_ip6)
+ ste->tcp_state = NAT64_TCP_STATE_V6_FIN_RCV;
+ else
+ ste->tcp_state = NAT64_TCP_STATE_V4_FIN_RCV;
+ }
+ else if (tcp->flags & TCP_FLAG_RST)
+ {
+ ste->tcp_state = NAT64_TCP_STATE_TRANS;
+ }
+ return;
+ }
+ case NAT64_TCP_STATE_V4_FIN_RCV:
+ {
+ if (is_ip6 && (tcp->flags & TCP_FLAG_FIN))
+ ste->tcp_state = NAT64_TCP_STATE_V6_FIN_V4_FIN_RCV;
+ return;
+ }
+ case NAT64_TCP_STATE_V6_FIN_RCV:
+ {
+ if (!is_ip6 && (tcp->flags & TCP_FLAG_FIN))
+ ste->tcp_state = NAT64_TCP_STATE_V6_FIN_V4_FIN_RCV;
+ return;
+ }
+ case NAT64_TCP_STATE_TRANS:
+ {
+ if (!(tcp->flags & TCP_FLAG_RST))
+ ste->tcp_state = NAT64_TCP_STATE_ESTABLISHED;
+ return;
+ }
+ default:
+ return;
+ }
+}
+
+int
+nat64_add_del_prefix (ip6_address_t * prefix, u8 plen, u32 vrf_id, u8 is_add)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_prefix_t *p = 0;
+ int i;
+
+ /* Verify prefix length */
+ if (plen != 32 && plen != 40 && plen != 48 && plen != 56 && plen != 64
+ && plen != 96)
+ return VNET_API_ERROR_INVALID_VALUE;
+
+ /* Check if tenant already have prefix */
+ for (i = 0; i < vec_len (nm->pref64); i++)
+ {
+ if (nm->pref64[i].vrf_id == vrf_id)
+ {
+ p = nm->pref64 + i;
+ break;
+ }
+ }
+
+ if (is_add)
+ {
+ if (!p)
+ {
+ vec_add2 (nm->pref64, p, 1);
+ p->fib_index =
+ fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6, vrf_id,
+ nm->fib_src_hi);
+ p->vrf_id = vrf_id;
+ }
+
+ p->prefix.as_u64[0] = prefix->as_u64[0];
+ p->prefix.as_u64[1] = prefix->as_u64[1];
+ p->plen = plen;
+ }
+ else
+ {
+ if (!p)
+ return VNET_API_ERROR_NO_SUCH_ENTRY;
+
+ // TODO: missing fib_table_unlock ?
+
+ vec_del1 (nm->pref64, i);
+ }
+
+ return 0;
+}
+
+void
+nat64_prefix_walk (nat64_prefix_walk_fn_t fn, void *ctx)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_prefix_t *p = 0;
+
+ /* *INDENT-OFF* */
+ vec_foreach (p, nm->pref64)
+ {
+ if (fn (p, ctx))
+ break;
+ };
+ /* *INDENT-ON* */
+}
+
+void
+nat64_compose_ip6 (ip6_address_t * ip6, ip4_address_t * ip4, u32 fib_index)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_prefix_t *p, *gp = 0, *prefix = 0;
+
+ /* *INDENT-OFF* */
+ vec_foreach (p, nm->pref64)
+ {
+ if (p->fib_index == fib_index)
+ {
+ prefix = p;
+ break;
+ }
+
+ if (p->fib_index == 0)
+ gp = p;
+ };
+ /* *INDENT-ON* */
+
+ if (!prefix)
+ prefix = gp;
+
+ if (prefix)
+ {
+ clib_memcpy_fast (ip6, &p->prefix, sizeof (ip6_address_t));
+ switch (p->plen)
+ {
+ case 32:
+ ip6->as_u32[1] = ip4->as_u32;
+ break;
+ case 40:
+ ip6->as_u8[5] = ip4->as_u8[0];
+ ip6->as_u8[6] = ip4->as_u8[1];
+ ip6->as_u8[7] = ip4->as_u8[2];
+ ip6->as_u8[9] = ip4->as_u8[3];
+ break;
+ case 48:
+ ip6->as_u8[6] = ip4->as_u8[0];
+ ip6->as_u8[7] = ip4->as_u8[1];
+ ip6->as_u8[9] = ip4->as_u8[2];
+ ip6->as_u8[10] = ip4->as_u8[3];
+ break;
+ case 56:
+ ip6->as_u8[7] = ip4->as_u8[0];
+ ip6->as_u8[9] = ip4->as_u8[1];
+ ip6->as_u8[10] = ip4->as_u8[2];
+ ip6->as_u8[11] = ip4->as_u8[3];
+ break;
+ case 64:
+ ip6->as_u8[9] = ip4->as_u8[0];
+ ip6->as_u8[10] = ip4->as_u8[1];
+ ip6->as_u8[11] = ip4->as_u8[2];
+ ip6->as_u8[12] = ip4->as_u8[3];
+ break;
+ case 96:
+ ip6->as_u32[3] = ip4->as_u32;
+ break;
+ default:
+ nat_elog_str ("invalid prefix length");
+ break;
+ }
+ }
+ else
+ {
+ clib_memcpy_fast (ip6, well_known_prefix, sizeof (ip6_address_t));
+ ip6->as_u32[3] = ip4->as_u32;
+ }
+}
+
+void
+nat64_extract_ip4 (ip6_address_t * ip6, ip4_address_t * ip4, u32 fib_index)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_prefix_t *p, *gp = 0;
+ u8 plen = 0;
+
+ /* *INDENT-OFF* */
+ vec_foreach (p, nm->pref64)
+ {
+ if (p->fib_index == fib_index)
+ {
+ plen = p->plen;
+ break;
+ }
+
+ if (p->vrf_id == 0)
+ gp = p;
+ };
+ /* *INDENT-ON* */
+
+ if (!plen)
+ {
+ if (gp)
+ plen = gp->plen;
+ else
+ plen = 96;
+ }
+
+ switch (plen)
+ {
+ case 32:
+ ip4->as_u32 = ip6->as_u32[1];
+ break;
+ case 40:
+ ip4->as_u8[0] = ip6->as_u8[5];
+ ip4->as_u8[1] = ip6->as_u8[6];
+ ip4->as_u8[2] = ip6->as_u8[7];
+ ip4->as_u8[3] = ip6->as_u8[9];
+ break;
+ case 48:
+ ip4->as_u8[0] = ip6->as_u8[6];
+ ip4->as_u8[1] = ip6->as_u8[7];
+ ip4->as_u8[2] = ip6->as_u8[9];
+ ip4->as_u8[3] = ip6->as_u8[10];
+ break;
+ case 56:
+ ip4->as_u8[0] = ip6->as_u8[7];
+ ip4->as_u8[1] = ip6->as_u8[9];
+ ip4->as_u8[2] = ip6->as_u8[10];
+ ip4->as_u8[3] = ip6->as_u8[11];
+ break;
+ case 64:
+ ip4->as_u8[0] = ip6->as_u8[9];
+ ip4->as_u8[1] = ip6->as_u8[10];
+ ip4->as_u8[2] = ip6->as_u8[11];
+ ip4->as_u8[3] = ip6->as_u8[12];
+ break;
+ case 96:
+ ip4->as_u32 = ip6->as_u32[3];
+ break;
+ default:
+ nat_elog_str ("invalid prefix length");
+ break;
+ }
+}
+
+/**
+ * @brief Per worker process checking expire time for NAT64 sessions.
+ */
+static uword
+nat64_expire_worker_walk_fn (vlib_main_t * vm, vlib_node_runtime_t * rt,
+ vlib_frame_t * f)
+{
+ nat64_main_t *nm = &nat64_main;
+ u32 thread_index = vm->thread_index;
+ nat64_db_t *db;
+ u32 now;
+
+ // TODO: barier sync on plugin enabled
+ if (plugin_enabled () == 0)
+ return 0;
+
+ db = &nm->db[thread_index];
+ now = (u32) vlib_time_now (vm);
+
+ nad64_db_st_free_expired (thread_index, db, now);
+ vlib_set_simple_counter (&nm->total_bibs, thread_index, 0,
+ db->bib.bib_entries_num);
+ vlib_set_simple_counter (&nm->total_sessions, thread_index, 0,
+ db->st.st_entries_num);
+ return 0;
+}
+
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE (nat64_expire_worker_walk_node, static) = {
+ .function = nat64_expire_worker_walk_fn,
+ .type = VLIB_NODE_TYPE_INPUT,
+ .state = VLIB_NODE_STATE_INTERRUPT,
+ .name = "nat64-expire-worker-walk",
+};
+/* *INDENT-ON* */
+
+/**
+ * @brief Centralized process to drive per worker expire walk.
+ */
+static uword
+nat64_expire_walk_fn (vlib_main_t * vm, vlib_node_runtime_t * rt,
+ vlib_frame_t * f)
+{
+ nat64_main_t *nm = &nat64_main;
+ vlib_main_t **worker_vms = 0, *worker_vm;
+ int i;
+ uword event_type, *event_data = 0;
+
+ if (vec_len (vlib_mains) == 0)
+ vec_add1 (worker_vms, vm);
+ else
+ {
+ for (i = 0; i < vec_len (vlib_mains); i++)
+ {
+ worker_vm = vlib_mains[i];
+ if (worker_vm)
+ vec_add1 (worker_vms, worker_vm);
+ }
+ }
+
+ while (1)
+ {
+ if (nm->total_enabled_count)
+ {
+ vlib_process_wait_for_event_or_clock (vm, 10.0);
+ event_type = vlib_process_get_events (vm, &event_data);
+ }
+ else
+ {
+ vlib_process_wait_for_event (vm);
+ event_type = vlib_process_get_events (vm, &event_data);
+ }
+
+ switch (event_type)
+ {
+ case ~0:
+ break;
+ case NAT64_CLEANER_RESCHEDULE:
+ break;
+ default:
+ nat64_log_err ("unknown event %u", event_type);
+ break;
+ }
+
+ for (i = 0; i < vec_len (worker_vms); i++)
+ {
+ worker_vm = worker_vms[i];
+ vlib_node_set_interrupt_pending (worker_vm,
+ nm->expire_worker_walk_node_index);
+ }
+ }
+
+ return 0;
+}
+
+void
+nat64_create_expire_walk_process ()
+{
+ nat64_main_t *nm = &nat64_main;
+
+ if (nm->expire_walk_node_index)
+ return;
+ nm->expire_walk_node_index = vlib_process_create (vlib_get_main (),
+ "nat64-expire-walk",
+ nat64_expire_walk_fn,
+ 16 /* stack_bytes */ );
+}
+
+int
+nat64_plugin_enable (nat64_config_t c)
+{
+ nat64_main_t *nm = &nat64_main;
+
+ if (plugin_enabled () == 1)
+ {
+ nat64_log_err ("plugin already enabled!");
+ return 1;
+ }
+
+ if (!c.bib_buckets)
+ c.bib_buckets = 1024;
+
+ if (!c.bib_memory_size)
+ c.bib_memory_size = 128 << 20;
+
+ if (!c.st_buckets)
+ c.st_buckets = 2048;
+
+ if (!c.st_memory_size)
+ c.st_memory_size = 256 << 20;
+
+ nm->config = c;
+
+ if (nat64_init_hash (c))
+ {
+ nat64_log_err ("initializing hashes failed!");
+ return 1;
+ }
+
+ nat64_create_expire_walk_process ();
+
+ nm->enabled = 1;
+ return 0;
+}
+
+int
+nat64_plugin_disable ()
+{
+ nat64_main_t *nm = &nat64_main;
+ vnet_main_t *vnm = vnet_get_main ();
+ int rv = 0;
+
+ nat64_address_t *a;
+ nat64_interface_t *i, *interfaces;
+
+ if (plugin_enabled () == 0)
+ {
+ nat64_log_err ("plugin already disabled!");
+ return 1;
+ }
+ nm->enabled = 0;
+
+ interfaces = vec_dup (nm->interfaces);
+ vec_foreach (i, interfaces)
+ {
+ rv = nat64_interface_add_del (i->sw_if_index, i->flags, 1);
+ if (rv)
+ {
+ nat64_log_err ("%U %s interface del failed",
+ unformat_vnet_sw_interface,
+ i->flags & NAT64_INTERFACE_FLAG_IS_INSIDE ?
+ "inside" : "outside", vnm, i->sw_if_index);
+ }
+ }
+ vec_free (interfaces);
+ pool_free (nm->interfaces);
+
+ nat64_reset_timeouts ();
+
+ if (nat64_free_hash ())
+ {
+ rv = 1;
+ nat64_log_err ("freeing hashes failed!");
+ }
+
+ // TODO: based on nat64_add_del_prefix fib_table_unlock is not called
+ vec_free (nm->pref64);
+
+ if (vec_len (nm->addr_pool))
+ {
+ vec_foreach (a, nm->addr_pool)
+ {
+ if (a->fib_index != ~0)
+ fib_table_unlock (a->fib_index, FIB_PROTOCOL_IP6, nm->fib_src_hi);
+ }
+ vec_free (nm->addr_pool);
+ }
+ return rv;
+}
+
+uword
+unformat_nat_protocol (unformat_input_t * input, va_list * args)
+{
+ u32 *r = va_arg (*args, u32 *);
+
+ if (0);
+#define _(N, i, n, s) else if (unformat (input, s)) *r = NAT_PROTOCOL_##N;
+ foreach_nat_protocol
+#undef _
+ else
+ return 0;
+ return 1;
+}
+
+u8 *
+format_nat_protocol (u8 * s, va_list * args)
+{
+ u32 i = va_arg (*args, u32);
+ u8 *t = 0;
+
+ switch (i)
+ {
+#define _(N, j, n, str) case NAT_PROTOCOL_##N: t = (u8 *) str; break;
+ foreach_nat_protocol
+#undef _
+ default:
+ s = format (s, "unknown");
+ return s;
+ }
+ s = format (s, "%s", t);
+ return s;
+}
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */
diff --git a/src/plugins/nat/nat64/nat64.h b/src/plugins/nat/nat64/nat64.h
new file mode 100644
index 00000000000..1180f9df778
--- /dev/null
+++ b/src/plugins/nat/nat64/nat64.h
@@ -0,0 +1,531 @@
+/*
+ * Copyright (c) 2017 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#ifndef __included_nat64_h__
+#define __included_nat64_h__
+
+#include <vnet/vnet.h>
+#include <vnet/ip/ip.h>
+#include <vnet/ethernet/ethernet.h>
+#include <vnet/ip/icmp46_packet.h>
+#include <vnet/api_errno.h>
+#include <vnet/fib/fib_source.h>
+#include <vppinfra/dlist.h>
+#include <vppinfra/error.h>
+#include <vlibapi/api.h>
+#include <vlib/log.h>
+#include <vnet/fib/fib_table.h>
+#include <vnet/fib/ip4_fib.h>
+#include <vnet/ip/reass/ip4_sv_reass.h>
+
+#include <nat/lib/lib.h>
+#include <nat/lib/inlines.h>
+#include <nat/lib/nat_inlines.h>
+
+#include <nat/nat64/nat64_db.h>
+
+typedef struct
+{
+ u16 identifier;
+ u16 sequence;
+} icmp_echo_header_t;
+
+typedef struct
+{
+ u16 src_port, dst_port;
+} tcp_udp_header_t;
+
+#define foreach_nat64_tcp_ses_state \
+ _(0, CLOSED, "closed") \
+ _(1, V4_INIT, "v4-init") \
+ _(2, V6_INIT, "v6-init") \
+ _(3, ESTABLISHED, "established") \
+ _(4, V4_FIN_RCV, "v4-fin-rcv") \
+ _(5, V6_FIN_RCV, "v6-fin-rcv") \
+ _(6, V6_FIN_V4_FIN_RCV, "v6-fin-v4-fin-rcv") \
+ _(7, TRANS, "trans")
+
+typedef enum
+{
+#define _(v, N, s) NAT64_TCP_STATE_##N = v,
+ foreach_nat64_tcp_ses_state
+#undef _
+} nat64_tcp_ses_state_t;
+
+typedef enum
+{
+ NAT64_CLEANER_RESCHEDULE = 1,
+} nat64_cleaner_process_event_e;
+
+typedef struct
+{
+ ip6_address_t prefix;
+ u8 plen;
+ u32 vrf_id;
+ u32 fib_index;
+} nat64_prefix_t;
+
+typedef struct
+{
+ ip6_address_t in_addr;
+ u16 in_port;
+ ip4_address_t out_addr;
+ u16 out_port;
+ u32 fib_index;
+ u32 thread_index;
+ u8 proto;
+ u8 is_add;
+ u8 done;
+} nat64_static_bib_to_update_t;
+
+typedef struct
+{
+ ip4_address_t addr;
+ u32 fib_index;
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ u16 busy_##n##_ports; \
+ u16 * busy_##n##_ports_per_thread; \
+ u32 busy_##n##_port_refcounts[65535];
+ foreach_nat_protocol
+#undef _
+/* *INDENT-ON* */
+} nat64_address_t;
+
+typedef struct
+{
+ u32 sw_if_index;
+ u8 flags;
+} nat64_interface_t;
+
+typedef struct
+{
+ u32 enabled;
+
+ nat64_config_t config;
+
+ /* API message ID base */
+ u16 msg_id_base;
+
+ /* log class */
+ vlib_log_class_t log_class;
+
+ /** Interface pool */
+ nat64_interface_t *interfaces;
+
+ /** Address pool vector */
+ nat64_address_t *addr_pool;
+
+ /** sw_if_indices whose interface addresses should be auto-added */
+ u32 *auto_add_sw_if_indices;
+
+ /** Pref64 vector */
+ nat64_prefix_t *pref64;
+
+ /** BIB and session DB per thread */
+ nat64_db_t *db;
+
+ /** Worker handoff */
+ u32 fq_in2out_index;
+ u32 fq_out2in_index;
+
+ /** Pool of static BIB entries to be added/deleted in worker threads */
+ nat64_static_bib_to_update_t *static_bibs;
+
+ /** config parameters */
+ u32 bib_buckets;
+ uword bib_memory_size;
+ u32 st_buckets;
+ uword st_memory_size;
+
+ /** values of various timeouts */
+ u32 udp_timeout;
+ u32 icmp_timeout;
+ u32 tcp_trans_timeout;
+ u32 tcp_est_timeout;
+
+ /* Total count of interfaces enabled */
+ u32 total_enabled_count;
+
+ /* Expire walk process node index */
+ u32 expire_walk_node_index;
+
+ /* Expire worker walk process node index */
+ u32 expire_worker_walk_node_index;
+
+ /* counters/gauges */
+ vlib_simple_counter_main_t total_bibs;
+ vlib_simple_counter_main_t total_sessions;
+
+ /** node index **/
+ u32 error_node_index;
+
+ u32 in2out_node_index;
+ u32 in2out_slowpath_node_index;
+
+ u32 out2in_node_index;
+
+#define _(x) vlib_simple_counter_main_t x;
+ struct
+ {
+ struct
+ {
+ foreach_nat_counter;
+ } in2out;
+
+ struct
+ {
+ foreach_nat_counter;
+ } out2in;
+ } counters;
+#undef _
+
+ /* convenience */
+ ip4_main_t *ip4_main;
+
+ /* required */
+ vnet_main_t *vnet_main;
+
+ /* Randomize port allocation order */
+ u32 random_seed;
+
+ /* TCP MSS clamping */
+ u16 mss_clamping;
+
+ fib_source_t fib_src_hi;
+ fib_source_t fib_src_low;
+
+ /* Thread settings */
+ u32 num_workers;
+ u32 first_worker_index;
+ u32 *workers;
+ u16 port_per_thread;
+
+} nat64_main_t;
+
+extern nat64_main_t nat64_main;
+extern vlib_node_registration_t nat64_in2out_node;
+extern vlib_node_registration_t nat64_out2in_node;
+
+/**
+ * @brief Add/delete address to NAT64 pool.
+ *
+ * @param thread_index Thread index used by ipfix nat logging (not address per thread).
+ * @param addr IPv4 address.
+ * @param vrf_id VRF id of tenant, ~0 means independent of VRF.
+ * @param is_add 1 if add, 0 if delete.
+ *
+ * @returns 0 on success, non-zero value otherwise.
+ */
+int nat64_add_del_pool_addr (u32 thread_index,
+ ip4_address_t * addr, u32 vrf_id, u8 is_add);
+
+/**
+ * @brief Call back function when walking addresses in NAT64 pool, non-zero
+ * return value stop walk.
+ */
+typedef int (*nat64_pool_addr_walk_fn_t) (nat64_address_t * addr, void *ctx);
+
+/**
+ * @brief Walk NAT64 pool.
+ *
+ * @param fn The function to invoke on each entry visited.
+ * @param ctx A context passed in the visit function.
+ */
+void nat64_pool_addr_walk (nat64_pool_addr_walk_fn_t fn, void *ctx);
+
+/**
+ * @brief NAT64 pool address from specific (DHCP addressed) interface.
+ *
+ * @param sw_if_index Index of the interface.
+ * @param is_add 1 if add, 0 if delete.
+ *
+ * @returns 0 on success, non-zero value otherwise.
+ */
+int nat64_add_interface_address (u32 sw_if_index, int is_add);
+
+/**
+ * @brief Enable/disable NAT64 feature on the interface.
+ *
+ * @param sw_if_index Index of the interface.
+ * @param is_inside 1 if inside, 0 if outside.
+ * @param is_add 1 if add, 0 if delete.
+ *
+ * @returns 0 on success, non-zero value otherwise.
+ */
+int nat64_interface_add_del (u32 sw_if_index, u8 is_inside, u8 is_add);
+
+/**
+ * @brief Call back function when walking interfaces with NAT64 feature,
+ * non-zero return value stop walk.
+ */
+typedef int (*nat64_interface_walk_fn_t) (nat64_interface_t * i, void *ctx);
+
+/**
+ * @brief Walk NAT64 interfaces.
+ *
+ * @param fn The function to invoke on each entry visited.
+ * @param ctx A context passed in the visit function.
+ */
+void nat64_interfaces_walk (nat64_interface_walk_fn_t fn, void *ctx);
+
+/**
+ * @brief Initialize NAT64.
+ *
+ * @param vm vlib main.
+ *
+ * @return error code.
+ */
+clib_error_t *nat64_init (vlib_main_t * vm);
+
+/**
+ * @brief Add/delete static NAT64 BIB entry.
+ *
+ * @param in_addr Inside IPv6 address.
+ * @param out_addr Outside IPv4 address.
+ * @param in_port Inside port number.
+ * @param out_port Outside port number.
+ * @param proto L4 protocol.
+ * @param vrf_id VRF id of tenant.
+ * @param is_add 1 if add, 0 if delete.
+ *
+ * @returns 0 on success, non-zero value otherwise.
+ */
+int nat64_add_del_static_bib_entry (ip6_address_t * in_addr,
+ ip4_address_t * out_addr, u16 in_port,
+ u16 out_port, u8 proto, u32 vrf_id,
+ u8 is_add);
+
+/**
+ * @brief Alloce IPv4 address and port pair from NAT64 pool.
+ *
+ * @param fib_index FIB index of tenant.
+ * @param proto L4 protocol.
+ * @param addr Allocated IPv4 address.
+ * @param port Allocated port number.
+ * @param thread_index Thread index.
+ *
+ * @returns 0 on success, non-zero value otherwise.
+ */
+int nat64_alloc_out_addr_and_port (u32 fib_index, nat_protocol_t proto,
+ ip4_address_t * addr, u16 * port,
+ u32 thread_index);
+
+/**
+ * @brief Set UDP session timeout.
+ *
+ * @param timeout Timeout value in seconds (if 0 reset to default value 300sec).
+ *
+ * @returns 0 on success, non-zero value otherwise.
+ */
+int nat64_set_udp_timeout (u32 timeout);
+
+/**
+ * @brief Get UDP session timeout.
+ *
+ * @returns UDP session timeout in seconds.
+ */
+u32 nat64_get_udp_timeout (void);
+
+/**
+ * @brief Set ICMP session timeout.
+ *
+ * @param timeout Timeout value in seconds (if 0 reset to default value 60sec).
+ *
+ * @returns 0 on success, non-zero value otherwise.
+ */
+int nat64_set_icmp_timeout (u32 timeout);
+
+/**
+ * @brief Get ICMP session timeout.
+ *
+ * @returns ICMP session timeout in seconds.
+ */
+u32 nat64_get_icmp_timeout (void);
+
+/**
+ * @brief Set TCP session timeouts.
+ *
+ * @param trans Transitory timeout in seconds (if 0 reset to default value 240sec).
+ * @param est Established timeout in seconds (if 0 reset to default value 7440sec).
+ *
+ * @returns 0 on success, non-zero value otherwise.
+ */
+int nat64_set_tcp_timeouts (u32 trans, u32 est);
+
+/**
+ * @brief Get TCP transitory timeout.
+ *
+ * @returns TCP transitory timeout in seconds.
+ */
+u32 nat64_get_tcp_trans_timeout (void);
+
+/**
+ * @brief Get TCP established timeout.
+ *
+ * @returns TCP established timeout in seconds.
+ */
+u32 nat64_get_tcp_est_timeout (void);
+
+/**
+ * @brief Reset NAT64 session timeout.
+ *
+ * @param ste Session table entry.
+ * @param vm VLIB main.
+ **/
+void nat64_session_reset_timeout (nat64_db_st_entry_t * ste,
+ vlib_main_t * vm);
+
+/**
+ * @brief Set NAT64 TCP session state.
+ *
+ * @param ste Session table entry.
+ * @param tcp TCP header.
+ * @param is_ip6 1 if IPv6 packet, 0 if IPv4.
+ */
+void nat64_tcp_session_set_state (nat64_db_st_entry_t * ste,
+ tcp_header_t * tcp, u8 is_ip6);
+
+/**
+ * @brief Add/delete NAT64 prefix.
+ *
+ * @param prefix NAT64 prefix.
+ * @param plen Prefix length.
+ * @param vrf_id VRF id of tenant.
+ * @param is_add 1 if add, 0 if delete.
+ *
+ * @returns 0 on success, non-zero value otherwise.
+ */
+int nat64_add_del_prefix (ip6_address_t * prefix, u8 plen, u32 vrf_id,
+ u8 is_add);
+
+/**
+ * @brief Call back function when walking addresses in NAT64 prefixes, non-zero
+ * return value stop walk.
+ */
+typedef int (*nat64_prefix_walk_fn_t) (nat64_prefix_t * pref64, void *ctx);
+
+/**
+ * @brief Walk NAT64 prefixes.
+ *
+ * @param fn The function to invoke on each entry visited.
+ * @param ctx A context passed in the visit function.
+ */
+void nat64_prefix_walk (nat64_prefix_walk_fn_t fn, void *ctx);
+
+/**
+ * Compose IPv4-embedded IPv6 addresses.
+ * @param ip6 IPv4-embedded IPv6 addresses.
+ * @param ip4 IPv4 address.
+ * @param fib_index Tenant FIB index.
+ */
+void nat64_compose_ip6 (ip6_address_t * ip6, ip4_address_t * ip4,
+ u32 fib_index);
+
+/**
+ * Extract IPv4 address from the IPv4-embedded IPv6 addresses.
+ *
+ * @param ip6 IPv4-embedded IPv6 addresses.
+ * @param ip4 IPv4 address.
+ * @param fib_index Tenant FIB index.
+ */
+void nat64_extract_ip4 (ip6_address_t * ip6, ip4_address_t * ip4,
+ u32 fib_index);
+
+/**
+ * @brief Set NAT64 hash tables configuration.
+ *
+ * @param bib_buckets Number of BIB hash buckets.
+ * @param bib_memory_size Memory size of BIB hash.
+ * @param st_buckets Number of session table hash buckets.
+ * @param st_memory_size Memory size of session table hash.
+ */
+void nat64_set_hash (u32 bib_buckets, uword bib_memory_size, u32 st_buckets,
+ uword st_memory_size);
+
+/**
+ * @brief Get worker thread index for NAT64 in2out.
+ *
+ * @param addr IPv6 src address.
+ *
+ * @returns worker thread index.
+ */
+u32 nat64_get_worker_in2out (ip6_address_t * addr);
+
+/**
+ * @brief Get worker thread index for NAT64 out2in.
+ *
+ * @param ip IPv4 header.
+ *
+ * @returns worker thread index.
+ */
+u32 nat64_get_worker_out2in (vlib_buffer_t * b, ip4_header_t * ip);
+
+/* NAT64 interface flags */
+#define NAT64_INTERFACE_FLAG_IS_INSIDE 1
+#define NAT64_INTERFACE_FLAG_IS_OUTSIDE 2
+
+/** \brief Check if NAT64 interface is inside.
+ @param i NAT64 interface
+ @return 1 if inside interface
+*/
+#define nat64_interface_is_inside(i) i->flags & NAT64_INTERFACE_FLAG_IS_INSIDE
+
+/** \brief Check if NAT64 interface is outside.
+ @param i NAT64 interface
+ @return 1 if outside interface
+*/
+#define nat64_interface_is_outside(i) i->flags & NAT64_INTERFACE_FLAG_IS_OUTSIDE
+
+static_always_inline u8
+plugin_enabled ()
+{
+ nat64_main_t *nm = &nat64_main;
+ return nm->enabled;
+}
+
+void
+nat64_add_del_addr_to_fib (ip4_address_t * addr, u8 p_len, u32 sw_if_index,
+ int is_add);
+
+int nat64_plugin_enable (nat64_config_t c);
+int nat64_plugin_disable ();
+void nat64_reset_timeouts ();
+
+format_function_t format_nat_protocol;
+unformat_function_t unformat_nat_protocol;
+
+/* logging */
+#define nat64_log_err(...) \
+ vlib_log(VLIB_LOG_LEVEL_ERR, nat64_main.log_class, __VA_ARGS__)
+#define nat64_log_warn(...) \
+ vlib_log(VLIB_LOG_LEVEL_WARNING, nat64_main.log_class, __VA_ARGS__)
+#define nat64_log_notice(...) \
+ vlib_log(VLIB_LOG_LEVEL_NOTICE, nat64_main.log_class, __VA_ARGS__)
+#define nat64_log_info(...) \
+ vlib_log(VLIB_LOG_LEVEL_INFO, nat64_main.log_class, __VA_ARGS__)
+#define nat64_log_debug(...)\
+ vlib_log(VLIB_LOG_LEVEL_DEBUG, nat64_main.log_class, __VA_ARGS__)
+
+clib_error_t *nat64_api_hookup (vlib_main_t * vm);
+
+#endif /* __included_nat64_h__ */
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */
diff --git a/src/plugins/nat/nat64/nat64_api.c b/src/plugins/nat/nat64/nat64_api.c
new file mode 100644
index 00000000000..e64b6434fd2
--- /dev/null
+++ b/src/plugins/nat/nat64/nat64_api.c
@@ -0,0 +1,458 @@
+/*
+ * Copyright (c) 2020 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vnet/ip/ip_types_api.h>
+#include <vlibmemory/api.h>
+#include <nat/nat64/nat64.h>
+#include <nat/nat64/nat64.api_enum.h>
+#include <nat/nat64/nat64.api_types.h>
+#include <vnet/fib/fib_table.h>
+#include <vnet/ip/ip.h>
+
+#define REPLY_MSG_ID_BASE nm->msg_id_base
+#include <vlibapi/api_helper_macros.h>
+
+static void
+ vl_api_nat64_plugin_enable_disable_t_handler
+ (vl_api_nat64_plugin_enable_disable_t * mp)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_nat64_plugin_enable_disable_reply_t *rmp;
+ nat64_config_t c = { 0 };
+ int rv = 0;
+ if (mp->enable)
+ {
+ c.bib_buckets = ntohl (mp->bib_buckets);
+ c.bib_memory_size = ntohl (mp->bib_memory_size);
+ c.st_buckets = ntohl (mp->st_buckets);
+ c.st_memory_size = ntohl (mp->st_memory_size);
+ rv = nat64_plugin_enable (c);
+ }
+ else
+ {
+ rv = nat64_plugin_disable ();
+ }
+ REPLY_MACRO (VL_API_NAT64_PLUGIN_ENABLE_DISABLE_REPLY);
+}
+
+static void
+vl_api_nat64_set_timeouts_t_handler (vl_api_nat64_set_timeouts_t * mp)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_nat64_set_timeouts_reply_t *rmp;
+ int rv = 0;
+
+ nm->udp_timeout = ntohl (mp->udp);
+ nm->tcp_est_timeout = ntohl (mp->tcp_established);
+ nm->tcp_trans_timeout = ntohl (mp->tcp_transitory);
+ nm->icmp_timeout = ntohl (mp->icmp);
+
+ REPLY_MACRO (VL_API_NAT64_SET_TIMEOUTS_REPLY);
+}
+
+static void
+vl_api_nat64_get_timeouts_t_handler (vl_api_nat64_get_timeouts_t * mp)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_nat64_get_timeouts_reply_t *rmp;
+ int rv = 0;
+
+ /* *INDENT-OFF* */
+ REPLY_MACRO2 (VL_API_NAT64_GET_TIMEOUTS_REPLY,
+ ({
+ rmp->udp = htonl (nm->udp_timeout);
+ rmp->tcp_established = htonl (nm->tcp_est_timeout);
+ rmp->tcp_transitory = htonl (nm->tcp_trans_timeout);
+ rmp->icmp = htonl (nm->icmp_timeout);
+ }))
+ /* *INDENT-ON* */
+}
+
+static void
+ vl_api_nat64_add_del_pool_addr_range_t_handler
+ (vl_api_nat64_add_del_pool_addr_range_t * mp)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_nat64_add_del_pool_addr_range_reply_t *rmp;
+ int rv = 0;
+ ip4_address_t this_addr;
+ u32 start_host_order, end_host_order;
+ u32 vrf_id;
+ int i, count;
+ u32 *tmp;
+
+ tmp = (u32 *) mp->start_addr;
+ start_host_order = clib_host_to_net_u32 (tmp[0]);
+ tmp = (u32 *) mp->end_addr;
+ end_host_order = clib_host_to_net_u32 (tmp[0]);
+
+ count = (end_host_order - start_host_order) + 1;
+
+ vrf_id = clib_host_to_net_u32 (mp->vrf_id);
+
+ memcpy (&this_addr.as_u8, mp->start_addr, 4);
+
+ for (i = 0; i < count; i++)
+ {
+ if ((rv = nat64_add_del_pool_addr (0, &this_addr, vrf_id, mp->is_add)))
+ goto send_reply;
+
+ increment_v4_address (&this_addr);
+ }
+
+send_reply:
+ REPLY_MACRO (VL_API_NAT64_ADD_DEL_POOL_ADDR_RANGE_REPLY);
+}
+
+typedef struct nat64_api_walk_ctx_t_
+{
+ vl_api_registration_t *reg;
+ u32 context;
+ nat64_db_t *db;
+} nat64_api_walk_ctx_t;
+
+static int
+nat64_api_pool_walk (nat64_address_t * a, void *arg)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_nat64_pool_addr_details_t *rmp;
+ nat64_api_walk_ctx_t *ctx = arg;
+
+ rmp = vl_msg_api_alloc (sizeof (*rmp));
+ clib_memset (rmp, 0, sizeof (*rmp));
+ rmp->_vl_msg_id = ntohs (VL_API_NAT64_POOL_ADDR_DETAILS + nm->msg_id_base);
+ clib_memcpy (rmp->address, &(a->addr), 4);
+ if (a->fib_index != ~0)
+ {
+ fib_table_t *fib = fib_table_get (a->fib_index, FIB_PROTOCOL_IP6);
+ if (!fib)
+ return -1;
+ rmp->vrf_id = ntohl (fib->ft_table_id);
+ }
+ else
+ rmp->vrf_id = ~0;
+ rmp->context = ctx->context;
+
+ vl_api_send_msg (ctx->reg, (u8 *) rmp);
+
+ return 0;
+}
+
+static void
+vl_api_nat64_pool_addr_dump_t_handler (vl_api_nat64_pool_addr_dump_t * mp)
+{
+ vl_api_registration_t *reg;
+
+ reg = vl_api_client_index_to_registration (mp->client_index);
+ if (!reg)
+ return;
+
+ nat64_api_walk_ctx_t ctx = {
+ .reg = reg,
+ .context = mp->context,
+ };
+
+ nat64_pool_addr_walk (nat64_api_pool_walk, &ctx);
+}
+
+static void
+vl_api_nat64_add_del_interface_t_handler (vl_api_nat64_add_del_interface_t *
+ mp)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_nat64_add_del_interface_reply_t *rmp;
+ int rv = 0;
+
+ VALIDATE_SW_IF_INDEX (mp);
+
+ rv =
+ nat64_interface_add_del (ntohl (mp->sw_if_index),
+ mp->flags & NAT_API_IS_INSIDE, mp->is_add);
+
+ BAD_SW_IF_INDEX_LABEL;
+
+ REPLY_MACRO (VL_API_NAT64_ADD_DEL_INTERFACE_REPLY);
+}
+
+static int
+nat64_api_interface_walk (nat64_interface_t * i, void *arg)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_nat64_interface_details_t *rmp;
+ nat64_api_walk_ctx_t *ctx = arg;
+
+ rmp = vl_msg_api_alloc (sizeof (*rmp));
+ clib_memset (rmp, 0, sizeof (*rmp));
+ rmp->_vl_msg_id = ntohs (VL_API_NAT64_INTERFACE_DETAILS + nm->msg_id_base);
+ rmp->sw_if_index = ntohl (i->sw_if_index);
+
+ if (nat64_interface_is_inside (i))
+ rmp->flags |= NAT_API_IS_INSIDE;
+ if (nat64_interface_is_outside (i))
+ rmp->flags |= NAT_API_IS_OUTSIDE;
+
+ rmp->context = ctx->context;
+
+ vl_api_send_msg (ctx->reg, (u8 *) rmp);
+
+ return 0;
+}
+
+static void
+vl_api_nat64_interface_dump_t_handler (vl_api_nat64_interface_dump_t * mp)
+{
+ vl_api_registration_t *reg;
+
+ reg = vl_api_client_index_to_registration (mp->client_index);
+ if (!reg)
+ return;
+
+ nat64_api_walk_ctx_t ctx = {
+ .reg = reg,
+ .context = mp->context,
+ };
+
+ nat64_interfaces_walk (nat64_api_interface_walk, &ctx);
+}
+
+static void
+ vl_api_nat64_add_del_static_bib_t_handler
+ (vl_api_nat64_add_del_static_bib_t * mp)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_nat64_add_del_static_bib_reply_t *rmp;
+ ip6_address_t in_addr;
+ ip4_address_t out_addr;
+ int rv = 0;
+
+ memcpy (&in_addr.as_u8, mp->i_addr, 16);
+ memcpy (&out_addr.as_u8, mp->o_addr, 4);
+
+ rv =
+ nat64_add_del_static_bib_entry (&in_addr, &out_addr,
+ clib_net_to_host_u16 (mp->i_port),
+ clib_net_to_host_u16 (mp->o_port),
+ mp->proto,
+ clib_net_to_host_u32 (mp->vrf_id),
+ mp->is_add);
+
+ REPLY_MACRO (VL_API_NAT64_ADD_DEL_STATIC_BIB_REPLY);
+}
+
+static int
+nat64_api_bib_walk (nat64_db_bib_entry_t * bibe, void *arg)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_nat64_bib_details_t *rmp;
+ nat64_api_walk_ctx_t *ctx = arg;
+ fib_table_t *fib;
+
+ fib = fib_table_get (bibe->fib_index, FIB_PROTOCOL_IP6);
+ if (!fib)
+ return -1;
+
+ rmp = vl_msg_api_alloc (sizeof (*rmp));
+ clib_memset (rmp, 0, sizeof (*rmp));
+ rmp->_vl_msg_id = ntohs (VL_API_NAT64_BIB_DETAILS + nm->msg_id_base);
+ rmp->context = ctx->context;
+ clib_memcpy (rmp->i_addr, &(bibe->in_addr), 16);
+ clib_memcpy (rmp->o_addr, &(bibe->out_addr), 4);
+ rmp->i_port = bibe->in_port;
+ rmp->o_port = bibe->out_port;
+ rmp->vrf_id = ntohl (fib->ft_table_id);
+ rmp->proto = bibe->proto;
+ if (bibe->is_static)
+ rmp->flags |= NAT_API_IS_STATIC;
+ rmp->ses_num = ntohl (bibe->ses_num);
+
+ vl_api_send_msg (ctx->reg, (u8 *) rmp);
+
+ return 0;
+}
+
+static void
+vl_api_nat64_bib_dump_t_handler (vl_api_nat64_bib_dump_t * mp)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_registration_t *reg;
+ nat64_db_t *db;
+
+ reg = vl_api_client_index_to_registration (mp->client_index);
+ if (!reg)
+ return;
+
+ nat64_api_walk_ctx_t ctx = {
+ .reg = reg,
+ .context = mp->context,
+ };
+
+ /* *INDENT-OFF* */
+ vec_foreach (db, nm->db)
+ nat64_db_bib_walk (db, mp->proto, nat64_api_bib_walk, &ctx);
+ /* *INDENT-ON* */
+}
+
+static int
+nat64_api_st_walk (nat64_db_st_entry_t * ste, void *arg)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_nat64_st_details_t *rmp;
+ nat64_api_walk_ctx_t *ctx = arg;
+ nat64_db_bib_entry_t *bibe;
+ fib_table_t *fib;
+
+ bibe = nat64_db_bib_entry_by_index (ctx->db, ste->proto, ste->bibe_index);
+ if (!bibe)
+ return -1;
+
+ fib = fib_table_get (bibe->fib_index, FIB_PROTOCOL_IP6);
+ if (!fib)
+ return -1;
+
+ rmp = vl_msg_api_alloc (sizeof (*rmp));
+ clib_memset (rmp, 0, sizeof (*rmp));
+ rmp->_vl_msg_id = ntohs (VL_API_NAT64_ST_DETAILS + nm->msg_id_base);
+ rmp->context = ctx->context;
+ clib_memcpy (rmp->il_addr, &(bibe->in_addr), 16);
+ clib_memcpy (rmp->ol_addr, &(bibe->out_addr), 4);
+ rmp->il_port = bibe->in_port;
+ rmp->ol_port = bibe->out_port;
+ clib_memcpy (rmp->ir_addr, &(ste->in_r_addr), 16);
+ clib_memcpy (rmp->or_addr, &(ste->out_r_addr), 4);
+ rmp->il_port = ste->r_port;
+ rmp->vrf_id = ntohl (fib->ft_table_id);
+ rmp->proto = ste->proto;
+
+ vl_api_send_msg (ctx->reg, (u8 *) rmp);
+
+ return 0;
+}
+
+static void
+vl_api_nat64_st_dump_t_handler (vl_api_nat64_st_dump_t * mp)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_registration_t *reg;
+ nat64_db_t *db;
+
+ reg = vl_api_client_index_to_registration (mp->client_index);
+ if (!reg)
+ return;
+
+ nat64_api_walk_ctx_t ctx = {
+ .reg = reg,
+ .context = mp->context,
+ };
+
+ /* *INDENT-OFF* */
+ vec_foreach (db, nm->db)
+ {
+ ctx.db = db;
+ nat64_db_st_walk (db, mp->proto, nat64_api_st_walk, &ctx);
+ }
+ /* *INDENT-ON* */
+}
+
+static void
+vl_api_nat64_add_del_prefix_t_handler (vl_api_nat64_add_del_prefix_t * mp)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_nat64_add_del_prefix_reply_t *rmp;
+ ip6_address_t prefix;
+ int rv = 0;
+
+ memcpy (&prefix.as_u8, mp->prefix.address, 16);
+
+ rv =
+ nat64_add_del_prefix (&prefix, mp->prefix.len,
+ clib_net_to_host_u32 (mp->vrf_id), mp->is_add);
+ REPLY_MACRO (VL_API_NAT64_ADD_DEL_PREFIX_REPLY);
+}
+
+static int
+nat64_api_prefix_walk (nat64_prefix_t * p, void *arg)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_nat64_prefix_details_t *rmp;
+ nat64_api_walk_ctx_t *ctx = arg;
+
+ rmp = vl_msg_api_alloc (sizeof (*rmp));
+ clib_memset (rmp, 0, sizeof (*rmp));
+ rmp->_vl_msg_id = ntohs (VL_API_NAT64_PREFIX_DETAILS + nm->msg_id_base);
+ clib_memcpy (rmp->prefix.address, &(p->prefix), 16);
+ rmp->prefix.len = p->plen;
+ rmp->vrf_id = ntohl (p->vrf_id);
+ rmp->context = ctx->context;
+
+ vl_api_send_msg (ctx->reg, (u8 *) rmp);
+
+ return 0;
+}
+
+static void
+vl_api_nat64_prefix_dump_t_handler (vl_api_nat64_prefix_dump_t * mp)
+{
+ vl_api_registration_t *reg;
+
+ reg = vl_api_client_index_to_registration (mp->client_index);
+ if (!reg)
+ return;
+
+ nat64_api_walk_ctx_t ctx = {
+ .reg = reg,
+ .context = mp->context,
+ };
+
+ nat64_prefix_walk (nat64_api_prefix_walk, &ctx);
+}
+
+static void
+ vl_api_nat64_add_del_interface_addr_t_handler
+ (vl_api_nat64_add_del_interface_addr_t * mp)
+{
+ nat64_main_t *nm = &nat64_main;
+ vl_api_nat64_add_del_interface_addr_reply_t *rmp;
+ u32 sw_if_index = ntohl (mp->sw_if_index);
+ int rv = 0;
+
+ VALIDATE_SW_IF_INDEX (mp);
+
+ rv = nat64_add_interface_address (sw_if_index, mp->is_add);
+
+ BAD_SW_IF_INDEX_LABEL;
+
+ REPLY_MACRO (VL_API_NAT64_ADD_DEL_INTERFACE_ADDR_REPLY);
+}
+
+/* API definitions */
+#include <vnet/format_fns.h>
+#include <nat/nat64/nat64.api.c>
+
+/* Set up the API message handling tables */
+clib_error_t *
+nat64_api_hookup (vlib_main_t * vm)
+{
+ nat64_main_t *nm = &nat64_main;
+ nm->msg_id_base = setup_message_id_table ();
+ return 0;
+}
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */
diff --git a/src/plugins/nat/nat64/nat64_cli.c b/src/plugins/nat/nat64/nat64_cli.c
new file mode 100644
index 00000000000..a7dd9ab9147
--- /dev/null
+++ b/src/plugins/nat/nat64/nat64_cli.c
@@ -0,0 +1,992 @@
+/*
+ * Copyright (c) 2020 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vnet/fib/fib_table.h>
+#include <nat/nat64/nat64.h>
+
+static clib_error_t *
+nat64_plugin_enable_disable_command_fn (vlib_main_t * vm,
+ unformat_input_t * input,
+ vlib_cli_command_t * cmd)
+{
+ unformat_input_t _line_input, *line_input = &_line_input;
+ u8 enable = 0, is_set = 0;
+ clib_error_t *error = 0;
+ nat64_config_t c = { 0 };
+
+ if (!unformat_user (input, unformat_line_input, line_input))
+ return 0;
+
+ while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
+ {
+ if (!is_set && unformat (line_input, "enable"))
+ {
+ unformat (line_input, "bib-buckets %u", &c.bib_buckets);
+ unformat (line_input, "bib-memory %u", &c.bib_memory_size);
+ unformat (line_input, "st-buckets %u", &c.st_buckets);
+ unformat (line_input, "st-memory %u", &c.st_memory_size);
+ enable = 1;
+ }
+ else if (!is_set && unformat (line_input, "disable"));
+ else
+ {
+ error = clib_error_return (0, "unknown input '%U'",
+ format_unformat_error, line_input);
+ goto done;
+ }
+ is_set = 1;
+ }
+
+ if (enable)
+ {
+ if (nat64_plugin_enable (c))
+ error = clib_error_return (0, "plugin enable failed");
+ }
+ else
+ {
+ if (nat64_plugin_disable ())
+ error = clib_error_return (0, "plugin disable failed");
+ }
+done:
+ unformat_free (line_input);
+ return error;
+}
+
+static clib_error_t *
+nat64_add_del_pool_addr_command_fn (vlib_main_t * vm,
+ unformat_input_t * input,
+ vlib_cli_command_t * cmd)
+{
+ unformat_input_t _line_input, *line_input = &_line_input;
+ ip4_address_t start_addr, end_addr, this_addr;
+ u32 start_host_order, end_host_order;
+ int i, count, rv;
+ u32 vrf_id = ~0;
+ u8 is_add = 1;
+ clib_error_t *error = 0;
+
+ /* Get a line of input. */
+ if (!unformat_user (input, unformat_line_input, line_input))
+ return 0;
+
+ while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
+ {
+ if (unformat (line_input, "%U - %U",
+ unformat_ip4_address, &start_addr,
+ unformat_ip4_address, &end_addr))
+ ;
+ else if (unformat (line_input, "tenant-vrf %u", &vrf_id))
+ ;
+ else if (unformat (line_input, "%U", unformat_ip4_address, &start_addr))
+ end_addr = start_addr;
+ else if (unformat (line_input, "del"))
+ is_add = 0;
+ else
+ {
+ error = clib_error_return (0, "unknown input '%U'",
+ format_unformat_error, line_input);
+ goto done;
+ }
+ }
+
+ start_host_order = clib_host_to_net_u32 (start_addr.as_u32);
+ end_host_order = clib_host_to_net_u32 (end_addr.as_u32);
+
+ if (end_host_order < start_host_order)
+ {
+ error = clib_error_return (0, "end address less than start address");
+ goto done;
+ }
+
+ count = (end_host_order - start_host_order) + 1;
+ this_addr = start_addr;
+
+ for (i = 0; i < count; i++)
+ {
+ rv = nat64_add_del_pool_addr (0, &this_addr, vrf_id, is_add);
+
+ switch (rv)
+ {
+ case VNET_API_ERROR_NO_SUCH_ENTRY:
+ error =
+ clib_error_return (0, "NAT64 pool address %U not exist.",
+ format_ip4_address, &this_addr);
+ goto done;
+ case VNET_API_ERROR_VALUE_EXIST:
+ error =
+ clib_error_return (0, "NAT64 pool address %U exist.",
+ format_ip4_address, &this_addr);
+ goto done;
+ default:
+ break;
+
+ }
+ increment_v4_address (&this_addr);
+ }
+
+done:
+ unformat_free (line_input);
+
+ return error;
+}
+
+static int
+nat64_cli_pool_walk (nat64_address_t * ap, void *ctx)
+{
+ vlib_main_t *vm = ctx;
+
+ if (ap->fib_index != ~0)
+ {
+ fib_table_t *fib;
+ fib = fib_table_get (ap->fib_index, FIB_PROTOCOL_IP6);
+ if (!fib)
+ return -1;
+ vlib_cli_output (vm, " %U tenant VRF: %u", format_ip4_address,
+ &ap->addr, fib->ft_table_id);
+ }
+ else
+ vlib_cli_output (vm, " %U", format_ip4_address, &ap->addr);
+
+#define _(N, i, n, s) \
+ vlib_cli_output (vm, " %d busy %s ports", ap->busy_##n##_ports, s);
+ foreach_nat_protocol
+#undef _
+ return 0;
+}
+
+static clib_error_t *
+nat64_show_pool_command_fn (vlib_main_t * vm,
+ unformat_input_t * input,
+ vlib_cli_command_t * cmd)
+{
+ vlib_cli_output (vm, "NAT64 pool:");
+ nat64_pool_addr_walk (nat64_cli_pool_walk, vm);
+
+ return 0;
+}
+
+static clib_error_t *
+nat64_interface_feature_command_fn (vlib_main_t * vm,
+ unformat_input_t *
+ input, vlib_cli_command_t * cmd)
+{
+ unformat_input_t _line_input, *line_input = &_line_input;
+ vnet_main_t *vnm = vnet_get_main ();
+ clib_error_t *error = 0;
+ u32 sw_if_index;
+ u32 *inside_sw_if_indices = 0;
+ u32 *outside_sw_if_indices = 0;
+ u8 is_add = 1;
+ int i, rv;
+
+ /* Get a line of input. */
+ if (!unformat_user (input, unformat_line_input, line_input))
+ return 0;
+
+ while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
+ {
+ if (unformat (line_input, "in %U", unformat_vnet_sw_interface,
+ vnm, &sw_if_index))
+ vec_add1 (inside_sw_if_indices, sw_if_index);
+ else if (unformat (line_input, "out %U", unformat_vnet_sw_interface,
+ vnm, &sw_if_index))
+ vec_add1 (outside_sw_if_indices, sw_if_index);
+ else if (unformat (line_input, "del"))
+ is_add = 0;
+ else
+ {
+ error = clib_error_return (0, "unknown input '%U'",
+ format_unformat_error, line_input);
+ goto done;
+ }
+ }
+
+ if (vec_len (inside_sw_if_indices))
+ {
+ for (i = 0; i < vec_len (inside_sw_if_indices); i++)
+ {
+ sw_if_index = inside_sw_if_indices[i];
+ rv = nat64_interface_add_del (sw_if_index, 1, is_add);
+ switch (rv)
+ {
+ case VNET_API_ERROR_NO_SUCH_ENTRY:
+ error =
+ clib_error_return (0, "%U NAT64 feature not enabled.",
+ format_vnet_sw_if_index_name, vnm,
+ sw_if_index);
+ goto done;
+ case VNET_API_ERROR_VALUE_EXIST:
+ error =
+ clib_error_return (0, "%U NAT64 feature already enabled.",
+ format_vnet_sw_if_index_name, vnm,
+ vnm, sw_if_index);
+ goto done;
+ case VNET_API_ERROR_INVALID_VALUE:
+ case VNET_API_ERROR_INVALID_VALUE_2:
+ error =
+ clib_error_return (0,
+ "%U NAT64 feature enable/disable failed.",
+ format_vnet_sw_if_index_name, vnm,
+ sw_if_index);
+ goto done;
+ default:
+ break;
+
+ }
+ }
+ }
+
+ if (vec_len (outside_sw_if_indices))
+ {
+ for (i = 0; i < vec_len (outside_sw_if_indices); i++)
+ {
+ sw_if_index = outside_sw_if_indices[i];
+ rv = nat64_interface_add_del (sw_if_index, 0, is_add);
+ switch (rv)
+ {
+ case VNET_API_ERROR_NO_SUCH_ENTRY:
+ error =
+ clib_error_return (0, "%U NAT64 feature not enabled.",
+ format_vnet_sw_if_index_name, vnm,
+ sw_if_index);
+ goto done;
+ case VNET_API_ERROR_VALUE_EXIST:
+ error =
+ clib_error_return (0, "%U NAT64 feature already enabled.",
+ format_vnet_sw_if_index_name, vnm,
+ sw_if_index);
+ goto done;
+ case VNET_API_ERROR_INVALID_VALUE:
+ case VNET_API_ERROR_INVALID_VALUE_2:
+ error =
+ clib_error_return (0,
+ "%U NAT64 feature enable/disable failed.",
+ format_vnet_sw_if_index_name, vnm,
+ sw_if_index);
+ goto done;
+ default:
+ break;
+
+ }
+ }
+ }
+
+done:
+ unformat_free (line_input);
+ vec_free (inside_sw_if_indices);
+ vec_free (outside_sw_if_indices);
+
+ return error;
+}
+
+static int
+nat64_cli_interface_walk (nat64_interface_t * i, void *ctx)
+{
+ vlib_main_t *vm = ctx;
+ vnet_main_t *vnm = vnet_get_main ();
+
+ vlib_cli_output (vm, " %U %s", format_vnet_sw_if_index_name, vnm,
+ i->sw_if_index,
+ (nat64_interface_is_inside (i)
+ && nat64_interface_is_outside (i)) ? "in out" :
+ nat64_interface_is_inside (i) ? "in" : "out");
+ return 0;
+}
+
+static clib_error_t *
+nat64_show_interfaces_command_fn (vlib_main_t * vm,
+ unformat_input_t *
+ input, vlib_cli_command_t * cmd)
+{
+ vlib_cli_output (vm, "NAT64 interfaces:");
+ nat64_interfaces_walk (nat64_cli_interface_walk, vm);
+
+ return 0;
+}
+
+static clib_error_t *
+nat64_add_del_static_bib_command_fn (vlib_main_t *
+ vm,
+ unformat_input_t
+ * input, vlib_cli_command_t * cmd)
+{
+ unformat_input_t _line_input, *line_input = &_line_input;
+ clib_error_t *error = 0;
+ u8 is_add = 1;
+ ip6_address_t in_addr;
+ ip4_address_t out_addr;
+ u32 in_port = 0;
+ u32 out_port = 0;
+ u32 vrf_id = 0, protocol;
+ nat_protocol_t proto = 0;
+ u8 p = 0;
+ int rv;
+
+ if (!unformat_user (input, unformat_line_input, line_input))
+ return 0;
+
+ while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
+ {
+ if (unformat (line_input, "%U %u", unformat_ip6_address,
+ &in_addr, &in_port))
+ ;
+ else if (unformat (line_input, "%U %u", unformat_ip4_address,
+ &out_addr, &out_port))
+ ;
+ else if (unformat (line_input, "vrf %u", &vrf_id))
+ ;
+ else if (unformat (line_input, "%U", unformat_nat_protocol, &proto))
+ ;
+ else
+ if (unformat
+ (line_input, "%U %U %u", unformat_ip6_address, &in_addr,
+ unformat_ip4_address, &out_addr, &protocol))
+ p = (u8) protocol;
+ else if (unformat (line_input, "del"))
+ is_add = 0;
+ else
+ {
+ error = clib_error_return (0, "unknown input: '%U'",
+ format_unformat_error, line_input);
+ goto done;
+ }
+ }
+
+ if (!p)
+ {
+ if (!in_port)
+ {
+ error =
+ clib_error_return (0, "inside port and address must be set");
+ goto done;
+ }
+
+ if (!out_port)
+ {
+ error =
+ clib_error_return (0, "outside port and address must be set");
+ goto done;
+ }
+
+ p = nat_proto_to_ip_proto (proto);
+ }
+
+ rv =
+ nat64_add_del_static_bib_entry (&in_addr, &out_addr, (u16) in_port,
+ (u16) out_port, p, vrf_id, is_add);
+
+ switch (rv)
+ {
+ case VNET_API_ERROR_NO_SUCH_ENTRY:
+ error = clib_error_return (0, "NAT64 BIB entry not exist.");
+ goto done;
+ case VNET_API_ERROR_VALUE_EXIST:
+ error = clib_error_return (0, "NAT64 BIB entry exist.");
+ goto done;
+ case VNET_API_ERROR_UNSPECIFIED:
+ error = clib_error_return (0, "Crerate NAT64 BIB entry failed.");
+ goto done;
+ case VNET_API_ERROR_INVALID_VALUE:
+ error =
+ clib_error_return (0,
+ "Outside address %U and port %u already in use.",
+ format_ip4_address, &out_addr, out_port);
+ goto done;
+ case VNET_API_ERROR_INVALID_VALUE_2:
+ error = clib_error_return (0, "Invalid outside port.");
+ default:
+ break;
+ }
+
+done:
+ unformat_free (line_input);
+
+ return error;
+}
+
+static int
+nat64_cli_bib_walk (nat64_db_bib_entry_t * bibe, void *ctx)
+{
+ vlib_main_t *vm = ctx;
+ fib_table_t *fib;
+
+ fib = fib_table_get (bibe->fib_index, FIB_PROTOCOL_IP6);
+ if (!fib)
+ return -1;
+
+ switch (bibe->proto)
+ {
+ case IP_PROTOCOL_ICMP:
+ case IP_PROTOCOL_TCP:
+ case IP_PROTOCOL_UDP:
+ vlib_cli_output (vm, " %U %u %U %u protocol %U vrf %u %s %u sessions",
+ format_ip6_address, &bibe->in_addr,
+ clib_net_to_host_u16 (bibe->in_port),
+ format_ip4_address, &bibe->out_addr,
+ clib_net_to_host_u16 (bibe->out_port),
+ format_nat_protocol,
+ ip_proto_to_nat_proto (bibe->proto), fib->ft_table_id,
+ bibe->is_static ? "static" : "dynamic", bibe->ses_num);
+ break;
+ default:
+ vlib_cli_output (vm, " %U %U protocol %u vrf %u %s %u sessions",
+ format_ip6_address, &bibe->in_addr,
+ format_ip4_address, &bibe->out_addr,
+ bibe->proto, fib->ft_table_id,
+ bibe->is_static ? "static" : "dynamic", bibe->ses_num);
+ }
+ return 0;
+}
+
+static clib_error_t *
+nat64_show_bib_command_fn (vlib_main_t * vm,
+ unformat_input_t * input, vlib_cli_command_t * cmd)
+{
+ nat64_main_t *nm = &nat64_main;
+ unformat_input_t _line_input, *line_input = &_line_input;
+ clib_error_t *error = 0;
+ u32 proto = NAT_PROTOCOL_OTHER;
+ u8 p = 255;
+ nat64_db_t *db;
+
+ if (!unformat_user (input, unformat_line_input, line_input))
+ return 0;
+
+ if (unformat (line_input, "%U", unformat_nat_protocol, &proto))
+ p = nat_proto_to_ip_proto (proto);
+ else if (unformat (line_input, "unknown"))
+ p = 0;
+ else if (unformat (line_input, "all"))
+ ;
+ else
+ {
+ error = clib_error_return (0, "unknown input: '%U'",
+ format_unformat_error, line_input);
+ goto done;
+ }
+
+ if (p == 255)
+ vlib_cli_output (vm, "NAT64 BIB entries:");
+ else
+ vlib_cli_output (vm, "NAT64 %U BIB entries:", format_nat_protocol, proto);
+
+ /* *INDENT-OFF* */
+ vec_foreach (db, nm->db)
+ nat64_db_bib_walk (db, p, nat64_cli_bib_walk, vm);
+ /* *INDENT-ON* */
+
+done:
+ unformat_free (line_input);
+
+ return error;
+}
+
+typedef struct nat64_cli_st_walk_ctx_t_
+{
+ vlib_main_t *vm;
+ nat64_db_t *db;
+} nat64_cli_st_walk_ctx_t;
+
+static int
+nat64_cli_st_walk (nat64_db_st_entry_t * ste, void *arg)
+{
+ nat64_cli_st_walk_ctx_t *ctx = arg;
+ vlib_main_t *vm = ctx->vm;
+ nat64_db_bib_entry_t *bibe;
+ fib_table_t *fib;
+
+ bibe = nat64_db_bib_entry_by_index (ctx->db, ste->proto, ste->bibe_index);
+ if (!bibe)
+ return -1;
+
+ fib = fib_table_get (bibe->fib_index, FIB_PROTOCOL_IP6);
+ if (!fib)
+ return -1;
+
+ u32 vrf_id = fib->ft_table_id;
+
+ if (ste->proto == IP_PROTOCOL_ICMP)
+ vlib_cli_output (vm, " %U %U %u %U %U %u protocol %U vrf %u",
+ format_ip6_address, &bibe->in_addr,
+ format_ip6_address, &ste->in_r_addr,
+ clib_net_to_host_u16 (bibe->in_port),
+ format_ip4_address, &bibe->out_addr,
+ format_ip4_address, &ste->out_r_addr,
+ clib_net_to_host_u16 (bibe->out_port),
+ format_nat_protocol,
+ ip_proto_to_nat_proto (bibe->proto), vrf_id);
+ else if (ste->proto == IP_PROTOCOL_TCP || ste->proto == IP_PROTOCOL_UDP)
+ vlib_cli_output (vm, " %U %u %U %u %U %u %U %u protcol %U vrf %u",
+ format_ip6_address, &bibe->in_addr,
+ clib_net_to_host_u16 (bibe->in_port),
+ format_ip6_address, &ste->in_r_addr,
+ clib_net_to_host_u16 (ste->r_port),
+ format_ip4_address, &bibe->out_addr,
+ clib_net_to_host_u16 (bibe->out_port),
+ format_ip4_address, &ste->out_r_addr,
+ clib_net_to_host_u16 (ste->r_port),
+ format_nat_protocol,
+ ip_proto_to_nat_proto (bibe->proto), vrf_id);
+ else
+ vlib_cli_output (vm, " %U %U %U %U protocol %u vrf %u",
+ format_ip6_address, &bibe->in_addr,
+ format_ip6_address, &ste->in_r_addr,
+ format_ip4_address, &bibe->out_addr,
+ format_ip4_address, &ste->out_r_addr,
+ bibe->proto, vrf_id);
+
+ return 0;
+}
+
+static clib_error_t *
+nat64_show_st_command_fn (vlib_main_t * vm,
+ unformat_input_t * input, vlib_cli_command_t * cmd)
+{
+ nat64_main_t *nm = &nat64_main;
+ unformat_input_t _line_input, *line_input = &_line_input;
+ clib_error_t *error = 0;
+ u32 proto = NAT_PROTOCOL_OTHER;
+ u8 p = 255;
+ nat64_db_t *db;
+ nat64_cli_st_walk_ctx_t ctx = {
+ .vm = vm,
+ };
+
+ if (!unformat_user (input, unformat_line_input, line_input))
+ return 0;
+
+ if (unformat (line_input, "%U", unformat_nat_protocol, &proto))
+ p = nat_proto_to_ip_proto (proto);
+ else if (unformat (line_input, "unknown"))
+ p = 0;
+ else if (unformat (line_input, "all"))
+ ;
+ else
+ {
+ error = clib_error_return (0, "unknown input: '%U'",
+ format_unformat_error, line_input);
+ goto done;
+ }
+
+ if (p == 255)
+ vlib_cli_output (vm, "NAT64 sessions:");
+ else
+ vlib_cli_output (vm, "NAT64 %U sessions:", format_nat_protocol, proto);
+ /* *INDENT-OFF* */
+ vec_foreach (db, nm->db)
+ {
+ ctx.db = db;
+ nat64_db_st_walk (db, p, nat64_cli_st_walk, &ctx);
+ }
+ /* *INDENT-ON* */
+
+done:
+ unformat_free (line_input);
+
+ return error;
+}
+
+static clib_error_t *
+nat64_add_del_prefix_command_fn (vlib_main_t * vm, unformat_input_t * input,
+ vlib_cli_command_t * cmd)
+{
+ nat64_main_t *nm = &nat64_main;
+ vnet_main_t *vnm = vnet_get_main ();
+ clib_error_t *error = 0;
+ unformat_input_t _line_input, *line_input = &_line_input;
+ u8 is_add = 1;
+ u32 vrf_id = 0, sw_if_index = ~0;
+ ip6_address_t prefix;
+ u32 plen = 0;
+ int rv;
+
+ if (!unformat_user (input, unformat_line_input, line_input))
+ return 0;
+
+ while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
+ {
+ if (unformat
+ (line_input, "%U/%u", unformat_ip6_address, &prefix, &plen))
+ ;
+ else if (unformat (line_input, "tenant-vrf %u", &vrf_id))
+ ;
+ else if (unformat (line_input, "del"))
+ is_add = 0;
+ else
+ if (unformat
+ (line_input, "interface %U", unformat_vnet_sw_interface, vnm,
+ &sw_if_index))
+ ;
+ else
+ {
+ error = clib_error_return (0, "unknown input: '%U'",
+ format_unformat_error, line_input);
+ goto done;
+ }
+ }
+
+ if (!plen)
+ {
+ error = clib_error_return (0, "NAT64 prefix must be set.");
+ goto done;
+ }
+
+ rv = nat64_add_del_prefix (&prefix, (u8) plen, vrf_id, is_add);
+
+ switch (rv)
+ {
+ case VNET_API_ERROR_NO_SUCH_ENTRY:
+ error = clib_error_return (0, "NAT64 prefix not exist.");
+ goto done;
+ case VNET_API_ERROR_INVALID_VALUE:
+ error = clib_error_return (0, "Invalid prefix length.");
+ goto done;
+ default:
+ break;
+ }
+
+ /*
+ * Add RX interface route, whenNAT isn't running on the real input
+ * interface
+ */
+ if (sw_if_index != ~0)
+ {
+ u32 fib_index;
+ fib_prefix_t fibpfx = {
+ .fp_len = plen,
+ .fp_proto = FIB_PROTOCOL_IP6,
+ .fp_addr = {
+ .ip6 = prefix}
+ };
+
+ if (is_add)
+ {
+ fib_index =
+ fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP6,
+ vrf_id, nm->fib_src_hi);
+ fib_table_entry_update_one_path (fib_index, &fibpfx,
+ nm->fib_src_hi,
+ FIB_ENTRY_FLAG_NONE,
+ DPO_PROTO_IP6, NULL,
+ sw_if_index, ~0, 0,
+ NULL, FIB_ROUTE_PATH_INTF_RX);
+ }
+ else
+ {
+ fib_index = fib_table_find (FIB_PROTOCOL_IP6, vrf_id);
+ fib_table_entry_path_remove (fib_index, &fibpfx,
+ nm->fib_src_hi,
+ DPO_PROTO_IP6, NULL,
+ sw_if_index, ~0, 1,
+ FIB_ROUTE_PATH_INTF_RX);
+ fib_table_unlock (fib_index, FIB_PROTOCOL_IP6, nm->fib_src_hi);
+ }
+ }
+
+done:
+ unformat_free (line_input);
+
+ return error;
+}
+
+static int
+nat64_cli_prefix_walk (nat64_prefix_t * p, void *ctx)
+{
+ vlib_main_t *vm = ctx;
+
+ vlib_cli_output (vm, " %U/%u tenant-vrf %u",
+ format_ip6_address, &p->prefix, p->plen, p->vrf_id);
+
+ return 0;
+}
+
+static clib_error_t *
+nat64_show_prefix_command_fn (vlib_main_t * vm,
+ unformat_input_t * input,
+ vlib_cli_command_t * cmd)
+{
+ vlib_cli_output (vm, "NAT64 prefix:");
+ nat64_prefix_walk (nat64_cli_prefix_walk, vm);
+
+ return 0;
+}
+
+static clib_error_t *
+nat64_add_interface_address_command_fn (vlib_main_t * vm,
+ unformat_input_t * input,
+ vlib_cli_command_t * cmd)
+{
+ vnet_main_t *vnm = vnet_get_main ();
+ unformat_input_t _line_input, *line_input = &_line_input;
+ u32 sw_if_index;
+ int rv;
+ int is_add = 1;
+ clib_error_t *error = 0;
+
+ /* Get a line of input. */
+ if (!unformat_user (input, unformat_line_input, line_input))
+ return 0;
+
+ while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT)
+ {
+ if (unformat
+ (line_input, "%U", unformat_vnet_sw_interface, vnm, &sw_if_index));
+ else if (unformat (line_input, "del"))
+ is_add = 0;
+ else
+ {
+ error = clib_error_return (0, "unknown input '%U'",
+ format_unformat_error, line_input);
+ goto done;
+ }
+ }
+
+ rv = nat64_add_interface_address (sw_if_index, is_add);
+
+ switch (rv)
+ {
+ case VNET_API_ERROR_NO_SUCH_ENTRY:
+ error = clib_error_return (0, "entry not exist");
+ break;
+ case VNET_API_ERROR_VALUE_EXIST:
+ error = clib_error_return (0, "entry exist");
+ break;
+ default:
+ break;
+ }
+
+done:
+ unformat_free (line_input);
+
+ return error;
+}
+
+/* *INDENT-OFF* */
+/*?
+ * @cliexpar
+ * @cliexstart{nat64 plugin}
+ * Enable/disable NAT64 plugin.
+ * To enable NAT64 plugin use:
+ * vpp# nat64 plugin enable
+ * To enable NAT64 plugin and configure buckets/memory:
+ * vpp# nat64 plugin enable bib-buckets <n> bib-memory <s> \
+ * st-buckets <n> st-memory <s>
+ * To disable NAT64 plugin:
+ * vpp# nat64 plugin disable
+ * @cliexend
+?*/
+VLIB_CLI_COMMAND (nat64_plugin_enable_disable_command, static) =
+{
+ .path = "nat64 plugin",
+ .short_help = "nat64 plugin <enable "
+ "[bib-buckets <count>] [bib-memory <size>] "
+ "[st-buckets <count>] [st-memory <size>] | disable>",
+ .function = nat64_plugin_enable_disable_command_fn,
+};
+
+/*?
+ * @cliexpar
+ * @cliexstart{nat64 add pool address}
+ * Add/delete NAT64 pool address.
+ * To add single NAT64 pool address use:
+ * vpp# nat64 add pool address 10.1.1.10
+ * To add NAT64 pool address range use:
+ * vpp# nat64 add pool address 10.1.1.2 - 10.1.1.5
+ * To add NAT64 pool address for specific tenant use:
+ * vpp# nat64 add pool address 10.1.1.100 tenant-vrf 100
+ * @cliexend
+?*/
+VLIB_CLI_COMMAND (nat64_add_pool_address_command, static) = {
+ .path = "nat64 add pool address",
+ .short_help = "nat64 add pool address <ip4-range-start> [- <ip4-range-end>] "
+ "[tenant-vrf <vrf-id>] [del]",
+ .function = nat64_add_del_pool_addr_command_fn,
+};
+
+/*?
+ * @cliexpar
+ * @cliexstart{show nat64 pool}
+ * Show NAT64 pool.
+ * vpp# show nat64 pool
+ * NAT64 pool:
+ * 10.1.1.3 tenant VRF: 0
+ * 10.1.1.10 tenant VRF: 10
+ * @cliexend
+?*/
+VLIB_CLI_COMMAND (show_nat64_pool_command, static) = {
+ .path = "show nat64 pool",
+ .short_help = "show nat64 pool",
+ .function = nat64_show_pool_command_fn,
+};
+
+/*?
+ * @cliexpar
+ * @cliexstart{set interface nat64}
+ * Enable/disable NAT64 feature on the interface.
+ * To enable NAT64 feature with local (IPv6) network interface
+ * GigabitEthernet0/8/0 and external (IPv4) network interface
+ * GigabitEthernet0/a/0 use:
+ * vpp# set interface nat64 in GigabitEthernet0/8/0 out GigabitEthernet0/a/0
+ * @cliexend
+?*/
+VLIB_CLI_COMMAND (set_interface_nat64_command, static) = {
+ .path = "set interface nat64",
+ .short_help = "set interface nat64 in|out <intfc> [del]",
+ .function = nat64_interface_feature_command_fn,
+};
+
+/*?
+ * @cliexpar
+ * @cliexstart{show nat64 interfaces}
+ * Show interfaces with NAT64 feature.
+ * To show interfaces with NAT64 feature use:
+ * vpp# show nat64 interfaces
+ * NAT64 interfaces:
+ * GigabitEthernet0/8/0 in
+ * GigabitEthernet0/a/0 out
+ * @cliexend
+?*/
+VLIB_CLI_COMMAND (show_nat64_interfaces_command, static) = {
+ .path = "show nat64 interfaces",
+ .short_help = "show nat64 interfaces",
+ .function = nat64_show_interfaces_command_fn,
+};
+
+/*?
+ * @cliexpar
+ * @cliexstart{nat64 add static bib}
+ * Add/delete NAT64 static BIB entry.
+ * To create NAT64 satatic BIB entry use:
+ * vpp# nat64 add static bib 2001:db8:c000:221:: 1234 10.1.1.3 5678 tcp
+ * vpp# nat64 add static bib 2001:db8:c000:221:: 1234 10.1.1.3 5678 udp vrf 10
+ * @cliexend
+?*/
+VLIB_CLI_COMMAND (nat64_add_del_static_bib_command, static) = {
+ .path = "nat64 add static bib",
+ .short_help = "nat64 add static bib <ip6-addr> <port> <ip4-addr> <port> "
+ "tcp|udp|icmp [vfr <table-id>] [del]",
+ .function = nat64_add_del_static_bib_command_fn,
+};
+
+/*?
+ * @cliexpar
+ * @cliexstart{show nat64 bib}
+ * Show NAT64 BIB entries.
+ * To show NAT64 TCP BIB entries use:
+ * vpp# show nat64 bib tcp
+ * NAT64 tcp BIB:
+ * fd01:1::2 6303 10.0.0.3 62303 tcp vrf 0 dynamic 1 sessions
+ * 2001:db8:c000:221:: 1234 10.1.1.3 5678 tcp vrf 0 static 2 sessions
+ * To show NAT64 UDP BIB entries use:
+ * vpp# show nat64 bib udp
+ * NAT64 udp BIB:
+ * fd01:1::2 6304 10.0.0.3 10546 udp vrf 0 dynamic 10 sessions
+ * 2001:db8:c000:221:: 1234 10.1.1.3 5678 udp vrf 10 static 0 sessions
+ * To show NAT64 ICMP BIB entries use:
+ * vpp# show nat64 bib icmp
+ * NAT64 icmp BIB:
+ * fd01:1::2 6305 10.0.0.3 63209 icmp vrf 10 dynamic 1 sessions
+ * @cliexend
+?*/
+VLIB_CLI_COMMAND (show_nat64_bib_command, static) = {
+ .path = "show nat64 bib",
+ .short_help = "show nat64 bib all|tcp|udp|icmp|unknown",
+ .function = nat64_show_bib_command_fn,
+};
+
+/*?
+ * @cliexpar
+ * @cliexstart{show nat64 session table}
+ * Show NAT64 session table.
+ * To show NAT64 TCP session table use:
+ * vpp# show nat64 session table tcp
+ * NAT64 tcp session table:
+ * fd01:1::2 6303 64:ff9b::ac10:202 20 10.0.0.3 62303 172.16.2.2 20 tcp vrf 0
+ * fd01:3::2 6303 64:ff9b::ac10:202 20 10.0.10.3 21300 172.16.2.2 20 tcp vrf 10
+ * To show NAT64 UDP session table use:
+ * #vpp show nat64 session table udp
+ * NAT64 udp session table:
+ * fd01:1::2 6304 64:ff9b::ac10:202 20 10.0.0.3 10546 172.16.2.2 20 udp vrf 0
+ * fd01:3::2 6304 64:ff9b::ac10:202 20 10.0.10.3 58627 172.16.2.2 20 udp vrf 10
+ * fd01:1::2 1235 64:ff9b::a00:3 4023 10.0.0.3 24488 10.0.0.3 4023 udp vrf 0
+ * fd01:1::3 23 64:ff9b::a00:3 24488 10.0.0.3 4023 10.0.0.3 24488 udp vrf 0
+ * To show NAT64 ICMP session table use:
+ * #vpp show nat64 session table icmp
+ * NAT64 icmp session table:
+ * fd01:1::2 64:ff9b::ac10:202 6305 10.0.0.3 172.16.2.2 63209 icmp vrf 0
+ * @cliexend
+?*/
+VLIB_CLI_COMMAND (show_nat64_st_command, static) = {
+ .path = "show nat64 session table",
+ .short_help = "show nat64 session table all|tcp|udp|icmp|unknown",
+ .function = nat64_show_st_command_fn,
+};
+
+/*?
+ * @cliexpar
+ * @cliexstart{nat64 add prefix}
+ * Set NAT64 prefix for generating IPv6 representations of IPv4 addresses.
+ * To set NAT64 global prefix use:
+ * vpp# nat64 add prefix 2001:db8::/32
+ * To set NAT64 prefix for specific tenant use:
+ * vpp# nat64 add prefix 2001:db8:122:300::/56 tenant-vrf 10
+ * @cliexend
+?*/
+VLIB_CLI_COMMAND (nat64_add_del_prefix_command, static) = {
+ .path = "nat64 add prefix",
+ .short_help = "nat64 add prefix <ip6-prefix>/<plen> [tenant-vrf <vrf-id>] "
+ "[del] [interface <interface]",
+ .function = nat64_add_del_prefix_command_fn,
+};
+
+/*?
+ * @cliexpar
+ * @cliexstart{show nat64 prefix}
+ * Show NAT64 prefix.
+ * To show NAT64 prefix use:
+ * vpp# show nat64 prefix
+ * NAT64 prefix:
+ * 2001:db8::/32 tenant-vrf 0
+ * 2001:db8:122:300::/56 tenant-vrf 10
+ * @cliexend
+?*/
+VLIB_CLI_COMMAND (show_nat64_prefix_command, static) = {
+ .path = "show nat64 prefix",
+ .short_help = "show nat64 prefix",
+ .function = nat64_show_prefix_command_fn,
+};
+
+/*?
+ * @cliexpar
+ * @cliexstart{nat64 add interface address}
+ * Add/delete NAT64 pool address from specific (DHCP addressed) interface.
+ * To add NAT64 pool address from specific interface use:
+ * vpp# nat64 add interface address GigabitEthernet0/8/0
+ * @cliexend
+?*/
+VLIB_CLI_COMMAND (nat64_add_interface_address_command, static) = {
+ .path = "nat64 add interface address",
+ .short_help = "nat64 add interface address <interface> [del]",
+ .function = nat64_add_interface_address_command_fn,
+};
+/* *INDENT-ON* */
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */
diff --git a/src/plugins/nat/nat64/nat64_db.c b/src/plugins/nat/nat64/nat64_db.c
new file mode 100644
index 00000000000..ffc5e7e7e84
--- /dev/null
+++ b/src/plugins/nat/nat64/nat64_db.c
@@ -0,0 +1,742 @@
+/*
+ * Copyright (c) 2020 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <vnet/fib/fib_table.h>
+//#include <nat/nat_ipfix_logging.h>
+#include <nat/nat_syslog.h>
+#include <nat/lib/inlines.h>
+#include <nat/nat64/nat64_db.h>
+
+int
+nat64_db_init (nat64_db_t * db, nat64_config_t c,
+ nat64_db_free_addr_port_function_t free_addr_port_cb)
+{
+ clib_bihash_init_24_8 (&db->bib.in2out, "bib-in2out", c.bib_buckets,
+ c.bib_memory_size);
+
+ clib_bihash_init_24_8 (&db->bib.out2in, "bib-out2in", c.bib_buckets,
+ c.bib_memory_size);
+
+ clib_bihash_init_48_8 (&db->st.in2out, "st-in2out", c.st_buckets,
+ c.st_memory_size);
+
+ clib_bihash_init_48_8 (&db->st.out2in, "st-out2in", c.st_buckets,
+ c.st_memory_size);
+
+ db->free_addr_port_cb = free_addr_port_cb;
+ db->bib.limit = 10 * c.bib_buckets;
+ db->bib.bib_entries_num = 0;
+ db->st.limit = 10 * c.st_buckets;
+ db->st.st_entries_num = 0;
+ db->addr_free = 0;
+
+ return 0;
+}
+
+int
+nat64_db_free (nat64_db_t * db)
+{
+ clib_bihash_free_24_8 (&db->bib.in2out);
+ clib_bihash_free_24_8 (&db->bib.out2in);
+
+ clib_bihash_free_48_8 (&db->st.in2out);
+ clib_bihash_free_48_8 (&db->st.out2in);
+
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ pool_free (db->bib._##n##_bib); \
+ pool_free (db->st._##n##_st);
+ foreach_nat_protocol
+#undef _
+/* *INDENT-ON* */
+
+ pool_free (db->bib._unk_proto_bib);
+ pool_free (db->st._unk_proto_st);
+
+ return 0;
+}
+
+nat64_db_bib_entry_t *
+nat64_db_bib_entry_create (u32 thread_index, nat64_db_t * db,
+ ip6_address_t * in_addr,
+ ip4_address_t * out_addr, u16 in_port,
+ u16 out_port, u32 fib_index, u8 proto,
+ u8 is_static)
+{
+ nat64_db_bib_entry_t *bibe;
+ nat64_db_bib_entry_key_t bibe_key;
+ clib_bihash_kv_24_8_t kv;
+
+ if (db->bib.bib_entries_num >= db->bib.limit)
+ {
+ db->free_addr_port_cb (db, out_addr, out_port, proto);
+ //nat_ipfix_logging_max_bibs (thread_index, db->bib.limit);
+ return 0;
+ }
+
+ /* create pool entry */
+ switch (ip_proto_to_nat_proto (proto))
+ {
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ case NAT_PROTOCOL_##N: \
+ pool_get (db->bib._##n##_bib, bibe); \
+ kv.value = bibe - db->bib._##n##_bib; \
+ break;
+ foreach_nat_protocol
+#undef _
+/* *INDENT-ON* */
+ default:
+ pool_get (db->bib._unk_proto_bib, bibe);
+ kv.value = bibe - db->bib._unk_proto_bib;
+ break;
+ }
+
+ db->bib.bib_entries_num++;
+
+ clib_memset (bibe, 0, sizeof (*bibe));
+ bibe->in_addr.as_u64[0] = in_addr->as_u64[0];
+ bibe->in_addr.as_u64[1] = in_addr->as_u64[1];
+ bibe->in_port = in_port;
+ bibe->out_addr.as_u32 = out_addr->as_u32;
+ bibe->out_port = out_port;
+ bibe->fib_index = fib_index;
+ bibe->proto = proto;
+ bibe->is_static = is_static;
+
+ /* create hash lookup */
+ bibe_key.addr.as_u64[0] = bibe->in_addr.as_u64[0];
+ bibe_key.addr.as_u64[1] = bibe->in_addr.as_u64[1];
+ bibe_key.fib_index = bibe->fib_index;
+ bibe_key.port = bibe->in_port;
+ bibe_key.proto = bibe->proto;
+ bibe_key.rsvd = 0;
+ kv.key[0] = bibe_key.as_u64[0];
+ kv.key[1] = bibe_key.as_u64[1];
+ kv.key[2] = bibe_key.as_u64[2];
+ clib_bihash_add_del_24_8 (&db->bib.in2out, &kv, 1);
+
+ clib_memset (&bibe_key.addr, 0, sizeof (bibe_key.addr));
+ bibe_key.addr.ip4.as_u32 = bibe->out_addr.as_u32;
+ bibe_key.fib_index = 0;
+ bibe_key.port = bibe->out_port;
+ kv.key[0] = bibe_key.as_u64[0];
+ kv.key[1] = bibe_key.as_u64[1];
+ kv.key[2] = bibe_key.as_u64[2];
+ clib_bihash_add_del_24_8 (&db->bib.out2in, &kv, 1);
+
+ /*fib_table_t *fib = fib_table_get (bibe->fib_index, FIB_PROTOCOL_IP6);
+ nat_ipfix_logging_nat64_bib (thread_index, in_addr, out_addr, proto,
+ in_port, out_port, fib->ft_table_id, 1); */
+ return bibe;
+}
+
+void
+nat64_db_bib_entry_free (u32 thread_index, nat64_db_t * db,
+ nat64_db_bib_entry_t * bibe)
+{
+ nat64_db_bib_entry_key_t bibe_key;
+ clib_bihash_kv_24_8_t kv;
+ nat64_db_bib_entry_t *bib;
+ u32 *ste_to_be_free = 0, *ste_index, bibe_index;
+ nat64_db_st_entry_t *st, *ste;
+
+ switch (ip_proto_to_nat_proto (bibe->proto))
+ {
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ case NAT_PROTOCOL_##N: \
+ bib = db->bib._##n##_bib; \
+ st = db->st._##n##_st; \
+ break;
+ foreach_nat_protocol
+#undef _
+/* *INDENT-ON* */
+ default:
+ bib = db->bib._unk_proto_bib;
+ st = db->st._unk_proto_st;
+ break;
+ }
+
+ db->bib.bib_entries_num--;
+
+ bibe_index = bibe - bib;
+
+ /* delete ST entries for static BIB entry */
+ if (bibe->is_static)
+ {
+ pool_foreach (ste, st, (
+ {
+ if (ste->bibe_index == bibe_index)
+ vec_add1 (ste_to_be_free, ste - st);}
+ ));
+ vec_foreach (ste_index, ste_to_be_free)
+ nat64_db_st_entry_free (thread_index, db,
+ pool_elt_at_index (st, ste_index[0]));
+ vec_free (ste_to_be_free);
+ }
+
+ /* delete hash lookup */
+ bibe_key.addr.as_u64[0] = bibe->in_addr.as_u64[0];
+ bibe_key.addr.as_u64[1] = bibe->in_addr.as_u64[1];
+ bibe_key.fib_index = bibe->fib_index;
+ bibe_key.port = bibe->in_port;
+ bibe_key.proto = bibe->proto;
+ bibe_key.rsvd = 0;
+ kv.key[0] = bibe_key.as_u64[0];
+ kv.key[1] = bibe_key.as_u64[1];
+ kv.key[2] = bibe_key.as_u64[2];
+ clib_bihash_add_del_24_8 (&db->bib.in2out, &kv, 0);
+
+ clib_memset (&bibe_key.addr, 0, sizeof (bibe_key.addr));
+ bibe_key.addr.ip4.as_u32 = bibe->out_addr.as_u32;
+ bibe_key.fib_index = 0;
+ bibe_key.port = bibe->out_port;
+ kv.key[0] = bibe_key.as_u64[0];
+ kv.key[1] = bibe_key.as_u64[1];
+ kv.key[2] = bibe_key.as_u64[2];
+ clib_bihash_add_del_24_8 (&db->bib.out2in, &kv, 0);
+
+ if (!db->addr_free)
+ db->free_addr_port_cb (db, &bibe->out_addr, bibe->out_port, bibe->proto);
+
+ /*fib_table_t *fib = fib_table_get (bibe->fib_index, FIB_PROTOCOL_IP6);
+ nat_ipfix_logging_nat64_bib (thread_index, &bibe->in_addr, &bibe->out_addr,
+ bibe->proto, bibe->in_port, bibe->out_port,
+ fib->ft_table_id, 0); */
+
+ /* delete from pool */
+ pool_put (bib, bibe);
+}
+
+nat64_db_bib_entry_t *
+nat64_db_bib_entry_find (nat64_db_t * db, ip46_address_t * addr, u16 port,
+ u8 proto, u32 fib_index, u8 is_ip6)
+{
+ nat64_db_bib_entry_t *bibe = 0;
+ nat64_db_bib_entry_key_t bibe_key;
+ clib_bihash_kv_24_8_t kv, value;
+ nat64_db_bib_entry_t *bib;
+
+ switch (ip_proto_to_nat_proto (proto))
+ {
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ case NAT_PROTOCOL_##N: \
+ bib = db->bib._##n##_bib; \
+ break;
+ foreach_nat_protocol
+#undef _
+/* *INDENT-ON* */
+ default:
+ bib = db->bib._unk_proto_bib;
+ break;
+ }
+
+ bibe_key.addr.as_u64[0] = addr->as_u64[0];
+ bibe_key.addr.as_u64[1] = addr->as_u64[1];
+ bibe_key.fib_index = fib_index;
+ bibe_key.port = port;
+ bibe_key.proto = proto;
+ bibe_key.rsvd = 0;
+
+ kv.key[0] = bibe_key.as_u64[0];
+ kv.key[1] = bibe_key.as_u64[1];
+ kv.key[2] = bibe_key.as_u64[2];
+
+ if (!clib_bihash_search_24_8
+ (is_ip6 ? &db->bib.in2out : &db->bib.out2in, &kv, &value))
+ bibe = pool_elt_at_index (bib, value.value);
+
+ return bibe;
+}
+
+void
+nat64_db_bib_walk (nat64_db_t * db, u8 proto,
+ nat64_db_bib_walk_fn_t fn, void *ctx)
+{
+ nat64_db_bib_entry_t *bib, *bibe;
+
+ if (proto == 255)
+ {
+ /* *INDENT-OFF* */
+ #define _(N, i, n, s) \
+ bib = db->bib._##n##_bib; \
+ pool_foreach (bibe, bib, ({ \
+ if (fn (bibe, ctx)) \
+ return; \
+ }));
+ foreach_nat_protocol
+ #undef _
+ bib = db->bib._unk_proto_bib;
+ pool_foreach (bibe, bib, ({
+ if (fn (bibe, ctx))
+ return;
+ }));
+ /* *INDENT-ON* */
+ }
+ else
+ {
+ switch (ip_proto_to_nat_proto (proto))
+ {
+ /* *INDENT-OFF* */
+ #define _(N, i, n, s) \
+ case NAT_PROTOCOL_##N: \
+ bib = db->bib._##n##_bib; \
+ break;
+ foreach_nat_protocol
+ #undef _
+ /* *INDENT-ON* */
+ default:
+ bib = db->bib._unk_proto_bib;
+ break;
+ }
+
+ /* *INDENT-OFF* */
+ pool_foreach (bibe, bib,
+ ({
+ if (fn (bibe, ctx))
+ return;
+ }));
+ /* *INDENT-ON* */
+ }
+}
+
+nat64_db_bib_entry_t *
+nat64_db_bib_entry_by_index (nat64_db_t * db, u8 proto, u32 bibe_index)
+{
+ nat64_db_bib_entry_t *bib;
+
+ switch (ip_proto_to_nat_proto (proto))
+ {
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ case NAT_PROTOCOL_##N: \
+ bib = db->bib._##n##_bib; \
+ break;
+ foreach_nat_protocol
+#undef _
+/* *INDENT-ON* */
+ default:
+ bib = db->bib._unk_proto_bib;
+ break;
+ }
+
+ return pool_elt_at_index (bib, bibe_index);
+}
+
+void
+nat64_db_st_walk (nat64_db_t * db, u8 proto,
+ nat64_db_st_walk_fn_t fn, void *ctx)
+{
+ nat64_db_st_entry_t *st, *ste;
+
+ if (proto == 255)
+ {
+ /* *INDENT-OFF* */
+ #define _(N, i, n, s) \
+ st = db->st._##n##_st; \
+ pool_foreach (ste, st, ({ \
+ if (fn (ste, ctx)) \
+ return; \
+ }));
+ foreach_nat_protocol
+ #undef _
+ st = db->st._unk_proto_st;
+ pool_foreach (ste, st, ({
+ if (fn (ste, ctx))
+ return;
+ }));
+ /* *INDENT-ON* */
+ }
+ else
+ {
+ switch (ip_proto_to_nat_proto (proto))
+ {
+ /* *INDENT-OFF* */
+ #define _(N, i, n, s) \
+ case NAT_PROTOCOL_##N: \
+ st = db->st._##n##_st; \
+ break;
+ foreach_nat_protocol
+ #undef _
+ /* *INDENT-ON* */
+ default:
+ st = db->st._unk_proto_st;
+ break;
+ }
+
+ /* *INDENT-OFF* */
+ pool_foreach (ste, st,
+ ({
+ if (fn (ste, ctx))
+ return;
+ }));
+ /* *INDENT-ON* */
+ }
+}
+
+nat64_db_st_entry_t *
+nat64_db_st_entry_create (u32 thread_index, nat64_db_t * db,
+ nat64_db_bib_entry_t * bibe,
+ ip6_address_t * in_r_addr,
+ ip4_address_t * out_r_addr, u16 r_port)
+{
+ nat64_db_st_entry_t *ste;
+ nat64_db_bib_entry_t *bib;
+ nat64_db_st_entry_key_t ste_key;
+ clib_bihash_kv_48_8_t kv;
+
+ if (db->st.st_entries_num >= db->st.limit)
+ {
+ //nat_ipfix_logging_max_sessions (thread_index, db->st.limit);
+ return 0;
+ }
+
+ /* create pool entry */
+ switch (ip_proto_to_nat_proto (bibe->proto))
+ {
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ case NAT_PROTOCOL_##N: \
+ pool_get (db->st._##n##_st, ste); \
+ kv.value = ste - db->st._##n##_st; \
+ bib = db->bib._##n##_bib; \
+ break;
+ foreach_nat_protocol
+#undef _
+/* *INDENT-ON* */
+ default:
+ pool_get (db->st._unk_proto_st, ste);
+ kv.value = ste - db->st._unk_proto_st;
+ bib = db->bib._unk_proto_bib;
+ break;
+ }
+
+ db->st.st_entries_num++;
+
+ clib_memset (ste, 0, sizeof (*ste));
+ ste->in_r_addr.as_u64[0] = in_r_addr->as_u64[0];
+ ste->in_r_addr.as_u64[1] = in_r_addr->as_u64[1];
+ ste->out_r_addr.as_u32 = out_r_addr->as_u32;
+ ste->r_port = r_port;
+ ste->bibe_index = bibe - bib;
+ ste->proto = bibe->proto;
+
+ /* increment session number for BIB entry */
+ bibe->ses_num++;
+
+ /* create hash lookup */
+ clib_memset (&ste_key, 0, sizeof (ste_key));
+ ste_key.l_addr.as_u64[0] = bibe->in_addr.as_u64[0];
+ ste_key.l_addr.as_u64[1] = bibe->in_addr.as_u64[1];
+ ste_key.r_addr.as_u64[0] = ste->in_r_addr.as_u64[0];
+ ste_key.r_addr.as_u64[1] = ste->in_r_addr.as_u64[1];
+ ste_key.fib_index = bibe->fib_index;
+ ste_key.l_port = bibe->in_port;
+ ste_key.r_port = ste->r_port;
+ ste_key.proto = ste->proto;
+ kv.key[0] = ste_key.as_u64[0];
+ kv.key[1] = ste_key.as_u64[1];
+ kv.key[2] = ste_key.as_u64[2];
+ kv.key[3] = ste_key.as_u64[3];
+ kv.key[4] = ste_key.as_u64[4];
+ kv.key[5] = ste_key.as_u64[5];
+ clib_bihash_add_del_48_8 (&db->st.in2out, &kv, 1);
+
+ clib_memset (&ste_key, 0, sizeof (ste_key));
+ ste_key.l_addr.ip4.as_u32 = bibe->out_addr.as_u32;
+ ste_key.r_addr.ip4.as_u32 = ste->out_r_addr.as_u32;
+ ste_key.l_port = bibe->out_port;
+ ste_key.r_port = ste->r_port;
+ ste_key.proto = ste->proto;
+ kv.key[0] = ste_key.as_u64[0];
+ kv.key[1] = ste_key.as_u64[1];
+ kv.key[2] = ste_key.as_u64[2];
+ kv.key[3] = ste_key.as_u64[3];
+ kv.key[4] = ste_key.as_u64[4];
+ kv.key[5] = ste_key.as_u64[5];
+ clib_bihash_add_del_48_8 (&db->st.out2in, &kv, 1);
+
+ /*fib_table_t *fib = fib_table_get (bibe->fib_index, FIB_PROTOCOL_IP6);
+ nat_ipfix_logging_nat64_session (thread_index, &bibe->in_addr,
+ &bibe->out_addr, bibe->proto,
+ bibe->in_port, bibe->out_port,
+ &ste->in_r_addr, &ste->out_r_addr,
+ ste->r_port, ste->r_port, fib->ft_table_id,
+ 1); */
+ nat_syslog_nat64_sadd (bibe->fib_index, &bibe->in_addr, bibe->in_port,
+ &bibe->out_addr, bibe->out_port, &ste->out_r_addr,
+ ste->r_port, bibe->proto);
+ return ste;
+}
+
+void
+nat64_db_st_entry_free (u32 thread_index,
+ nat64_db_t * db, nat64_db_st_entry_t * ste)
+{
+ nat64_db_st_entry_t *st;
+ nat64_db_bib_entry_t *bib, *bibe;
+ nat64_db_st_entry_key_t ste_key;
+ clib_bihash_kv_48_8_t kv;
+
+ switch (ip_proto_to_nat_proto (ste->proto))
+ {
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ case NAT_PROTOCOL_##N: \
+ st = db->st._##n##_st; \
+ bib = db->bib._##n##_bib; \
+ break;
+ foreach_nat_protocol
+#undef _
+/* *INDENT-ON* */
+ default:
+ st = db->st._unk_proto_st;
+ bib = db->bib._unk_proto_bib;
+ break;
+ }
+
+ bibe = pool_elt_at_index (bib, ste->bibe_index);
+
+ db->st.st_entries_num--;
+
+ /* delete hash lookup */
+ clib_memset (&ste_key, 0, sizeof (ste_key));
+ ste_key.l_addr.as_u64[0] = bibe->in_addr.as_u64[0];
+ ste_key.l_addr.as_u64[1] = bibe->in_addr.as_u64[1];
+ ste_key.r_addr.as_u64[0] = ste->in_r_addr.as_u64[0];
+ ste_key.r_addr.as_u64[1] = ste->in_r_addr.as_u64[1];
+ ste_key.fib_index = bibe->fib_index;
+ ste_key.l_port = bibe->in_port;
+ ste_key.r_port = ste->r_port;
+ ste_key.proto = ste->proto;
+ kv.key[0] = ste_key.as_u64[0];
+ kv.key[1] = ste_key.as_u64[1];
+ kv.key[2] = ste_key.as_u64[2];
+ kv.key[3] = ste_key.as_u64[3];
+ kv.key[4] = ste_key.as_u64[4];
+ kv.key[5] = ste_key.as_u64[5];
+ clib_bihash_add_del_48_8 (&db->st.in2out, &kv, 0);
+
+ clib_memset (&ste_key, 0, sizeof (ste_key));
+ ste_key.l_addr.ip4.as_u32 = bibe->out_addr.as_u32;
+ ste_key.r_addr.ip4.as_u32 = ste->out_r_addr.as_u32;
+ ste_key.l_port = bibe->out_port;
+ ste_key.r_port = ste->r_port;
+ ste_key.proto = ste->proto;
+ kv.key[0] = ste_key.as_u64[0];
+ kv.key[1] = ste_key.as_u64[1];
+ kv.key[2] = ste_key.as_u64[2];
+ kv.key[3] = ste_key.as_u64[3];
+ kv.key[4] = ste_key.as_u64[4];
+ kv.key[5] = ste_key.as_u64[5];
+ clib_bihash_add_del_48_8 (&db->st.out2in, &kv, 0);
+
+ /*fib_table_t *fib = fib_table_get (bibe->fib_index, FIB_PROTOCOL_IP6);
+ nat_ipfix_logging_nat64_session (thread_index, &bibe->in_addr,
+ &bibe->out_addr, bibe->proto,
+ bibe->in_port, bibe->out_port,
+ &ste->in_r_addr, &ste->out_r_addr,
+ ste->r_port, ste->r_port, fib->ft_table_id,
+ 0); */
+ nat_syslog_nat64_sdel (bibe->fib_index, &bibe->in_addr, bibe->in_port,
+ &bibe->out_addr, bibe->out_port, &ste->out_r_addr,
+ ste->r_port, bibe->proto);
+
+ /* delete from pool */
+ pool_put (st, ste);
+
+ /* decrement session number for BIB entry */
+ bibe->ses_num--;
+
+ /* delete BIB entry if last session and dynamic */
+ if (!bibe->is_static && !bibe->ses_num)
+ nat64_db_bib_entry_free (thread_index, db, bibe);
+}
+
+nat64_db_st_entry_t *
+nat64_db_st_entry_find (nat64_db_t * db, ip46_address_t * l_addr,
+ ip46_address_t * r_addr, u16 l_port, u16 r_port,
+ u8 proto, u32 fib_index, u8 is_ip6)
+{
+ nat64_db_st_entry_t *ste = 0;
+ nat64_db_st_entry_t *st;
+ nat64_db_st_entry_key_t ste_key;
+ clib_bihash_kv_48_8_t kv, value;
+
+ switch (ip_proto_to_nat_proto (proto))
+ {
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ case NAT_PROTOCOL_##N: \
+ st = db->st._##n##_st; \
+ break;
+ foreach_nat_protocol
+#undef _
+/* *INDENT-ON* */
+ default:
+ st = db->st._unk_proto_st;
+ break;
+ }
+
+ clib_memset (&ste_key, 0, sizeof (ste_key));
+ ste_key.l_addr.as_u64[0] = l_addr->as_u64[0];
+ ste_key.l_addr.as_u64[1] = l_addr->as_u64[1];
+ ste_key.r_addr.as_u64[0] = r_addr->as_u64[0];
+ ste_key.r_addr.as_u64[1] = r_addr->as_u64[1];
+ ste_key.fib_index = fib_index;
+ ste_key.l_port = l_port;
+ ste_key.r_port = r_port;
+ ste_key.proto = proto;
+ kv.key[0] = ste_key.as_u64[0];
+ kv.key[1] = ste_key.as_u64[1];
+ kv.key[2] = ste_key.as_u64[2];
+ kv.key[3] = ste_key.as_u64[3];
+ kv.key[4] = ste_key.as_u64[4];
+ kv.key[5] = ste_key.as_u64[5];
+
+ if (!clib_bihash_search_48_8
+ (is_ip6 ? &db->st.in2out : &db->st.out2in, &kv, &value))
+ ste = pool_elt_at_index (st, value.value);
+
+ return ste;
+}
+
+u32
+nat64_db_st_entry_get_index (nat64_db_t * db, nat64_db_st_entry_t * ste)
+{
+ nat64_db_st_entry_t *st;
+
+ switch (ip_proto_to_nat_proto (ste->proto))
+ {
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ case NAT_PROTOCOL_##N: \
+ st = db->st._##n##_st; \
+ break;
+ foreach_nat_protocol
+#undef _
+/* *INDENT-ON* */
+ default:
+ st = db->st._unk_proto_st;
+ return (u32) ~ 0;
+ }
+
+ return ste - st;
+}
+
+nat64_db_st_entry_t *
+nat64_db_st_entry_by_index (nat64_db_t * db, u8 proto, u32 ste_index)
+{
+ nat64_db_st_entry_t *st;
+
+ switch (ip_proto_to_nat_proto (proto))
+ {
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ case NAT_PROTOCOL_##N: \
+ st = db->st._##n##_st; \
+ break;
+ foreach_nat_protocol
+#undef _
+/* *INDENT-ON* */
+ default:
+ st = db->st._unk_proto_st;
+ break;
+ }
+
+ return pool_elt_at_index (st, ste_index);
+}
+
+void
+nad64_db_st_free_expired (u32 thread_index, nat64_db_t * db, u32 now)
+{
+ u32 *ste_to_be_free = 0, *ste_index;
+ nat64_db_st_entry_t *st, *ste;
+
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ st = db->st._##n##_st; \
+ pool_foreach (ste, st, ({\
+ if (i == NAT_PROTOCOL_TCP && !ste->tcp_state) \
+ continue; \
+ if (ste->expire < now) \
+ vec_add1 (ste_to_be_free, ste - st); \
+ })); \
+ vec_foreach (ste_index, ste_to_be_free) \
+ nat64_db_st_entry_free (thread_index, db, \
+ pool_elt_at_index(st, ste_index[0])); \
+ vec_free (ste_to_be_free); \
+ ste_to_be_free = 0;
+ foreach_nat_protocol
+#undef _
+ st = db->st._unk_proto_st;
+ pool_foreach (ste, st, ({
+ if (ste->expire < now)
+ vec_add1 (ste_to_be_free, ste - st);
+ }));
+ vec_foreach (ste_index, ste_to_be_free)
+ nat64_db_st_entry_free (thread_index, db,
+ pool_elt_at_index(st, ste_index[0]));
+ vec_free (ste_to_be_free);
+/* *INDENT-ON* */
+}
+
+void
+nat64_db_free_out_addr (u32 thread_index,
+ nat64_db_t * db, ip4_address_t * out_addr)
+{
+ u32 *ste_to_be_free = 0, *ste_index;
+ nat64_db_st_entry_t *st, *ste;
+ nat64_db_bib_entry_t *bibe;
+
+ db->addr_free = 1;
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ st = db->st._##n##_st; \
+ pool_foreach (ste, st, ({ \
+ bibe = pool_elt_at_index (db->bib._##n##_bib, ste->bibe_index); \
+ if (bibe->out_addr.as_u32 == out_addr->as_u32) \
+ vec_add1 (ste_to_be_free, ste - st); \
+ })); \
+ vec_foreach (ste_index, ste_to_be_free) \
+ nat64_db_st_entry_free (thread_index, db, \
+ pool_elt_at_index(st, ste_index[0])); \
+ vec_free (ste_to_be_free); \
+ ste_to_be_free = 0;
+ foreach_nat_protocol
+#undef _
+ st = db->st._unk_proto_st;
+ pool_foreach (ste, st, ({
+ bibe = pool_elt_at_index (db->bib._unk_proto_bib, ste->bibe_index);
+ if (bibe->out_addr.as_u32 == out_addr->as_u32)
+ vec_add1 (ste_to_be_free, ste - st);
+ }));
+ vec_foreach (ste_index, ste_to_be_free)
+ nat64_db_st_entry_free (thread_index, db,
+ pool_elt_at_index(st, ste_index[0]));
+ vec_free (ste_to_be_free);
+ db->addr_free = 0;
+/* *INDENT-ON* */
+}
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */
diff --git a/src/plugins/nat/nat64/nat64_db.h b/src/plugins/nat/nat64/nat64_db.h
new file mode 100644
index 00000000000..711b6bf6b03
--- /dev/null
+++ b/src/plugins/nat/nat64/nat64_db.h
@@ -0,0 +1,380 @@
+/*
+ * Copyright (c) 2020 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+#ifndef __included_nat64_db_h__
+#define __included_nat64_db_h__
+
+#include <vnet/vnet.h>
+#include <vnet/ip/ip.h>
+#include <vnet/fib/fib_source.h>
+
+#include <vppinfra/bihash_24_8.h>
+#include <vppinfra/bihash_48_8.h>
+
+typedef struct
+{
+ u32 bib_buckets;
+ u32 bib_memory_size;
+ u32 st_buckets;
+ u32 st_memory_size;
+} nat64_config_t;
+
+typedef struct
+{
+ union
+ {
+ struct
+ {
+ ip46_address_t addr;
+ u32 fib_index;
+ u16 port;
+ u8 proto;
+ u8 rsvd;
+ };
+ u64 as_u64[3];
+ };
+} nat64_db_bib_entry_key_t;
+
+/* *INDENT-OFF* */
+typedef CLIB_PACKED(struct
+{
+ ip6_address_t in_addr;
+ u16 in_port;
+ ip4_address_t out_addr;
+ u16 out_port;
+ u32 fib_index;
+ u32 ses_num;
+ u8 proto;
+ u8 is_static;
+}) nat64_db_bib_entry_t;
+/* *INDENT-ON* */
+
+typedef struct
+{
+ /* BIBs */
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ nat64_db_bib_entry_t *_##n##_bib;
+ foreach_nat_protocol
+#undef _
+/* *INDENT-ON* */
+ nat64_db_bib_entry_t *_unk_proto_bib;
+
+ /* BIB lookup */
+ clib_bihash_24_8_t in2out;
+ clib_bihash_24_8_t out2in;
+
+ u32 limit;
+ u32 bib_entries_num;
+} nat64_db_bib_t;
+
+typedef struct
+{
+ union
+ {
+ struct
+ {
+ ip46_address_t l_addr;
+ ip46_address_t r_addr;
+ u32 fib_index;
+ u16 l_port;
+ u16 r_port;
+ u8 proto;
+ u8 rsvd[7];
+ };
+ u64 as_u64[6];
+ };
+} nat64_db_st_entry_key_t;
+
+/* *INDENT-OFF* */
+typedef CLIB_PACKED(struct
+{
+ ip6_address_t in_r_addr;
+ ip4_address_t out_r_addr;
+ u16 r_port;
+ u32 bibe_index;
+ u32 expire;
+ u8 proto;
+ u8 tcp_state;
+}) nat64_db_st_entry_t;
+/* *INDENT-ON* */
+
+typedef struct
+{
+ /* session tables */
+/* *INDENT-OFF* */
+#define _(N, i, n, s) \
+ nat64_db_st_entry_t *_##n##_st;
+ foreach_nat_protocol
+#undef _
+/* *INDENT-ON* */
+ nat64_db_st_entry_t *_unk_proto_st;
+
+ /* session lookup */
+ clib_bihash_48_8_t in2out;
+ clib_bihash_48_8_t out2in;
+
+ u32 limit;
+ u32 st_entries_num;
+} nat64_db_st_t;
+
+struct nat64_db_s;
+
+/**
+ * @brief Call back function to free NAT64 pool address and port when BIB
+ * entry is deleted.
+ */
+typedef void (*nat64_db_free_addr_port_function_t) (struct nat64_db_s * db,
+ ip4_address_t * addr,
+ u16 port, u8 proto);
+
+typedef struct nat64_db_s
+{
+ nat64_db_bib_t bib;
+ nat64_db_st_t st;
+ nat64_db_free_addr_port_function_t free_addr_port_cb;
+ u8 addr_free;
+} nat64_db_t;
+
+/**
+ * @brief Initialize NAT64 DB.
+ *
+ * @param db NAT64 DB.
+ * @param c.bib_buckets Number of BIB hash buckets.
+ * @param c.bib_memory_size Memory size of BIB hash.
+ * @param c.st_buckets Number of session table hash buckets.
+ * @param c.st_memory_size Memory size of session table hash.
+ * @param free_addr_port_cb Call back function to free address and port.
+ *
+ * @returns 0 on success, non-zero value otherwise.
+ */
+int nat64_db_init (nat64_db_t * db, nat64_config_t c,
+ nat64_db_free_addr_port_function_t free_addr_port_cb);
+
+/**
+ * @brief Free NAT64 DB.
+ *
+ * @param db NAT64 DB.
+ *
+ * @returns 0 on success, non-zero value otherwise.
+ */
+int nat64_db_free (nat64_db_t * db);
+
+
+/**
+ * @brief Create new NAT64 BIB entry.
+ *
+ * @param thread_index thread index.
+ * @param db NAT64 DB.
+ * @param in_addr Inside IPv6 address.
+ * @param out_addr Outside IPv4 address.
+ * @param in_port Inside port number.
+ * @param out_port Outside port number.
+ * @param fib_index FIB index.
+ * @param proto L4 protocol.
+ * @param is_static 1 if static, 0 if dynamic.
+ *
+ * @returns BIB entry on success, 0 otherwise.
+ */
+nat64_db_bib_entry_t *nat64_db_bib_entry_create (u32 thread_index,
+ nat64_db_t * db,
+ ip6_address_t * in_addr,
+ ip4_address_t * out_addr,
+ u16 in_port, u16 out_port,
+ u32 fib_index, u8 proto,
+ u8 is_static);
+
+/**
+ * @brief Free NAT64 BIB entry.
+ *
+ * @param thread_index thread index.
+ * @param db NAT64 DB.
+ * @param bibe BIB entry.
+ */
+void nat64_db_bib_entry_free (u32 thread_index, nat64_db_t * db,
+ nat64_db_bib_entry_t * bibe);
+
+/**
+ * @brief Call back function when walking NAT64 BIB, non-zero
+ * return value stop walk.
+ */
+typedef int (*nat64_db_bib_walk_fn_t) (nat64_db_bib_entry_t * bibe,
+ void *ctx);
+/**
+ * @brief Walk NAT64 BIB.
+ *
+ * @param db NAT64 DB.
+ * @param proto BIB L4 protocol:
+ * - 255 all BIBs
+ * - 6 TCP BIB
+ * - 17 UDP BIB
+ * - 1/58 ICMP BIB
+ *
+ * u - otherwise "unknown" protocol BIB
+ * @param fn The function to invoke on each entry visited.
+ * @param ctx A context passed in the visit function.
+ */
+void nat64_db_bib_walk (nat64_db_t * db, u8 proto,
+ nat64_db_bib_walk_fn_t fn, void *ctx);
+
+/**
+ * @brief Find NAT64 BIB entry.
+ *
+ * @param db NAT64 DB.
+ * @param addr IP address.
+ * @param port Port number.
+ * @param proto L4 protocol.
+ * @param fib_index FIB index.
+ * @param is_ip6 1 if find by IPv6 (inside) address, 0 by IPv4 (outside).
+ *
+ * @return BIB entry if found.
+ */
+nat64_db_bib_entry_t *nat64_db_bib_entry_find (nat64_db_t * db,
+ ip46_address_t * addr,
+ u16 port,
+ u8 proto,
+ u32 fib_index, u8 is_ip6);
+
+/**
+ * @brief Get BIB entry by index and protocol.
+ *
+ * @param db NAT64 DB.
+ * @param proto L4 protocol.
+ * @param bibe_index BIB entry index.
+ *
+ * @return BIB entry if found.
+ */
+nat64_db_bib_entry_t *nat64_db_bib_entry_by_index (nat64_db_t * db,
+ u8 proto, u32 bibe_index);
+/**
+ * @brief Create new NAT64 session table entry.
+ *
+ * @param thread_index thread index.
+ * @param db NAT64 DB.
+ * @param bibe Corresponding BIB entry.
+ * @param in_r_addr Inside IPv6 address of the remote host.
+ * @param out_r_addr Outside IPv4 address of the remote host.
+ * @param r_port Remote host port number.
+ *
+ * @returns BIB entry on success, 0 otherwise.
+ */
+nat64_db_st_entry_t *nat64_db_st_entry_create (u32 thread_index,
+ nat64_db_t * db,
+ nat64_db_bib_entry_t * bibe,
+ ip6_address_t * in_r_addr,
+ ip4_address_t * out_r_addr,
+ u16 r_port);
+
+/**
+ * @brief Free NAT64 session table entry.
+ *
+ * @param thread_index thread index.
+ * @param db NAT64 DB.
+ * @param ste Session table entry.
+ */
+void nat64_db_st_entry_free (u32 thread_index, nat64_db_t * db,
+ nat64_db_st_entry_t * ste);
+
+/**
+ * @brief Find NAT64 session table entry.
+ *
+ * @param db NAT64 DB.
+ * @param l_addr Local host address.
+ * @param r_addr Remote host address.
+ * @param l_port Local host port number.
+ * @param r_port Remote host port number.
+ * @param proto L4 protocol.
+ * @param fib_index FIB index.
+ * @param is_ip6 1 if find by IPv6 (inside) address, 0 by IPv4 (outside).
+ *
+ * @return BIB entry if found.
+ */
+nat64_db_st_entry_t *nat64_db_st_entry_find (nat64_db_t * db,
+ ip46_address_t * l_addr,
+ ip46_address_t * r_addr,
+ u16 l_port, u16 r_port,
+ u8 proto,
+ u32 fib_index, u8 is_ip6);
+
+/**
+ * @brief Call back function when walking NAT64 session table, non-zero
+ * return value stop walk.
+ */
+typedef int (*nat64_db_st_walk_fn_t) (nat64_db_st_entry_t * ste, void *ctx);
+
+/**
+ * @brief Walk NAT64 session table.
+ *
+ * @param db NAT64 DB.
+ * @param proto L4 protocol:
+ * - 255 all session tables
+ * - 6 TCP session table
+ * - 17 UDP session table
+ * - 1/58 ICMP session table
+ * - otherwise "unknown" protocol session table
+ * @param fn The function to invoke on each entry visited.
+ * @param ctx A context passed in the visit function.
+ */
+void nat64_db_st_walk (nat64_db_t * db, u8 proto,
+ nat64_db_st_walk_fn_t fn, void *ctx);
+
+/**
+ * @brief Free expired session entries in session tables.
+ *
+ * @param thread_index thread index.
+ * @param db NAT64 DB.
+ * @param now Current time.
+ */
+void nad64_db_st_free_expired (u32 thread_index, nat64_db_t * db, u32 now);
+
+/**
+ * @brief Free sessions using specific outside address.
+ *
+ * @param thread_index thread index.
+ * @param db NAT64 DB.
+ * @param out_addr Outside address to match.
+ */
+void nat64_db_free_out_addr (u32 thread_index, nat64_db_t * db,
+ ip4_address_t * out_addr);
+
+/*
+ * @brief Get ST entry index.
+ *
+ * @param db NAT64 DB.
+ * @param ste ST entry.
+ *
+ * @return ST entry index on success, ~0 otherwise.
+ */
+u32 nat64_db_st_entry_get_index (nat64_db_t * db, nat64_db_st_entry_t * ste);
+
+/**
+ * @brief Get ST entry by index and protocol.
+ *
+ * @param db NAT64 DB.
+ * @param proto L4 protocol.
+ * @param bibe_index ST entry index.
+ *
+ * @return BIB entry if found.
+ */
+nat64_db_st_entry_t *nat64_db_st_entry_by_index (nat64_db_t * db,
+ u8 proto, u32 ste_index);
+#endif /* __included_nat64_db_h__ */
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */
diff --git a/src/plugins/nat/nat64/nat64_doc.md b/src/plugins/nat/nat64/nat64_doc.md
new file mode 100644
index 00000000000..f65b46338b0
--- /dev/null
+++ b/src/plugins/nat/nat64/nat64_doc.md
@@ -0,0 +1,73 @@
+# Stateful NAT64: Network Address and Protocol Translation from IPv6 Clients to IPv4 Servers {#nat64_doc}
+
+## Introduction
+
+Stateful NAT64 in VPP allows IPv6-only clients to contact IPv4 servers using unicast UDP, TCP, or ICMP based on RFC 6146.
+
+## Configuration
+
+### Enable/disable NAT64 feature on the interface
+
+> set interface nat64 in|out <intfc> [del]
+
+in: inside/local/IPv6 network
+out: outside/external/IPv4 network
+intfc: interface name
+
+### Add/delete NAT64 pool address
+
+One or more public IPv4 addresses assigned to a NAT64 are shared among several IPv6-only clients.
+
+> nat64 add pool address <ip4-range-start> [- <ip4-range-end>] [tenant-vrf <tenant-vrf-id>] [del]
+
+ip4-range-start: First IPv4 address of the range
+ip4-range-end: Last IPv4 address of the range (optional, not used for single address)
+tenant-vrf-id: VRF id of the tenant associated with the pool address (optional, if not set pool address is global)
+
+### Add/delete static BIB entry
+
+Stateful NAT64 also supports IPv4-initiated communications to a subset of the IPv6 hosts through staticaly configured bindings.
+
+> nat64 add static bib <ip6-addr> <in-port> <ip4-addr> <out-port> tcp|udp|icmp [vfr <table-id>] [del]
+
+ip6-addr: inside IPv6 address of the host
+in-port: inside port or ICMPv6 identifier
+ip4-addr: outside IPv4 address of the host
+out-port: outside port or ICMPv4 identifier
+table-id: VRF id of the tenant associated with the BIB entry (optional, default use global VRF)
+
+### Set NAT64 session timeouts
+
+Session is deleted when timer expires. If all sessions corresponding to a dynamically create BIB entry are deleted, then the BIB entry is also deleted. When packets are flowing sessiom timer is refreshed to keep the session alive.
+
+> set nat64 timeouts udp <sec> icmp <sec> tcp-trans <sec> tcp-est <sec> tcp-incoming-syn <sec> | reset
+
+udp: UDP session timeout value (default 300sec)
+icmp: ICMP session timeout value (default 60sec)
+tcp-trans: transitory TCP session timeout value (default 240sec)
+tcp-est: established TCP session timeout value (default 7440sec)
+tcp-incoming-syn: incoming SYN TCP session timeout value (default 6sec)
+reset: reset timers to default values
+
+### Set NAT64 prefix
+
+Stateful NAT64 support the algorithm for generating IPv6 representations of IPv4 addresses defined in RFC 6052. If no prefix is configured, Well-Known Prefix (64:ff9b::/96) is used.
+
+> nat64 add prefix <ip6-prefix>/<plen> [tenant-vrf <vrf-id>] [del]
+
+ip6-prefix: IPv6 prefix
+plen: prefix length (valid values: 32, 40, 48, 56, 64, or 96)
+tenant-vrf: VRF id of the tenant associated with the prefix
+
+### Show commands
+
+> show nat64 pool
+> show nat64 interfaces
+> show nat64 bib tcp|udp|icmp
+> show nat64 session table tcp|udp|icmp
+> show nat64 timeouts
+> show nat64 prefix
+
+## Notes
+
+Multi thread is not supported yet (CLI/API commands are disabled when VPP runs with multiple threads).
diff --git a/src/plugins/nat/nat64/nat64_in2out.c b/src/plugins/nat/nat64/nat64_in2out.c
new file mode 100644
index 00000000000..5d98277d443
--- /dev/null
+++ b/src/plugins/nat/nat64/nat64_in2out.c
@@ -0,0 +1,1411 @@
+/*
+ * Copyright (c) 2020 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <nat/nat64/nat64.h>
+#include <vnet/ip/ip6_to_ip4.h>
+#include <vnet/fib/fib_table.h>
+#include <nat/lib/nat_inlines.h>
+
+typedef struct
+{
+ u32 sw_if_index;
+ u32 next_index;
+ u8 is_slow_path;
+} nat64_in2out_trace_t;
+
+static u8 *
+format_nat64_in2out_trace (u8 * s, va_list * args)
+{
+ CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
+ CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
+ nat64_in2out_trace_t *t = va_arg (*args, nat64_in2out_trace_t *);
+ char *tag;
+
+ tag = t->is_slow_path ? "NAT64-in2out-slowpath" : "NAT64-in2out";
+
+ s =
+ format (s, "%s: sw_if_index %d, next index %d", tag, t->sw_if_index,
+ t->next_index);
+
+ return s;
+}
+
+#define foreach_nat64_in2out_error \
+_(UNSUPPORTED_PROTOCOL, "unsupported protocol") \
+_(NO_TRANSLATION, "no translation") \
+_(UNKNOWN, "unknown")
+
+
+typedef enum
+{
+#define _(sym,str) NAT64_IN2OUT_ERROR_##sym,
+ foreach_nat64_in2out_error
+#undef _
+ NAT64_IN2OUT_N_ERROR,
+} nat64_in2out_error_t;
+
+static char *nat64_in2out_error_strings[] = {
+#define _(sym,string) string,
+ foreach_nat64_in2out_error
+#undef _
+};
+
+typedef enum
+{
+ NAT64_IN2OUT_NEXT_IP4_LOOKUP,
+ NAT64_IN2OUT_NEXT_IP6_LOOKUP,
+ NAT64_IN2OUT_NEXT_DROP,
+ NAT64_IN2OUT_NEXT_SLOWPATH,
+ NAT64_IN2OUT_N_NEXT,
+} nat64_in2out_next_t;
+
+typedef struct nat64_in2out_set_ctx_t_
+{
+ vlib_buffer_t *b;
+ vlib_main_t *vm;
+ u32 thread_index;
+} nat64_in2out_set_ctx_t;
+
+static inline u8
+nat64_not_translate (u32 sw_if_index, ip6_address_t ip6_addr)
+{
+ ip6_address_t *addr;
+ ip6_main_t *im6 = &ip6_main;
+ ip_lookup_main_t *lm6 = &im6->lookup_main;
+ ip_interface_address_t *ia = 0;
+
+ /* *INDENT-OFF* */
+ foreach_ip_interface_address (lm6, ia, sw_if_index, 0,
+ ({
+ addr = ip_interface_address_get_address (lm6, ia);
+ if (0 == ip6_address_compare (addr, &ip6_addr))
+ return 1;
+ }));
+ /* *INDENT-ON* */
+
+ return 0;
+}
+
+/**
+ * @brief Check whether is a hairpinning.
+ *
+ * If the destination IP address of the packet is an IPv4 address assigned to
+ * the NAT64 itself, then the packet is a hairpin packet.
+ *
+ * param dst_addr Destination address of the packet.
+ *
+ * @returns 1 if hairpinning, otherwise 0.
+ */
+static_always_inline int
+is_hairpinning (ip6_address_t * dst_addr)
+{
+ nat64_main_t *nm = &nat64_main;
+ int i;
+
+ for (i = 0; i < vec_len (nm->addr_pool); i++)
+ {
+ if (nm->addr_pool[i].addr.as_u32 == dst_addr->as_u32[3])
+ return 1;
+ }
+
+ return 0;
+}
+
+static int
+nat64_in2out_tcp_udp (vlib_main_t * vm, vlib_buffer_t * p, u16 l4_offset,
+ u16 frag_hdr_offset, nat64_in2out_set_ctx_t * ctx)
+{
+ ip6_header_t *ip6;
+ ip_csum_t csum = 0;
+ ip4_header_t *ip4;
+ u16 fragment_id;
+ u8 frag_more;
+ u16 frag_offset;
+ nat64_main_t *nm = &nat64_main;
+ nat64_db_bib_entry_t *bibe;
+ nat64_db_st_entry_t *ste;
+ ip46_address_t old_saddr, old_daddr;
+ ip4_address_t new_daddr;
+ u32 sw_if_index, fib_index;
+ u8 proto = vnet_buffer (p)->ip.reass.ip_proto;
+ u16 sport = vnet_buffer (p)->ip.reass.l4_src_port;
+ u16 dport = vnet_buffer (p)->ip.reass.l4_dst_port;
+ nat64_db_t *db = &nm->db[ctx->thread_index];
+
+ ip6 = vlib_buffer_get_current (p);
+
+ vlib_buffer_advance (p, l4_offset - sizeof (*ip4));
+ ip4 = vlib_buffer_get_current (p);
+
+ u32 ip_version_traffic_class_and_flow_label =
+ ip6->ip_version_traffic_class_and_flow_label;
+ u16 payload_length = ip6->payload_length;
+ u8 hop_limit = ip6->hop_limit;
+
+ old_saddr.as_u64[0] = ip6->src_address.as_u64[0];
+ old_saddr.as_u64[1] = ip6->src_address.as_u64[1];
+ old_daddr.as_u64[0] = ip6->dst_address.as_u64[0];
+ old_daddr.as_u64[1] = ip6->dst_address.as_u64[1];
+
+ if (PREDICT_FALSE (frag_hdr_offset))
+ {
+ //Only the first fragment
+ ip6_frag_hdr_t *hdr =
+ (ip6_frag_hdr_t *) u8_ptr_add (ip6, frag_hdr_offset);
+ fragment_id = frag_id_6to4 (hdr->identification);
+ frag_more = ip6_frag_hdr_more (hdr);
+ frag_offset = ip6_frag_hdr_offset (hdr);
+ }
+ else
+ {
+ fragment_id = 0;
+ frag_offset = 0;
+ frag_more = 0;
+ }
+
+ ip4->ip_version_and_header_length =
+ IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS;
+ ip4->tos = ip6_translate_tos (ip_version_traffic_class_and_flow_label);
+ ip4->length =
+ u16_net_add (payload_length, sizeof (*ip4) + sizeof (*ip6) - l4_offset);
+ ip4->fragment_id = fragment_id;
+ ip4->flags_and_fragment_offset =
+ clib_host_to_net_u16 (frag_offset |
+ (frag_more ? IP4_HEADER_FLAG_MORE_FRAGMENTS : 0));
+ ip4->ttl = hop_limit;
+ ip4->protocol = (proto == IP_PROTOCOL_ICMP6) ? IP_PROTOCOL_ICMP : proto;
+
+ sw_if_index = vnet_buffer (ctx->b)->sw_if_index[VLIB_RX];
+ fib_index =
+ fib_table_get_index_for_sw_if_index (FIB_PROTOCOL_IP6, sw_if_index);
+
+ ste =
+ nat64_db_st_entry_find (db, &old_saddr, &old_daddr, sport, dport, proto,
+ fib_index, 1);
+
+ if (ste)
+ {
+ bibe = nat64_db_bib_entry_by_index (db, proto, ste->bibe_index);
+ if (!bibe)
+ return -1;
+ }
+ else
+ {
+ bibe =
+ nat64_db_bib_entry_find (db, &old_saddr, sport, proto, fib_index, 1);
+
+ if (!bibe)
+ {
+ u16 out_port;
+ ip4_address_t out_addr;
+ if (nat64_alloc_out_addr_and_port
+ (fib_index, ip_proto_to_nat_proto (proto), &out_addr,
+ &out_port, ctx->thread_index))
+ return -1;
+
+ bibe =
+ nat64_db_bib_entry_create (ctx->thread_index, db,
+ &old_saddr.ip6, &out_addr, sport,
+ out_port, fib_index, proto, 0);
+ if (!bibe)
+ return -1;
+
+ vlib_set_simple_counter (&nm->total_bibs, ctx->thread_index, 0,
+ db->bib.bib_entries_num);
+ }
+
+ nat64_extract_ip4 (&old_daddr.ip6, &new_daddr, fib_index);
+ ste =
+ nat64_db_st_entry_create (ctx->thread_index, db, bibe,
+ &old_daddr.ip6, &new_daddr, dport);
+ if (!ste)
+ return -1;
+
+ vlib_set_simple_counter (&nm->total_sessions, ctx->thread_index, 0,
+ db->st.st_entries_num);
+ }
+
+ ip4->src_address.as_u32 = bibe->out_addr.as_u32;
+ ip4->dst_address.as_u32 = ste->out_r_addr.as_u32;
+
+ ip4->checksum = ip4_header_checksum (ip4);
+
+ if (!vnet_buffer (p)->ip.reass.is_non_first_fragment)
+ {
+ udp_header_t *udp = (udp_header_t *) (ip4 + 1);
+ udp->src_port = bibe->out_port;
+
+ //UDP checksum is optional over IPv4
+ if (proto == IP_PROTOCOL_UDP)
+ {
+ udp->checksum = 0;
+ }
+ else
+ {
+ tcp_header_t *tcp = (tcp_header_t *) (ip4 + 1);
+ csum = ip_csum_sub_even (tcp->checksum, old_saddr.as_u64[0]);
+ csum = ip_csum_sub_even (csum, old_saddr.as_u64[1]);
+ csum = ip_csum_sub_even (csum, old_daddr.as_u64[0]);
+ csum = ip_csum_sub_even (csum, old_daddr.as_u64[1]);
+ csum = ip_csum_add_even (csum, ip4->dst_address.as_u32);
+ csum = ip_csum_add_even (csum, ip4->src_address.as_u32);
+ csum = ip_csum_sub_even (csum, sport);
+ csum = ip_csum_add_even (csum, udp->src_port);
+ mss_clamping (nm->mss_clamping, tcp, &csum);
+ tcp->checksum = ip_csum_fold (csum);
+
+ nat64_tcp_session_set_state (ste, tcp, 1);
+ }
+ }
+
+ nat64_session_reset_timeout (ste, ctx->vm);
+
+ return 0;
+}
+
+static int
+nat64_in2out_icmp_set_cb (ip6_header_t * ip6, ip4_header_t * ip4, void *arg)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_in2out_set_ctx_t *ctx = arg;
+ nat64_db_bib_entry_t *bibe;
+ nat64_db_st_entry_t *ste;
+ ip46_address_t saddr, daddr;
+ u32 sw_if_index, fib_index;
+ icmp46_header_t *icmp = ip6_next_header (ip6);
+ nat64_db_t *db = &nm->db[ctx->thread_index];
+
+ sw_if_index = vnet_buffer (ctx->b)->sw_if_index[VLIB_RX];
+ fib_index =
+ fib_table_get_index_for_sw_if_index (FIB_PROTOCOL_IP6, sw_if_index);
+
+ saddr.as_u64[0] = ip6->src_address.as_u64[0];
+ saddr.as_u64[1] = ip6->src_address.as_u64[1];
+ daddr.as_u64[0] = ip6->dst_address.as_u64[0];
+ daddr.as_u64[1] = ip6->dst_address.as_u64[1];
+
+ if (icmp->type == ICMP4_echo_request || icmp->type == ICMP4_echo_reply)
+ {
+ u16 in_id = ((u16 *) (icmp))[2];
+ ste =
+ nat64_db_st_entry_find (db, &saddr, &daddr, in_id, 0,
+ IP_PROTOCOL_ICMP, fib_index, 1);
+
+ if (ste)
+ {
+ bibe =
+ nat64_db_bib_entry_by_index (db, IP_PROTOCOL_ICMP,
+ ste->bibe_index);
+ if (!bibe)
+ return -1;
+ }
+ else
+ {
+ bibe =
+ nat64_db_bib_entry_find (db, &saddr, in_id,
+ IP_PROTOCOL_ICMP, fib_index, 1);
+
+ if (!bibe)
+ {
+ u16 out_id;
+ ip4_address_t out_addr;
+ if (nat64_alloc_out_addr_and_port
+ (fib_index, NAT_PROTOCOL_ICMP, &out_addr, &out_id,
+ ctx->thread_index))
+ return -1;
+
+ bibe =
+ nat64_db_bib_entry_create (ctx->thread_index, db,
+ &ip6->src_address, &out_addr,
+ in_id, out_id, fib_index,
+ IP_PROTOCOL_ICMP, 0);
+ if (!bibe)
+ return -1;
+
+ vlib_set_simple_counter (&nm->total_bibs, ctx->thread_index, 0,
+ db->bib.bib_entries_num);
+ }
+
+ nat64_extract_ip4 (&ip6->dst_address, &daddr.ip4, fib_index);
+ ste =
+ nat64_db_st_entry_create (ctx->thread_index, db, bibe,
+ &ip6->dst_address, &daddr.ip4, 0);
+ if (!ste)
+ return -1;
+
+ vlib_set_simple_counter (&nm->total_sessions, ctx->thread_index, 0,
+ db->st.st_entries_num);
+ }
+
+ nat64_session_reset_timeout (ste, ctx->vm);
+
+ ip4->src_address.as_u32 = bibe->out_addr.as_u32;
+ ((u16 *) (icmp))[2] = bibe->out_port;
+
+ ip4->dst_address.as_u32 = ste->out_r_addr.as_u32;
+ }
+ else
+ {
+ if (!vec_len (nm->addr_pool))
+ return -1;
+
+ ip4->src_address.as_u32 = nm->addr_pool[0].addr.as_u32;
+ nat64_extract_ip4 (&ip6->dst_address, &ip4->dst_address, fib_index);
+ }
+
+ return 0;
+}
+
+static int
+nat64_in2out_inner_icmp_set_cb (ip6_header_t * ip6, ip4_header_t * ip4,
+ void *arg)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_in2out_set_ctx_t *ctx = arg;
+ nat64_db_st_entry_t *ste;
+ nat64_db_bib_entry_t *bibe;
+ ip46_address_t saddr, daddr;
+ u32 sw_if_index, fib_index;
+ u8 proto = ip6->protocol;
+ nat64_db_t *db = &nm->db[ctx->thread_index];
+
+ sw_if_index = vnet_buffer (ctx->b)->sw_if_index[VLIB_RX];
+ fib_index =
+ fib_table_get_index_for_sw_if_index (FIB_PROTOCOL_IP6, sw_if_index);
+
+ saddr.as_u64[0] = ip6->src_address.as_u64[0];
+ saddr.as_u64[1] = ip6->src_address.as_u64[1];
+ daddr.as_u64[0] = ip6->dst_address.as_u64[0];
+ daddr.as_u64[1] = ip6->dst_address.as_u64[1];
+
+ if (proto == IP_PROTOCOL_ICMP6)
+ {
+ icmp46_header_t *icmp = ip6_next_header (ip6);
+ u16 in_id = ((u16 *) (icmp))[2];
+ proto = IP_PROTOCOL_ICMP;
+
+ if (!
+ (icmp->type == ICMP4_echo_request
+ || icmp->type == ICMP4_echo_reply))
+ return -1;
+
+ ste =
+ nat64_db_st_entry_find (db, &daddr, &saddr, in_id, 0, proto,
+ fib_index, 1);
+ if (!ste)
+ return -1;
+
+ bibe = nat64_db_bib_entry_by_index (db, proto, ste->bibe_index);
+ if (!bibe)
+ return -1;
+
+ ip4->dst_address.as_u32 = bibe->out_addr.as_u32;
+ ((u16 *) (icmp))[2] = bibe->out_port;
+ ip4->src_address.as_u32 = ste->out_r_addr.as_u32;
+ }
+ else
+ {
+ udp_header_t *udp = ip6_next_header (ip6);
+ tcp_header_t *tcp = ip6_next_header (ip6);
+ u16 *checksum;
+ ip_csum_t csum;
+
+ u16 sport = udp->src_port;
+ u16 dport = udp->dst_port;
+
+ ste =
+ nat64_db_st_entry_find (db, &daddr, &saddr, dport, sport, proto,
+ fib_index, 1);
+ if (!ste)
+ return -1;
+
+ bibe = nat64_db_bib_entry_by_index (db, proto, ste->bibe_index);
+ if (!bibe)
+ return -1;
+
+ ip4->dst_address.as_u32 = bibe->out_addr.as_u32;
+ udp->dst_port = bibe->out_port;
+ ip4->src_address.as_u32 = ste->out_r_addr.as_u32;
+
+ if (proto == IP_PROTOCOL_TCP)
+ checksum = &tcp->checksum;
+ else
+ checksum = &udp->checksum;
+ csum = ip_csum_sub_even (*checksum, dport);
+ csum = ip_csum_add_even (csum, udp->dst_port);
+ *checksum = ip_csum_fold (csum);
+ }
+
+ return 0;
+}
+
+typedef struct unk_proto_st_walk_ctx_t_
+{
+ ip6_address_t src_addr;
+ ip6_address_t dst_addr;
+ ip4_address_t out_addr;
+ u32 fib_index;
+ u32 thread_index;
+ u8 proto;
+} unk_proto_st_walk_ctx_t;
+
+static int
+unk_proto_st_walk (nat64_db_st_entry_t * ste, void *arg)
+{
+ nat64_main_t *nm = &nat64_main;
+ unk_proto_st_walk_ctx_t *ctx = arg;
+ nat64_db_bib_entry_t *bibe;
+ ip46_address_t saddr, daddr;
+ nat64_db_t *db = &nm->db[ctx->thread_index];
+
+ if (ip6_address_is_equal (&ste->in_r_addr, &ctx->dst_addr))
+ {
+ bibe = nat64_db_bib_entry_by_index (db, ste->proto, ste->bibe_index);
+ if (!bibe)
+ return -1;
+
+ if (ip6_address_is_equal (&bibe->in_addr, &ctx->src_addr)
+ && bibe->fib_index == ctx->fib_index)
+ {
+ clib_memset (&saddr, 0, sizeof (saddr));
+ saddr.ip4.as_u32 = bibe->out_addr.as_u32;
+ clib_memset (&daddr, 0, sizeof (daddr));
+ nat64_extract_ip4 (&ctx->dst_addr, &daddr.ip4, ctx->fib_index);
+
+ if (nat64_db_st_entry_find
+ (db, &daddr, &saddr, 0, 0, ctx->proto, ctx->fib_index, 0))
+ return -1;
+
+ ctx->out_addr.as_u32 = bibe->out_addr.as_u32;
+ return 1;
+ }
+ }
+
+ return 0;
+}
+
+static int
+nat64_in2out_unk_proto (vlib_main_t * vm, vlib_buffer_t * p, u8 l4_protocol,
+ u16 l4_offset, u16 frag_hdr_offset,
+ nat64_in2out_set_ctx_t * s_ctx)
+{
+ ip6_header_t *ip6;
+ ip4_header_t *ip4;
+ u16 fragment_id;
+ u16 frag_offset;
+ u8 frag_more;
+
+ ip6 = vlib_buffer_get_current (p);
+
+ ip4 = (ip4_header_t *) u8_ptr_add (ip6, l4_offset - sizeof (*ip4));
+
+ vlib_buffer_advance (p, l4_offset - sizeof (*ip4));
+
+ if (PREDICT_FALSE (frag_hdr_offset))
+ {
+ //Only the first fragment
+ ip6_frag_hdr_t *hdr =
+ (ip6_frag_hdr_t *) u8_ptr_add (ip6, frag_hdr_offset);
+ fragment_id = frag_id_6to4 (hdr->identification);
+ frag_offset = ip6_frag_hdr_offset (hdr);
+ frag_more = ip6_frag_hdr_more (hdr);
+ }
+ else
+ {
+ fragment_id = 0;
+ frag_offset = 0;
+ frag_more = 0;
+ }
+
+ nat64_main_t *nm = &nat64_main;
+ nat64_db_bib_entry_t *bibe;
+ nat64_db_st_entry_t *ste;
+ ip46_address_t saddr, daddr, addr;
+ u32 sw_if_index, fib_index;
+ int i;
+ nat64_db_t *db = &nm->db[s_ctx->thread_index];
+
+ sw_if_index = vnet_buffer (s_ctx->b)->sw_if_index[VLIB_RX];
+ fib_index =
+ fib_table_get_index_for_sw_if_index (FIB_PROTOCOL_IP6, sw_if_index);
+
+ saddr.as_u64[0] = ip6->src_address.as_u64[0];
+ saddr.as_u64[1] = ip6->src_address.as_u64[1];
+ daddr.as_u64[0] = ip6->dst_address.as_u64[0];
+ daddr.as_u64[1] = ip6->dst_address.as_u64[1];
+
+ ste =
+ nat64_db_st_entry_find (db, &saddr, &daddr, 0, 0, l4_protocol, fib_index,
+ 1);
+
+ if (ste)
+ {
+ bibe = nat64_db_bib_entry_by_index (db, l4_protocol, ste->bibe_index);
+ if (!bibe)
+ return -1;
+ }
+ else
+ {
+ bibe =
+ nat64_db_bib_entry_find (db, &saddr, 0, l4_protocol, fib_index, 1);
+
+ if (!bibe)
+ {
+ /* Choose same out address as for TCP/UDP session to same dst */
+ unk_proto_st_walk_ctx_t ctx = {
+ .src_addr.as_u64[0] = ip6->src_address.as_u64[0],
+ .src_addr.as_u64[1] = ip6->src_address.as_u64[1],
+ .dst_addr.as_u64[0] = ip6->dst_address.as_u64[0],
+ .dst_addr.as_u64[1] = ip6->dst_address.as_u64[1],
+ .out_addr.as_u32 = 0,
+ .fib_index = fib_index,
+ .proto = l4_protocol,
+ .thread_index = s_ctx->thread_index,
+ };
+
+ nat64_db_st_walk (db, IP_PROTOCOL_TCP, unk_proto_st_walk, &ctx);
+
+ if (!ctx.out_addr.as_u32)
+ nat64_db_st_walk (db, IP_PROTOCOL_UDP, unk_proto_st_walk, &ctx);
+
+ /* Verify if out address is not already in use for protocol */
+ clib_memset (&addr, 0, sizeof (addr));
+ addr.ip4.as_u32 = ctx.out_addr.as_u32;
+ if (nat64_db_bib_entry_find (db, &addr, 0, l4_protocol, 0, 0))
+ ctx.out_addr.as_u32 = 0;
+
+ if (!ctx.out_addr.as_u32)
+ {
+ for (i = 0; i < vec_len (nm->addr_pool); i++)
+ {
+ addr.ip4.as_u32 = nm->addr_pool[i].addr.as_u32;
+ if (!nat64_db_bib_entry_find
+ (db, &addr, 0, l4_protocol, 0, 0))
+ break;
+ }
+ }
+
+ if (!ctx.out_addr.as_u32)
+ return -1;
+
+ bibe =
+ nat64_db_bib_entry_create (s_ctx->thread_index, db,
+ &ip6->src_address, &ctx.out_addr,
+ 0, 0, fib_index, l4_protocol, 0);
+ if (!bibe)
+ return -1;
+
+ vlib_set_simple_counter (&nm->total_bibs, s_ctx->thread_index, 0,
+ db->bib.bib_entries_num);
+ }
+
+ nat64_extract_ip4 (&ip6->dst_address, &daddr.ip4, fib_index);
+ ste =
+ nat64_db_st_entry_create (s_ctx->thread_index, db, bibe,
+ &ip6->dst_address, &daddr.ip4, 0);
+ if (!ste)
+ return -1;
+
+ vlib_set_simple_counter (&nm->total_sessions, s_ctx->thread_index, 0,
+ db->st.st_entries_num);
+ }
+
+ nat64_session_reset_timeout (ste, s_ctx->vm);
+
+ ip4->src_address.as_u32 = bibe->out_addr.as_u32;
+ ip4->dst_address.as_u32 = ste->out_r_addr.as_u32;
+
+ ip4->ip_version_and_header_length =
+ IP4_VERSION_AND_HEADER_LENGTH_NO_OPTIONS;
+ ip4->tos = ip6_translate_tos (ip6->ip_version_traffic_class_and_flow_label);
+ ip4->length = u16_net_add (ip6->payload_length,
+ sizeof (*ip4) + sizeof (*ip6) - l4_offset);
+ ip4->fragment_id = fragment_id;
+ ip4->flags_and_fragment_offset =
+ clib_host_to_net_u16 (frag_offset |
+ (frag_more ? IP4_HEADER_FLAG_MORE_FRAGMENTS : 0));
+ ip4->ttl = ip6->hop_limit;
+ ip4->protocol = l4_protocol;
+ ip4->checksum = ip4_header_checksum (ip4);
+
+ return 0;
+}
+
+static int
+nat64_in2out_tcp_udp_hairpinning (vlib_main_t * vm, vlib_buffer_t * b,
+ ip6_header_t * ip6, u32 l4_offset,
+ u32 thread_index)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_db_bib_entry_t *bibe;
+ nat64_db_st_entry_t *ste;
+ ip46_address_t saddr, daddr;
+ u32 sw_if_index, fib_index;
+ udp_header_t *udp = (udp_header_t *) u8_ptr_add (ip6, l4_offset);
+ tcp_header_t *tcp = (tcp_header_t *) u8_ptr_add (ip6, l4_offset);
+ u8 proto = vnet_buffer (b)->ip.reass.ip_proto;
+ u16 sport = vnet_buffer (b)->ip.reass.l4_src_port;
+ u16 dport = vnet_buffer (b)->ip.reass.l4_dst_port;
+ u16 *checksum = NULL;
+ ip_csum_t csum = 0;
+ nat64_db_t *db = &nm->db[thread_index];
+
+ sw_if_index = vnet_buffer (b)->sw_if_index[VLIB_RX];
+ fib_index =
+ fib_table_get_index_for_sw_if_index (FIB_PROTOCOL_IP6, sw_if_index);
+
+ saddr.as_u64[0] = ip6->src_address.as_u64[0];
+ saddr.as_u64[1] = ip6->src_address.as_u64[1];
+ daddr.as_u64[0] = ip6->dst_address.as_u64[0];
+ daddr.as_u64[1] = ip6->dst_address.as_u64[1];
+
+ if (!vnet_buffer (b)->ip.reass.is_non_first_fragment)
+ {
+ if (proto == IP_PROTOCOL_UDP)
+ checksum = &udp->checksum;
+ else
+ checksum = &tcp->checksum;
+ csum = ip_csum_sub_even (*checksum, ip6->src_address.as_u64[0]);
+ csum = ip_csum_sub_even (csum, ip6->src_address.as_u64[1]);
+ csum = ip_csum_sub_even (csum, ip6->dst_address.as_u64[0]);
+ csum = ip_csum_sub_even (csum, ip6->dst_address.as_u64[1]);
+ }
+
+ ste =
+ nat64_db_st_entry_find (db, &saddr, &daddr, sport, dport, proto,
+ fib_index, 1);
+
+ if (ste)
+ {
+ bibe = nat64_db_bib_entry_by_index (db, proto, ste->bibe_index);
+ if (!bibe)
+ return -1;
+ }
+ else
+ {
+ bibe = nat64_db_bib_entry_find (db, &saddr, sport, proto, fib_index, 1);
+
+ if (!bibe)
+ {
+ u16 out_port;
+ ip4_address_t out_addr;
+ if (nat64_alloc_out_addr_and_port
+ (fib_index, ip_proto_to_nat_proto (proto), &out_addr,
+ &out_port, thread_index))
+ return -1;
+
+ bibe =
+ nat64_db_bib_entry_create (thread_index, db, &ip6->src_address,
+ &out_addr, sport, out_port, fib_index,
+ proto, 0);
+ if (!bibe)
+ return -1;
+
+ vlib_set_simple_counter (&nm->total_bibs, thread_index, 0,
+ db->bib.bib_entries_num);
+ }
+
+ nat64_extract_ip4 (&ip6->dst_address, &daddr.ip4, fib_index);
+ ste =
+ nat64_db_st_entry_create (thread_index, db, bibe, &ip6->dst_address,
+ &daddr.ip4, dport);
+ if (!ste)
+ return -1;
+
+ vlib_set_simple_counter (&nm->total_sessions, thread_index, 0,
+ db->st.st_entries_num);
+ }
+
+ if (proto == IP_PROTOCOL_TCP)
+ nat64_tcp_session_set_state (ste, tcp, 1);
+
+ nat64_session_reset_timeout (ste, vm);
+
+ if (!vnet_buffer (b)->ip.reass.is_non_first_fragment)
+ {
+ udp->src_port = bibe->out_port;
+ }
+
+ nat64_compose_ip6 (&ip6->src_address, &bibe->out_addr, fib_index);
+
+ clib_memset (&daddr, 0, sizeof (daddr));
+ daddr.ip4.as_u32 = ste->out_r_addr.as_u32;
+
+ bibe = 0;
+ /* *INDENT-OFF* */
+ vec_foreach (db, nm->db)
+ {
+ bibe = nat64_db_bib_entry_find (db, &daddr, dport, proto, 0, 0);
+
+ if (bibe)
+ break;
+ }
+ /* *INDENT-ON* */
+
+ if (!bibe)
+ return -1;
+
+ ip6->dst_address.as_u64[0] = bibe->in_addr.as_u64[0];
+ ip6->dst_address.as_u64[1] = bibe->in_addr.as_u64[1];
+
+ if (!vnet_buffer (b)->ip.reass.is_non_first_fragment)
+ {
+ csum = ip_csum_add_even (csum, ip6->src_address.as_u64[0]);
+ csum = ip_csum_add_even (csum, ip6->src_address.as_u64[1]);
+ csum = ip_csum_add_even (csum, ip6->dst_address.as_u64[0]);
+ csum = ip_csum_add_even (csum, ip6->dst_address.as_u64[1]);
+ csum = ip_csum_sub_even (csum, sport);
+ csum = ip_csum_sub_even (csum, dport);
+ udp->dst_port = bibe->in_port;
+ csum = ip_csum_add_even (csum, udp->src_port);
+ csum = ip_csum_add_even (csum, udp->dst_port);
+ *checksum = ip_csum_fold (csum);
+ }
+
+ return 0;
+}
+
+static int
+nat64_in2out_icmp_hairpinning (vlib_main_t * vm, vlib_buffer_t * b,
+ ip6_header_t * ip6, u32 thread_index)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_db_bib_entry_t *bibe;
+ nat64_db_st_entry_t *ste;
+ icmp46_header_t *icmp = ip6_next_header (ip6);
+ ip6_header_t *inner_ip6;
+ ip46_address_t saddr, daddr;
+ u32 sw_if_index, fib_index;
+ u8 proto;
+ udp_header_t *udp;
+ tcp_header_t *tcp;
+ u16 *checksum, sport, dport;
+ ip_csum_t csum;
+ nat64_db_t *db = &nm->db[thread_index];
+
+ if (icmp->type == ICMP6_echo_request || icmp->type == ICMP6_echo_reply)
+ return -1;
+
+ inner_ip6 = (ip6_header_t *) u8_ptr_add (icmp, 8);
+
+ proto = inner_ip6->protocol;
+
+ if (proto == IP_PROTOCOL_ICMP6)
+ return -1;
+
+ sw_if_index = vnet_buffer (b)->sw_if_index[VLIB_RX];
+ fib_index =
+ fib_table_get_index_for_sw_if_index (FIB_PROTOCOL_IP6, sw_if_index);
+
+ saddr.as_u64[0] = inner_ip6->src_address.as_u64[0];
+ saddr.as_u64[1] = inner_ip6->src_address.as_u64[1];
+ daddr.as_u64[0] = inner_ip6->dst_address.as_u64[0];
+ daddr.as_u64[1] = inner_ip6->dst_address.as_u64[1];
+
+ udp = ip6_next_header (inner_ip6);
+ tcp = ip6_next_header (inner_ip6);
+
+ sport = udp->src_port;
+ dport = udp->dst_port;
+
+ if (proto == IP_PROTOCOL_UDP)
+ checksum = &udp->checksum;
+ else
+ checksum = &tcp->checksum;
+
+ csum = ip_csum_sub_even (*checksum, inner_ip6->src_address.as_u64[0]);
+ csum = ip_csum_sub_even (csum, inner_ip6->src_address.as_u64[1]);
+ csum = ip_csum_sub_even (csum, inner_ip6->dst_address.as_u64[0]);
+ csum = ip_csum_sub_even (csum, inner_ip6->dst_address.as_u64[1]);
+ csum = ip_csum_sub_even (csum, sport);
+ csum = ip_csum_sub_even (csum, dport);
+
+ ste =
+ nat64_db_st_entry_find (db, &daddr, &saddr, dport, sport, proto,
+ fib_index, 1);
+ if (!ste)
+ return -1;
+
+ bibe = nat64_db_bib_entry_by_index (db, proto, ste->bibe_index);
+ if (!bibe)
+ return -1;
+
+ dport = udp->dst_port = bibe->out_port;
+ nat64_compose_ip6 (&inner_ip6->dst_address, &bibe->out_addr, fib_index);
+
+ clib_memset (&saddr, 0, sizeof (saddr));
+ clib_memset (&daddr, 0, sizeof (daddr));
+ saddr.ip4.as_u32 = ste->out_r_addr.as_u32;
+ daddr.ip4.as_u32 = bibe->out_addr.as_u32;
+
+ ste = 0;
+ /* *INDENT-OFF* */
+ vec_foreach (db, nm->db)
+ {
+ ste = nat64_db_st_entry_find (db, &saddr, &daddr, sport, dport, proto,
+ 0, 0);
+
+ if (ste)
+ break;
+ }
+ /* *INDENT-ON* */
+
+ if (!ste)
+ return -1;
+
+ bibe = nat64_db_bib_entry_by_index (db, proto, ste->bibe_index);
+ if (!bibe)
+ return -1;
+
+ inner_ip6->src_address.as_u64[0] = bibe->in_addr.as_u64[0];
+ inner_ip6->src_address.as_u64[1] = bibe->in_addr.as_u64[1];
+ udp->src_port = bibe->in_port;
+
+ csum = ip_csum_add_even (csum, inner_ip6->src_address.as_u64[0]);
+ csum = ip_csum_add_even (csum, inner_ip6->src_address.as_u64[1]);
+ csum = ip_csum_add_even (csum, inner_ip6->dst_address.as_u64[0]);
+ csum = ip_csum_add_even (csum, inner_ip6->dst_address.as_u64[1]);
+ csum = ip_csum_add_even (csum, udp->src_port);
+ csum = ip_csum_add_even (csum, udp->dst_port);
+ *checksum = ip_csum_fold (csum);
+
+ if (!vec_len (nm->addr_pool))
+ return -1;
+
+ nat64_compose_ip6 (&ip6->src_address, &nm->addr_pool[0].addr, fib_index);
+ ip6->dst_address.as_u64[0] = inner_ip6->src_address.as_u64[0];
+ ip6->dst_address.as_u64[1] = inner_ip6->src_address.as_u64[1];
+
+ icmp->checksum = 0;
+ csum = ip_csum_with_carry (0, ip6->payload_length);
+ csum = ip_csum_with_carry (csum, clib_host_to_net_u16 (ip6->protocol));
+ csum = ip_csum_with_carry (csum, ip6->src_address.as_u64[0]);
+ csum = ip_csum_with_carry (csum, ip6->src_address.as_u64[1]);
+ csum = ip_csum_with_carry (csum, ip6->dst_address.as_u64[0]);
+ csum = ip_csum_with_carry (csum, ip6->dst_address.as_u64[1]);
+ csum =
+ ip_incremental_checksum (csum, icmp,
+ clib_net_to_host_u16 (ip6->payload_length));
+ icmp->checksum = ~ip_csum_fold (csum);
+
+ return 0;
+}
+
+static int
+nat64_in2out_unk_proto_hairpinning (vlib_main_t * vm, vlib_buffer_t * b,
+ ip6_header_t * ip6, u32 thread_index)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_db_bib_entry_t *bibe;
+ nat64_db_st_entry_t *ste;
+ ip46_address_t saddr, daddr, addr;
+ u32 sw_if_index, fib_index;
+ u8 proto = ip6->protocol;
+ int i;
+ nat64_db_t *db = &nm->db[thread_index];
+
+ sw_if_index = vnet_buffer (b)->sw_if_index[VLIB_RX];
+ fib_index =
+ fib_table_get_index_for_sw_if_index (FIB_PROTOCOL_IP6, sw_if_index);
+
+ saddr.as_u64[0] = ip6->src_address.as_u64[0];
+ saddr.as_u64[1] = ip6->src_address.as_u64[1];
+ daddr.as_u64[0] = ip6->dst_address.as_u64[0];
+ daddr.as_u64[1] = ip6->dst_address.as_u64[1];
+
+ ste =
+ nat64_db_st_entry_find (db, &saddr, &daddr, 0, 0, proto, fib_index, 1);
+
+ if (ste)
+ {
+ bibe = nat64_db_bib_entry_by_index (db, proto, ste->bibe_index);
+ if (!bibe)
+ return -1;
+ }
+ else
+ {
+ bibe = nat64_db_bib_entry_find (db, &saddr, 0, proto, fib_index, 1);
+
+ if (!bibe)
+ {
+ /* Choose same out address as for TCP/UDP session to same dst */
+ unk_proto_st_walk_ctx_t ctx = {
+ .src_addr.as_u64[0] = ip6->src_address.as_u64[0],
+ .src_addr.as_u64[1] = ip6->src_address.as_u64[1],
+ .dst_addr.as_u64[0] = ip6->dst_address.as_u64[0],
+ .dst_addr.as_u64[1] = ip6->dst_address.as_u64[1],
+ .out_addr.as_u32 = 0,
+ .fib_index = fib_index,
+ .proto = proto,
+ .thread_index = thread_index,
+ };
+
+ nat64_db_st_walk (db, IP_PROTOCOL_TCP, unk_proto_st_walk, &ctx);
+
+ if (!ctx.out_addr.as_u32)
+ nat64_db_st_walk (db, IP_PROTOCOL_UDP, unk_proto_st_walk, &ctx);
+
+ /* Verify if out address is not already in use for protocol */
+ clib_memset (&addr, 0, sizeof (addr));
+ addr.ip4.as_u32 = ctx.out_addr.as_u32;
+ if (nat64_db_bib_entry_find (db, &addr, 0, proto, 0, 0))
+ ctx.out_addr.as_u32 = 0;
+
+ if (!ctx.out_addr.as_u32)
+ {
+ for (i = 0; i < vec_len (nm->addr_pool); i++)
+ {
+ addr.ip4.as_u32 = nm->addr_pool[i].addr.as_u32;
+ if (!nat64_db_bib_entry_find (db, &addr, 0, proto, 0, 0))
+ break;
+ }
+ }
+
+ if (!ctx.out_addr.as_u32)
+ return -1;
+
+ bibe =
+ nat64_db_bib_entry_create (thread_index, db, &ip6->src_address,
+ &ctx.out_addr, 0, 0, fib_index, proto,
+ 0);
+ if (!bibe)
+ return -1;
+
+ vlib_set_simple_counter (&nm->total_bibs, thread_index, 0,
+ db->bib.bib_entries_num);
+ }
+
+ nat64_extract_ip4 (&ip6->dst_address, &daddr.ip4, fib_index);
+ ste =
+ nat64_db_st_entry_create (thread_index, db, bibe, &ip6->dst_address,
+ &daddr.ip4, 0);
+ if (!ste)
+ return -1;
+
+ vlib_set_simple_counter (&nm->total_sessions, thread_index, 0,
+ db->st.st_entries_num);
+ }
+
+ nat64_session_reset_timeout (ste, vm);
+
+ nat64_compose_ip6 (&ip6->src_address, &bibe->out_addr, fib_index);
+
+ clib_memset (&daddr, 0, sizeof (daddr));
+ daddr.ip4.as_u32 = ste->out_r_addr.as_u32;
+
+ bibe = 0;
+ /* *INDENT-OFF* */
+ vec_foreach (db, nm->db)
+ {
+ bibe = nat64_db_bib_entry_find (db, &daddr, 0, proto, 0, 0);
+
+ if (bibe)
+ break;
+ }
+ /* *INDENT-ON* */
+
+ if (!bibe)
+ return -1;
+
+ ip6->dst_address.as_u64[0] = bibe->in_addr.as_u64[0];
+ ip6->dst_address.as_u64[1] = bibe->in_addr.as_u64[1];
+
+ return 0;
+}
+
+static inline uword
+nat64_in2out_node_fn_inline (vlib_main_t * vm, vlib_node_runtime_t * node,
+ vlib_frame_t * frame, u8 is_slow_path)
+{
+ u32 n_left_from, *from, *to_next;
+ nat64_in2out_next_t next_index;
+ u32 thread_index = vm->thread_index;
+ nat64_main_t *nm = &nat64_main;
+
+ from = vlib_frame_vector_args (frame);
+ n_left_from = frame->n_vectors;
+ next_index = node->cached_next_index;
+
+ while (n_left_from > 0)
+ {
+ u32 n_left_to_next;
+
+ vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next);
+
+ while (n_left_from > 0 && n_left_to_next > 0)
+ {
+ u32 bi0;
+ vlib_buffer_t *b0;
+ u32 next0;
+ ip6_header_t *ip60;
+ u16 l4_offset0, frag_hdr_offset0;
+ u8 l4_protocol0;
+ u32 proto0;
+ nat64_in2out_set_ctx_t ctx0;
+ u32 sw_if_index0;
+
+ /* speculatively enqueue b0 to the current next frame */
+ bi0 = from[0];
+ to_next[0] = bi0;
+ from += 1;
+ to_next += 1;
+ n_left_from -= 1;
+ n_left_to_next -= 1;
+
+ b0 = vlib_get_buffer (vm, bi0);
+ ip60 = vlib_buffer_get_current (b0);
+
+ sw_if_index0 = vnet_buffer (b0)->sw_if_index[VLIB_RX];
+
+ ctx0.b = b0;
+ ctx0.vm = vm;
+ ctx0.thread_index = thread_index;
+
+ next0 = NAT64_IN2OUT_NEXT_IP4_LOOKUP;
+
+ if (PREDICT_FALSE
+ (ip6_parse
+ (vm, b0, ip60, b0->current_length, &l4_protocol0, &l4_offset0,
+ &frag_hdr_offset0)))
+ {
+ next0 = NAT64_IN2OUT_NEXT_DROP;
+ b0->error = node->errors[NAT64_IN2OUT_ERROR_UNKNOWN];
+ goto trace0;
+ }
+
+ if (nat64_not_translate (sw_if_index0, ip60->dst_address))
+ {
+ next0 = NAT64_IN2OUT_NEXT_IP6_LOOKUP;
+ goto trace0;
+ }
+
+ proto0 = ip_proto_to_nat_proto (l4_protocol0);
+
+ if (is_slow_path)
+ {
+ if (PREDICT_TRUE (proto0 == NAT_PROTOCOL_OTHER))
+ {
+ vlib_increment_simple_counter (&nm->counters.in2out.other,
+ thread_index, sw_if_index0,
+ 1);
+ if (is_hairpinning (&ip60->dst_address))
+ {
+ next0 = NAT64_IN2OUT_NEXT_IP6_LOOKUP;
+ if (nat64_in2out_unk_proto_hairpinning
+ (vm, b0, ip60, thread_index))
+ {
+ next0 = NAT64_IN2OUT_NEXT_DROP;
+ b0->error =
+ node->errors[NAT64_IN2OUT_ERROR_NO_TRANSLATION];
+ }
+ goto trace0;
+ }
+
+ if (nat64_in2out_unk_proto
+ (vm, b0, l4_protocol0, l4_offset0, frag_hdr_offset0,
+ &ctx0))
+ {
+ next0 = NAT64_IN2OUT_NEXT_DROP;
+ b0->error =
+ node->errors[NAT64_IN2OUT_ERROR_NO_TRANSLATION];
+ goto trace0;
+ }
+ }
+ goto trace0;
+ }
+ else
+ {
+ if (PREDICT_FALSE (proto0 == NAT_PROTOCOL_OTHER))
+ {
+ next0 = NAT64_IN2OUT_NEXT_SLOWPATH;
+ goto trace0;
+ }
+ }
+
+ if (proto0 == NAT_PROTOCOL_ICMP)
+ {
+ vlib_increment_simple_counter (&nm->counters.in2out.icmp,
+ thread_index, sw_if_index0, 1);
+ if (is_hairpinning (&ip60->dst_address))
+ {
+ next0 = NAT64_IN2OUT_NEXT_IP6_LOOKUP;
+ if (nat64_in2out_icmp_hairpinning
+ (vm, b0, ip60, thread_index))
+ {
+ next0 = NAT64_IN2OUT_NEXT_DROP;
+ b0->error =
+ node->errors[NAT64_IN2OUT_ERROR_NO_TRANSLATION];
+ }
+ goto trace0;
+ }
+
+ if (icmp6_to_icmp
+ (vm, b0, nat64_in2out_icmp_set_cb, &ctx0,
+ nat64_in2out_inner_icmp_set_cb, &ctx0))
+ {
+ next0 = NAT64_IN2OUT_NEXT_DROP;
+ b0->error = node->errors[NAT64_IN2OUT_ERROR_NO_TRANSLATION];
+ goto trace0;
+ }
+ }
+ else if (proto0 == NAT_PROTOCOL_TCP || proto0 == NAT_PROTOCOL_UDP)
+ {
+ if (proto0 == NAT_PROTOCOL_TCP)
+ vlib_increment_simple_counter (&nm->counters.in2out.tcp,
+ thread_index, sw_if_index0, 1);
+ else
+ vlib_increment_simple_counter (&nm->counters.in2out.udp,
+ thread_index, sw_if_index0, 1);
+
+ if (is_hairpinning (&ip60->dst_address))
+ {
+ next0 = NAT64_IN2OUT_NEXT_IP6_LOOKUP;
+ if (nat64_in2out_tcp_udp_hairpinning
+ (vm, b0, ip60, l4_offset0, thread_index))
+ {
+ next0 = NAT64_IN2OUT_NEXT_DROP;
+ b0->error =
+ node->errors[NAT64_IN2OUT_ERROR_NO_TRANSLATION];
+ }
+ goto trace0;
+ }
+
+ if (nat64_in2out_tcp_udp
+ (vm, b0, l4_offset0, frag_hdr_offset0, &ctx0))
+ {
+ next0 = NAT64_IN2OUT_NEXT_DROP;
+ b0->error = node->errors[NAT64_IN2OUT_ERROR_NO_TRANSLATION];
+ goto trace0;
+ }
+ }
+
+ trace0:
+ if (PREDICT_FALSE ((node->flags & VLIB_NODE_FLAG_TRACE)
+ && (b0->flags & VLIB_BUFFER_IS_TRACED)))
+ {
+ nat64_in2out_trace_t *t =
+ vlib_add_trace (vm, node, b0, sizeof (*t));
+ t->sw_if_index = vnet_buffer (b0)->sw_if_index[VLIB_RX];
+ t->next_index = next0;
+ t->is_slow_path = is_slow_path;
+ }
+
+ if (next0 == NAT64_IN2OUT_NEXT_DROP)
+ {
+ vlib_increment_simple_counter (&nm->counters.in2out.drops,
+ thread_index, sw_if_index0, 1);
+ }
+
+
+ /* verify speculative enqueue, maybe switch current next frame */
+ vlib_validate_buffer_enqueue_x1 (vm, node, next_index, to_next,
+ n_left_to_next, bi0, next0);
+ }
+ vlib_put_next_frame (vm, node, next_index, n_left_to_next);
+ }
+
+ return frame->n_vectors;
+}
+
+VLIB_NODE_FN (nat64_in2out_node) (vlib_main_t * vm,
+ vlib_node_runtime_t * node,
+ vlib_frame_t * frame)
+{
+ return nat64_in2out_node_fn_inline (vm, node, frame, 0);
+}
+
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE (nat64_in2out_node) = {
+ .name = "nat64-in2out",
+ .vector_size = sizeof (u32),
+ .format_trace = format_nat64_in2out_trace,
+ .type = VLIB_NODE_TYPE_INTERNAL,
+ .n_errors = ARRAY_LEN (nat64_in2out_error_strings),
+ .error_strings = nat64_in2out_error_strings,
+ .n_next_nodes = NAT64_IN2OUT_N_NEXT,
+ /* edit / add dispositions here */
+ .next_nodes = {
+ [NAT64_IN2OUT_NEXT_DROP] = "error-drop",
+ [NAT64_IN2OUT_NEXT_IP4_LOOKUP] = "ip4-lookup",
+ [NAT64_IN2OUT_NEXT_IP6_LOOKUP] = "ip6-lookup",
+ [NAT64_IN2OUT_NEXT_SLOWPATH] = "nat64-in2out-slowpath",
+ },
+};
+/* *INDENT-ON* */
+
+VLIB_NODE_FN (nat64_in2out_slowpath_node) (vlib_main_t * vm,
+ vlib_node_runtime_t * node,
+ vlib_frame_t * frame)
+{
+ return nat64_in2out_node_fn_inline (vm, node, frame, 1);
+}
+
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE (nat64_in2out_slowpath_node) = {
+ .name = "nat64-in2out-slowpath",
+ .vector_size = sizeof (u32),
+ .format_trace = format_nat64_in2out_trace,
+ .type = VLIB_NODE_TYPE_INTERNAL,
+ .n_errors = ARRAY_LEN (nat64_in2out_error_strings),
+ .error_strings = nat64_in2out_error_strings,
+ .n_next_nodes = NAT64_IN2OUT_N_NEXT,
+ /* edit / add dispositions here */
+ .next_nodes = {
+ [NAT64_IN2OUT_NEXT_DROP] = "error-drop",
+ [NAT64_IN2OUT_NEXT_IP4_LOOKUP] = "ip4-lookup",
+ [NAT64_IN2OUT_NEXT_IP6_LOOKUP] = "ip6-lookup",
+ [NAT64_IN2OUT_NEXT_SLOWPATH] = "nat64-in2out-slowpath",
+ },
+};
+/* *INDENT-ON* */
+
+typedef struct nat64_in2out_frag_set_ctx_t_
+{
+ vlib_main_t *vm;
+ u32 sess_index;
+ u32 thread_index;
+ u16 l4_offset;
+ u8 proto;
+ u8 first_frag;
+} nat64_in2out_frag_set_ctx_t;
+
+
+#define foreach_nat64_in2out_handoff_error \
+_(CONGESTION_DROP, "congestion drop") \
+_(SAME_WORKER, "same worker") \
+_(DO_HANDOFF, "do handoff")
+
+typedef enum
+{
+#define _(sym,str) NAT64_IN2OUT_HANDOFF_ERROR_##sym,
+ foreach_nat64_in2out_handoff_error
+#undef _
+ NAT64_IN2OUT_HANDOFF_N_ERROR,
+} nat64_in2out_handoff_error_t;
+
+static char *nat64_in2out_handoff_error_strings[] = {
+#define _(sym,string) string,
+ foreach_nat64_in2out_handoff_error
+#undef _
+};
+
+typedef struct
+{
+ u32 next_worker_index;
+} nat64_in2out_handoff_trace_t;
+
+static u8 *
+format_nat64_in2out_handoff_trace (u8 * s, va_list * args)
+{
+ CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
+ CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
+ nat64_in2out_handoff_trace_t *t =
+ va_arg (*args, nat64_in2out_handoff_trace_t *);
+
+ s =
+ format (s, "NAT64-IN2OUT-HANDOFF: next-worker %d", t->next_worker_index);
+
+ return s;
+}
+
+VLIB_NODE_FN (nat64_in2out_handoff_node) (vlib_main_t * vm,
+ vlib_node_runtime_t * node,
+ vlib_frame_t * frame)
+{
+ nat64_main_t *nm = &nat64_main;
+ vlib_buffer_t *bufs[VLIB_FRAME_SIZE], **b;
+ u32 n_enq, n_left_from, *from;
+ u16 thread_indices[VLIB_FRAME_SIZE], *ti;
+ u32 fq_index;
+ u32 thread_index = vm->thread_index;
+ u32 do_handoff = 0, same_worker = 0;
+
+ from = vlib_frame_vector_args (frame);
+ n_left_from = frame->n_vectors;
+ vlib_get_buffers (vm, from, bufs, n_left_from);
+
+ b = bufs;
+ ti = thread_indices;
+
+ fq_index = nm->fq_in2out_index;
+
+ while (n_left_from > 0)
+ {
+ ip6_header_t *ip0;
+
+ ip0 = vlib_buffer_get_current (b[0]);
+ ti[0] = nat64_get_worker_in2out (&ip0->src_address);
+
+ if (ti[0] != thread_index)
+ do_handoff++;
+ else
+ same_worker++;
+
+ if (PREDICT_FALSE
+ ((node->flags & VLIB_NODE_FLAG_TRACE)
+ && (b[0]->flags & VLIB_BUFFER_IS_TRACED)))
+ {
+ nat64_in2out_handoff_trace_t *t =
+ vlib_add_trace (vm, node, b[0], sizeof (*t));
+ t->next_worker_index = ti[0];
+ }
+
+ n_left_from -= 1;
+ ti += 1;
+ b += 1;
+ }
+
+ n_enq =
+ vlib_buffer_enqueue_to_thread (vm, fq_index, from, thread_indices,
+ frame->n_vectors, 1);
+
+ if (n_enq < frame->n_vectors)
+ vlib_node_increment_counter (vm, node->node_index,
+ NAT64_IN2OUT_HANDOFF_ERROR_CONGESTION_DROP,
+ frame->n_vectors - n_enq);
+ vlib_node_increment_counter (vm, node->node_index,
+ NAT64_IN2OUT_HANDOFF_ERROR_SAME_WORKER,
+ same_worker);
+ vlib_node_increment_counter (vm, node->node_index,
+ NAT64_IN2OUT_HANDOFF_ERROR_DO_HANDOFF,
+ do_handoff);
+
+ return frame->n_vectors;
+}
+
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE (nat64_in2out_handoff_node) = {
+ .name = "nat64-in2out-handoff",
+ .vector_size = sizeof (u32),
+ .format_trace = format_nat64_in2out_handoff_trace,
+ .type = VLIB_NODE_TYPE_INTERNAL,
+ .n_errors = ARRAY_LEN(nat64_in2out_handoff_error_strings),
+ .error_strings = nat64_in2out_handoff_error_strings,
+
+ .n_next_nodes = 1,
+
+ .next_nodes = {
+ [0] = "error-drop",
+ },
+};
+/* *INDENT-ON* */
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */
diff --git a/src/plugins/nat/nat64/nat64_out2in.c b/src/plugins/nat/nat64/nat64_out2in.c
new file mode 100644
index 00000000000..108edf0bdd1
--- /dev/null
+++ b/src/plugins/nat/nat64/nat64_out2in.c
@@ -0,0 +1,796 @@
+/*
+ * Copyright (c) 2020 Cisco and/or its affiliates.
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at:
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <nat/nat64/nat64.h>
+#include <vnet/ip/ip4_to_ip6.h>
+#include <vnet/fib/ip4_fib.h>
+#include <vnet/udp/udp.h>
+
+typedef struct
+{
+ u32 sw_if_index;
+ u32 next_index;
+} nat64_out2in_trace_t;
+
+static u8 *
+format_nat64_out2in_trace (u8 * s, va_list * args)
+{
+ CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
+ CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
+ nat64_out2in_trace_t *t = va_arg (*args, nat64_out2in_trace_t *);
+
+ s =
+ format (s, "NAT64-out2in: sw_if_index %d, next index %d", t->sw_if_index,
+ t->next_index);
+
+ return s;
+}
+
+#define foreach_nat64_out2in_error \
+_(UNSUPPORTED_PROTOCOL, "unsupported protocol") \
+_(NO_TRANSLATION, "no translation") \
+_(UNKNOWN, "unknown")
+
+typedef enum
+{
+#define _(sym,str) NAT64_OUT2IN_ERROR_##sym,
+ foreach_nat64_out2in_error
+#undef _
+ NAT64_OUT2IN_N_ERROR,
+} nat64_out2in_error_t;
+
+static char *nat64_out2in_error_strings[] = {
+#define _(sym,string) string,
+ foreach_nat64_out2in_error
+#undef _
+};
+
+typedef enum
+{
+ NAT64_OUT2IN_NEXT_IP6_LOOKUP,
+ NAT64_OUT2IN_NEXT_IP4_LOOKUP,
+ NAT64_OUT2IN_NEXT_DROP,
+ NAT64_OUT2IN_N_NEXT,
+} nat64_out2in_next_t;
+
+typedef struct nat64_out2in_set_ctx_t_
+{
+ vlib_buffer_t *b;
+ vlib_main_t *vm;
+ u32 thread_index;
+} nat64_out2in_set_ctx_t;
+
+static int
+nat64_out2in_tcp_udp (vlib_main_t * vm, vlib_buffer_t * b,
+ nat64_out2in_set_ctx_t * ctx)
+{
+ ip4_header_t *ip4;
+ ip6_header_t *ip6;
+ ip_csum_t csum;
+ u16 *checksum = NULL;
+ ip6_frag_hdr_t *frag;
+ u32 frag_id;
+ ip4_address_t old_src, old_dst;
+
+ nat64_main_t *nm = &nat64_main;
+ nat64_db_bib_entry_t *bibe;
+ nat64_db_st_entry_t *ste;
+ ip46_address_t saddr;
+ ip46_address_t daddr;
+ ip6_address_t ip6_saddr;
+ u8 proto = vnet_buffer (b)->ip.reass.ip_proto;
+ u16 dport = vnet_buffer (b)->ip.reass.l4_dst_port;
+ u16 sport = vnet_buffer (b)->ip.reass.l4_src_port;
+ u32 sw_if_index, fib_index;
+ nat64_db_t *db = &nm->db[ctx->thread_index];
+
+ ip4 = vlib_buffer_get_current (b);
+
+ udp_header_t *udp = ip4_next_header (ip4);
+ tcp_header_t *tcp = ip4_next_header (ip4);
+ if (!vnet_buffer (b)->ip.reass.is_non_first_fragment)
+ {
+ if (ip4->protocol == IP_PROTOCOL_UDP)
+ {
+ checksum = &udp->checksum;
+ //UDP checksum is optional over IPv4 but mandatory for IPv6
+ //We do not check udp->length sanity but use our safe computed value instead
+ if (PREDICT_FALSE (!*checksum))
+ {
+ u16 udp_len =
+ clib_host_to_net_u16 (ip4->length) - sizeof (*ip4);
+ csum = ip_incremental_checksum (0, udp, udp_len);
+ csum =
+ ip_csum_with_carry (csum, clib_host_to_net_u16 (udp_len));
+ csum =
+ ip_csum_with_carry (csum,
+ clib_host_to_net_u16 (IP_PROTOCOL_UDP));
+ csum =
+ ip_csum_with_carry (csum, *((u64 *) (&ip4->src_address)));
+ *checksum = ~ip_csum_fold (csum);
+ }
+ }
+ else
+ {
+ checksum = &tcp->checksum;
+ }
+ }
+
+ old_src.as_u32 = ip4->src_address.as_u32;
+ old_dst.as_u32 = ip4->dst_address.as_u32;
+
+ // Deal with fragmented packets
+ u16 frag_offset = ip4_get_fragment_offset (ip4);
+ if (PREDICT_FALSE (ip4_get_fragment_more (ip4) || frag_offset))
+ {
+ ip6 =
+ (ip6_header_t *) u8_ptr_add (ip4,
+ sizeof (*ip4) - sizeof (*ip6) -
+ sizeof (*frag));
+ frag =
+ (ip6_frag_hdr_t *) u8_ptr_add (ip4, sizeof (*ip4) - sizeof (*frag));
+ frag_id = frag_id_4to6 (ip4->fragment_id);
+ vlib_buffer_advance (b, sizeof (*ip4) - sizeof (*ip6) - sizeof (*frag));
+ }
+ else
+ {
+ ip6 = (ip6_header_t *) (((u8 *) ip4) + sizeof (*ip4) - sizeof (*ip6));
+ vlib_buffer_advance (b, sizeof (*ip4) - sizeof (*ip6));
+ frag = NULL;
+ }
+
+ ip6->ip_version_traffic_class_and_flow_label =
+ clib_host_to_net_u32 ((6 << 28) + (ip4->tos << 20));
+ ip6->payload_length = u16_net_add (ip4->length, -sizeof (*ip4));
+ ip6->hop_limit = ip4->ttl;
+ ip6->protocol = ip4->protocol;
+
+ sw_if_index = vnet_buffer (ctx->b)->sw_if_index[VLIB_RX];
+ fib_index = ip4_fib_table_get_index_for_sw_if_index (sw_if_index);
+
+ clib_memset (&saddr, 0, sizeof (saddr));
+ saddr.ip4.as_u32 = ip4->src_address.as_u32;
+ clib_memset (&daddr, 0, sizeof (daddr));
+ daddr.ip4.as_u32 = ip4->dst_address.as_u32;
+
+ ste =
+ nat64_db_st_entry_find (db, &daddr, &saddr, dport, sport, proto,
+ fib_index, 0);
+ if (ste)
+ {
+ bibe = nat64_db_bib_entry_by_index (db, proto, ste->bibe_index);
+ if (!bibe)
+ return -1;
+ }
+ else
+ {
+ bibe = nat64_db_bib_entry_find (db, &daddr, dport, proto, fib_index, 0);
+
+ if (!bibe)
+ return -1;
+
+ nat64_compose_ip6 (&ip6_saddr, &old_src, bibe->fib_index);
+ ste =
+ nat64_db_st_entry_create (ctx->thread_index, db, bibe, &ip6_saddr,
+ &saddr.ip4, sport);
+
+ if (!ste)
+ return -1;
+
+ vlib_set_simple_counter (&nm->total_sessions, ctx->thread_index, 0,
+ db->st.st_entries_num);
+ }
+
+ ip6->src_address.as_u64[0] = ste->in_r_addr.as_u64[0];
+ ip6->src_address.as_u64[1] = ste->in_r_addr.as_u64[1];
+
+ ip6->dst_address.as_u64[0] = bibe->in_addr.as_u64[0];
+ ip6->dst_address.as_u64[1] = bibe->in_addr.as_u64[1];
+
+ vnet_buffer (ctx->b)->sw_if_index[VLIB_TX] = bibe->fib_index;
+
+ nat64_session_reset_timeout (ste, ctx->vm);
+
+ if (PREDICT_FALSE (frag != NULL))
+ {
+ frag->next_hdr = ip6->protocol;
+ frag->identification = frag_id;
+ frag->rsv = 0;
+ frag->fragment_offset_and_more =
+ ip6_frag_hdr_offset_and_more (frag_offset, 1);
+ ip6->protocol = IP_PROTOCOL_IPV6_FRAGMENTATION;
+ ip6->payload_length = u16_net_add (ip6->payload_length, sizeof (*frag));
+ }
+
+ if (!vnet_buffer (b)->ip.reass.is_non_first_fragment)
+ {
+ udp->dst_port = bibe->in_port;
+
+ if (proto == IP_PROTOCOL_TCP)
+ {
+ nat64_tcp_session_set_state (ste, tcp, 0);
+ }
+
+ csum = ip_csum_sub_even (*checksum, dport);
+ csum = ip_csum_add_even (csum, udp->dst_port);
+ csum = ip_csum_sub_even (csum, old_src.as_u32);
+ csum = ip_csum_sub_even (csum, old_dst.as_u32);
+ csum = ip_csum_add_even (csum, ip6->src_address.as_u64[0]);
+ csum = ip_csum_add_even (csum, ip6->src_address.as_u64[1]);
+ csum = ip_csum_add_even (csum, ip6->dst_address.as_u64[0]);
+ csum = ip_csum_add_even (csum, ip6->dst_address.as_u64[1]);
+ *checksum = ip_csum_fold (csum);
+ }
+
+ return 0;
+}
+
+static int
+nat64_out2in_icmp_set_cb (vlib_buffer_t * b, ip4_header_t * ip4,
+ ip6_header_t * ip6, void *arg)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_out2in_set_ctx_t *ctx = arg;
+ nat64_db_bib_entry_t *bibe;
+ nat64_db_st_entry_t *ste;
+ ip46_address_t saddr, daddr;
+ ip6_address_t ip6_saddr;
+ u32 sw_if_index, fib_index;
+ icmp46_header_t *icmp = ip4_next_header (ip4);
+ nat64_db_t *db = &nm->db[ctx->thread_index];
+
+ sw_if_index = vnet_buffer (ctx->b)->sw_if_index[VLIB_RX];
+ fib_index = ip4_fib_table_get_index_for_sw_if_index (sw_if_index);
+
+ clib_memset (&saddr, 0, sizeof (saddr));
+ saddr.ip4.as_u32 = ip4->src_address.as_u32;
+ clib_memset (&daddr, 0, sizeof (daddr));
+ daddr.ip4.as_u32 = ip4->dst_address.as_u32;
+
+ if (icmp->type == ICMP6_echo_request || icmp->type == ICMP6_echo_reply)
+ {
+ u16 out_id = ((u16 *) (icmp))[2];
+ ste =
+ nat64_db_st_entry_find (db, &daddr, &saddr, out_id, 0,
+ IP_PROTOCOL_ICMP, fib_index, 0);
+
+ if (ste)
+ {
+ bibe =
+ nat64_db_bib_entry_by_index (db, IP_PROTOCOL_ICMP,
+ ste->bibe_index);
+ if (!bibe)
+ return -1;
+ }
+ else
+ {
+ bibe =
+ nat64_db_bib_entry_find (db, &daddr, out_id,
+ IP_PROTOCOL_ICMP, fib_index, 0);
+ if (!bibe)
+ return -1;
+
+ nat64_compose_ip6 (&ip6_saddr, &ip4->src_address, bibe->fib_index);
+ ste =
+ nat64_db_st_entry_create (ctx->thread_index, db,
+ bibe, &ip6_saddr, &saddr.ip4, 0);
+
+ if (!ste)
+ return -1;
+
+ vlib_set_simple_counter (&nm->total_sessions, ctx->thread_index, 0,
+ db->st.st_entries_num);
+ }
+
+ nat64_session_reset_timeout (ste, ctx->vm);
+
+ ip6->src_address.as_u64[0] = ste->in_r_addr.as_u64[0];
+ ip6->src_address.as_u64[1] = ste->in_r_addr.as_u64[1];
+
+ ip6->dst_address.as_u64[0] = bibe->in_addr.as_u64[0];
+ ip6->dst_address.as_u64[1] = bibe->in_addr.as_u64[1];
+ ((u16 *) (icmp))[2] = bibe->in_port;
+
+ vnet_buffer (ctx->b)->sw_if_index[VLIB_TX] = bibe->fib_index;
+ }
+ else
+ {
+ ip6_header_t *inner_ip6 = (ip6_header_t *) u8_ptr_add (icmp, 8);
+
+ nat64_compose_ip6 (&ip6->src_address, &ip4->src_address,
+ vnet_buffer (ctx->b)->sw_if_index[VLIB_TX]);
+ ip6->dst_address.as_u64[0] = inner_ip6->src_address.as_u64[0];
+ ip6->dst_address.as_u64[1] = inner_ip6->src_address.as_u64[1];
+ }
+
+ return 0;
+}
+
+static int
+nat64_out2in_inner_icmp_set_cb (vlib_buffer_t * b, ip4_header_t * ip4,
+ ip6_header_t * ip6, void *arg)
+{
+ nat64_main_t *nm = &nat64_main;
+ nat64_out2in_set_ctx_t *ctx = arg;
+ nat64_db_bib_entry_t *bibe;
+ nat64_db_st_entry_t *ste;
+ ip46_address_t saddr, daddr;
+ u32 sw_if_index, fib_index;
+ u8 proto = ip4->protocol;
+ nat64_db_t *db = &nm->db[ctx->thread_index];
+
+ sw_if_index = vnet_buffer (ctx->b)->sw_if_index[VLIB_RX];
+ fib_index =
+ fib_table_get_index_for_sw_if_index (FIB_PROTOCOL_IP6, sw_if_index);
+
+ clib_memset (&saddr, 0, sizeof (saddr));
+ saddr.ip4.as_u32 = ip4->src_address.as_u32;
+ clib_memset (&daddr, 0, sizeof (daddr));
+ daddr.ip4.as_u32 = ip4->dst_address.as_u32;
+
+ if (proto == IP_PROTOCOL_ICMP6)
+ {
+ icmp46_header_t *icmp = ip4_next_header (ip4);
+ u16 out_id = ((u16 *) (icmp))[2];
+ proto = IP_PROTOCOL_ICMP;
+
+ if (!
+ (icmp->type == ICMP6_echo_request
+ || icmp->type == ICMP6_echo_reply))
+ return -1;
+
+ ste =
+ nat64_db_st_entry_find (db, &saddr, &daddr, out_id, 0, proto,
+ fib_index, 0);
+ if (!ste)
+ return -1;
+
+ bibe = nat64_db_bib_entry_by_index (db, proto, ste->bibe_index);
+ if (!bibe)
+ return -1;
+
+ ip6->dst_address.as_u64[0] = ste->in_r_addr.as_u64[0];
+ ip6->dst_address.as_u64[1] = ste->in_r_addr.as_u64[1];
+ ip6->src_address.as_u64[0] = bibe->in_addr.as_u64[0];
+ ip6->src_address.as_u64[1] = bibe->in_addr.as_u64[1];
+ ((u16 *) (icmp))[2] = bibe->in_port;
+
+ vnet_buffer (ctx->b)->sw_if_index[VLIB_TX] = bibe->fib_index;
+ }
+ else
+ {
+ udp_header_t *udp = ip4_next_header (ip4);
+ tcp_header_t *tcp = ip4_next_header (ip4);
+ u16 dport = udp->dst_port;
+ u16 sport = udp->src_port;
+ u16 *checksum;
+ ip_csum_t csum;
+
+ ste =
+ nat64_db_st_entry_find (db, &saddr, &daddr, sport, dport, proto,
+ fib_index, 0);
+ if (!ste)
+ return -1;
+
+ bibe = nat64_db_bib_entry_by_index (db, proto, ste->bibe_index);
+ if (!bibe)
+ return -1;
+
+ nat64_compose_ip6 (&ip6->dst_address, &daddr.ip4, bibe->fib_index);
+ ip6->src_address.as_u64[0] = bibe->in_addr.as_u64[0];
+ ip6->src_address.as_u64[1] = bibe->in_addr.as_u64[1];
+ udp->src_port = bibe->in_port;
+
+ if (proto == IP_PROTOCOL_UDP)
+ checksum = &udp->checksum;
+ else
+ checksum = &tcp->checksum;
+ if (*checksum)
+ {
+ csum = ip_csum_sub_even (*checksum, sport);
+ csum = ip_csum_add_even (csum, udp->src_port);
+ *checksum = ip_csum_fold (csum);
+ }
+
+ vnet_buffer (ctx->b)->sw_if_index[VLIB_TX] = bibe->fib_index;
+ }
+
+ return 0;
+}
+
+static int
+nat64_out2in_unk_proto (vlib_main_t * vm, vlib_buffer_t * p,
+ nat64_out2in_set_ctx_t * ctx)
+{
+ ip4_header_t *ip4 = vlib_buffer_get_current (p);
+ ip6_header_t *ip6;
+ ip6_frag_hdr_t *frag;
+ u32 frag_id;
+
+ nat64_main_t *nm = &nat64_main;
+ nat64_db_bib_entry_t *bibe;
+ nat64_db_st_entry_t *ste;
+ ip46_address_t saddr, daddr;
+ ip6_address_t ip6_saddr;
+ u32 sw_if_index, fib_index;
+ u8 proto = ip4->protocol;
+ nat64_db_t *db = &nm->db[ctx->thread_index];
+
+ // Deal with fragmented packets
+ u16 frag_offset = ip4_get_fragment_offset (ip4);
+ if (PREDICT_FALSE (ip4_get_fragment_more (ip4) || frag_offset))
+ {
+ ip6 =
+ (ip6_header_t *) u8_ptr_add (ip4,
+ sizeof (*ip4) - sizeof (*ip6) -
+ sizeof (*frag));
+ frag =
+ (ip6_frag_hdr_t *) u8_ptr_add (ip4, sizeof (*ip4) - sizeof (*frag));
+ frag_id = frag_id_4to6 (ip4->fragment_id);
+ vlib_buffer_advance (p, sizeof (*ip4) - sizeof (*ip6) - sizeof (*frag));
+ }
+ else
+ {
+ ip6 = (ip6_header_t *) (((u8 *) ip4) + sizeof (*ip4) - sizeof (*ip6));
+ vlib_buffer_advance (p, sizeof (*ip4) - sizeof (*ip6));
+ frag = NULL;
+ }
+
+ ip6->ip_version_traffic_class_and_flow_label =
+ clib_host_to_net_u32 ((6 << 28) + (ip4->tos << 20));
+ ip6->payload_length = u16_net_add (ip4->length, -sizeof (*ip4));
+ ip6->hop_limit = ip4->ttl;
+ ip6->protocol = ip4->protocol;
+
+ if (PREDICT_FALSE (frag != NULL))
+ {
+ frag->next_hdr = ip6->protocol;
+ frag->identification = frag_id;
+ frag->rsv = 0;
+ frag->fragment_offset_and_more =
+ ip6_frag_hdr_offset_and_more (frag_offset, 1);
+ ip6->protocol = IP_PROTOCOL_IPV6_FRAGMENTATION;
+ ip6->payload_length = u16_net_add (ip6->payload_length, sizeof (*frag));
+ }
+
+ sw_if_index = vnet_buffer (ctx->b)->sw_if_index[VLIB_RX];
+ fib_index = ip4_fib_table_get_index_for_sw_if_index (sw_if_index);
+
+ clib_memset (&saddr, 0, sizeof (saddr));
+ saddr.ip4.as_u32 = ip4->src_address.as_u32;
+ clib_memset (&daddr, 0, sizeof (daddr));
+ daddr.ip4.as_u32 = ip4->dst_address.as_u32;
+
+ ste =
+ nat64_db_st_entry_find (db, &daddr, &saddr, 0, 0, proto, fib_index, 0);
+ if (ste)
+ {
+ bibe = nat64_db_bib_entry_by_index (db, proto, ste->bibe_index);
+ if (!bibe)
+ return -1;
+ }
+ else
+ {
+ bibe = nat64_db_bib_entry_find (db, &daddr, 0, proto, fib_index, 0);
+
+ if (!bibe)
+ return -1;
+
+ nat64_compose_ip6 (&ip6_saddr, &ip4->src_address, bibe->fib_index);
+ ste = nat64_db_st_entry_create (ctx->thread_index, db,
+ bibe, &ip6_saddr, &saddr.ip4, 0);
+
+ if (!ste)
+ return -1;
+
+ vlib_set_simple_counter (&nm->total_sessions, ctx->thread_index, 0,
+ db->st.st_entries_num);
+ }
+
+ nat64_session_reset_timeout (ste, ctx->vm);
+
+ ip6->src_address.as_u64[0] = ste->in_r_addr.as_u64[0];
+ ip6->src_address.as_u64[1] = ste->in_r_addr.as_u64[1];
+
+ ip6->dst_address.as_u64[0] = bibe->in_addr.as_u64[0];
+ ip6->dst_address.as_u64[1] = bibe->in_addr.as_u64[1];
+
+ vnet_buffer (ctx->b)->sw_if_index[VLIB_TX] = bibe->fib_index;
+
+ return 0;
+}
+
+VLIB_NODE_FN (nat64_out2in_node) (vlib_main_t * vm,
+ vlib_node_runtime_t * node,
+ vlib_frame_t * frame)
+{
+ u32 n_left_from, *from, *to_next;
+ nat64_out2in_next_t next_index;
+ nat64_main_t *nm = &nat64_main;
+ u32 thread_index = vm->thread_index;
+
+ from = vlib_frame_vector_args (frame);
+ n_left_from = frame->n_vectors;
+ next_index = node->cached_next_index;
+ while (n_left_from > 0)
+ {
+ u32 n_left_to_next;
+
+ vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next);
+
+ while (n_left_from > 0 && n_left_to_next > 0)
+ {
+ u32 bi0;
+ vlib_buffer_t *b0;
+ u32 next0;
+ ip4_header_t *ip40;
+ u32 proto0;
+ nat64_out2in_set_ctx_t ctx0;
+ udp_header_t *udp0;
+ u32 sw_if_index0;
+
+ /* speculatively enqueue b0 to the current next frame */
+ bi0 = from[0];
+ to_next[0] = bi0;
+ from += 1;
+ to_next += 1;
+ n_left_from -= 1;
+ n_left_to_next -= 1;
+
+ b0 = vlib_get_buffer (vm, bi0);
+ ip40 = vlib_buffer_get_current (b0);
+
+ ctx0.b = b0;
+ ctx0.vm = vm;
+ ctx0.thread_index = thread_index;
+
+ next0 = NAT64_OUT2IN_NEXT_IP6_LOOKUP;
+
+ proto0 = ip_proto_to_nat_proto (ip40->protocol);
+ sw_if_index0 = vnet_buffer (b0)->sw_if_index[VLIB_RX];
+ if (PREDICT_FALSE (proto0 == NAT_PROTOCOL_OTHER))
+ {
+ if (nat64_out2in_unk_proto (vm, b0, &ctx0))
+ {
+ next0 = NAT64_OUT2IN_NEXT_DROP;
+ b0->error = node->errors[NAT64_OUT2IN_ERROR_NO_TRANSLATION];
+ }
+ vlib_increment_simple_counter (&nm->counters.out2in.other,
+ thread_index, sw_if_index0, 1);
+ goto trace0;
+ }
+
+ if (proto0 == NAT_PROTOCOL_ICMP)
+ {
+ vlib_increment_simple_counter (&nm->counters.out2in.icmp,
+ thread_index, sw_if_index0, 1);
+ if (icmp_to_icmp6
+ (b0, nat64_out2in_icmp_set_cb, &ctx0,
+ nat64_out2in_inner_icmp_set_cb, &ctx0))
+ {
+ next0 = NAT64_OUT2IN_NEXT_DROP;
+ b0->error = node->errors[NAT64_OUT2IN_ERROR_NO_TRANSLATION];
+ goto trace0;
+ }
+ }
+ else
+ {
+ if (proto0 == NAT_PROTOCOL_TCP)
+ vlib_increment_simple_counter (&nm->counters.out2in.tcp,
+ thread_index, sw_if_index0, 1);
+ else
+ vlib_increment_simple_counter (&nm->counters.out2in.udp,
+ thread_index, sw_if_index0, 1);
+
+ if (nat64_out2in_tcp_udp (vm, b0, &ctx0))
+ {
+ udp0 = ip4_next_header (ip40);
+ /*
+ * Send DHCP packets to the ipv4 stack, or we won't
+ * be able to use dhcp client on the outside interface
+ */
+ if ((proto0 == NAT_PROTOCOL_UDP)
+ && (udp0->dst_port ==
+ clib_host_to_net_u16 (UDP_DST_PORT_dhcp_to_client)))
+ {
+ next0 = NAT64_OUT2IN_NEXT_IP4_LOOKUP;
+ goto trace0;
+ }
+ next0 = NAT64_OUT2IN_NEXT_DROP;
+ b0->error = node->errors[NAT64_OUT2IN_ERROR_NO_TRANSLATION];
+ goto trace0;
+ }
+ }
+
+ trace0:
+ if (PREDICT_FALSE ((node->flags & VLIB_NODE_FLAG_TRACE)
+ && (b0->flags & VLIB_BUFFER_IS_TRACED)))
+ {
+ nat64_out2in_trace_t *t =
+ vlib_add_trace (vm, node, b0, sizeof (*t));
+ t->sw_if_index = vnet_buffer (b0)->sw_if_index[VLIB_RX];
+ t->next_index = next0;
+ }
+
+ if (next0 == NAT64_OUT2IN_NEXT_DROP)
+ {
+ vlib_increment_simple_counter (&nm->counters.out2in.drops,
+ thread_index, sw_if_index0, 1);
+ }
+
+ /* verify speculative enqueue, maybe switch current next frame */
+ vlib_validate_buffer_enqueue_x1 (vm, node, next_index, to_next,
+ n_left_to_next, bi0, next0);
+ }
+ vlib_put_next_frame (vm, node, next_index, n_left_to_next);
+ }
+ return frame->n_vectors;
+}
+
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE (nat64_out2in_node) = {
+ .name = "nat64-out2in",
+ .vector_size = sizeof (u32),
+ .format_trace = format_nat64_out2in_trace,
+ .type = VLIB_NODE_TYPE_INTERNAL,
+ .n_errors = ARRAY_LEN (nat64_out2in_error_strings),
+ .error_strings = nat64_out2in_error_strings,
+ .n_next_nodes = NAT64_OUT2IN_N_NEXT,
+ /* edit / add dispositions here */
+ .next_nodes = {
+ [NAT64_OUT2IN_NEXT_DROP] = "error-drop",
+ [NAT64_OUT2IN_NEXT_IP6_LOOKUP] = "ip6-lookup",
+ [NAT64_OUT2IN_NEXT_IP4_LOOKUP] = "ip4-lookup",
+ },
+};
+/* *INDENT-ON* */
+
+typedef struct nat64_out2in_frag_set_ctx_t_
+{
+ vlib_main_t *vm;
+ vlib_buffer_t *b;
+ u32 sess_index;
+ u32 thread_index;
+ u8 proto;
+ u8 first_frag;
+} nat64_out2in_frag_set_ctx_t;
+
+#define foreach_nat64_out2in_handoff_error \
+_(CONGESTION_DROP, "congestion drop") \
+_(SAME_WORKER, "same worker") \
+_(DO_HANDOFF, "do handoff")
+
+typedef enum
+{
+#define _(sym,str) NAT64_OUT2IN_HANDOFF_ERROR_##sym,
+ foreach_nat64_out2in_handoff_error
+#undef _
+ NAT64_OUT2IN_HANDOFF_N_ERROR,
+} nat64_out2in_handoff_error_t;
+
+static char *nat64_out2in_handoff_error_strings[] = {
+#define _(sym,string) string,
+ foreach_nat64_out2in_handoff_error
+#undef _
+};
+
+typedef struct
+{
+ u32 next_worker_index;
+} nat64_out2in_handoff_trace_t;
+
+static u8 *
+format_nat64_out2in_handoff_trace (u8 * s, va_list * args)
+{
+ CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *);
+ CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *);
+ nat64_out2in_handoff_trace_t *t =
+ va_arg (*args, nat64_out2in_handoff_trace_t *);
+
+ s =
+ format (s, "NAT64-OUT2IN-HANDOFF: next-worker %d", t->next_worker_index);
+
+ return s;
+}
+
+VLIB_NODE_FN (nat64_out2in_handoff_node) (vlib_main_t * vm,
+ vlib_node_runtime_t * node,
+ vlib_frame_t * frame)
+{
+ nat64_main_t *nm = &nat64_main;
+ vlib_buffer_t *bufs[VLIB_FRAME_SIZE], **b;
+ u32 n_enq, n_left_from, *from;
+ u16 thread_indices[VLIB_FRAME_SIZE], *ti;
+ u32 fq_index;
+ u32 thread_index = vm->thread_index;
+ u32 do_handoff = 0, same_worker = 0;
+
+ from = vlib_frame_vector_args (frame);
+ n_left_from = frame->n_vectors;
+ vlib_get_buffers (vm, from, bufs, n_left_from);
+
+ b = bufs;
+ ti = thread_indices;
+
+ fq_index = nm->fq_out2in_index;
+
+ while (n_left_from > 0)
+ {
+ ip4_header_t *ip0;
+
+ ip0 = vlib_buffer_get_current (b[0]);
+ ti[0] = nat64_get_worker_out2in (b[0], ip0);
+
+ if (ti[0] != thread_index)
+ do_handoff++;
+ else
+ same_worker++;
+
+ if (PREDICT_FALSE
+ ((node->flags & VLIB_NODE_FLAG_TRACE)
+ && (b[0]->flags & VLIB_BUFFER_IS_TRACED)))
+ {
+ nat64_out2in_handoff_trace_t *t =
+ vlib_add_trace (vm, node, b[0], sizeof (*t));
+ t->next_worker_index = ti[0];
+ }
+
+ n_left_from -= 1;
+ ti += 1;
+ b += 1;
+ }
+
+ n_enq =
+ vlib_buffer_enqueue_to_thread (vm, fq_index, from, thread_indices,
+ frame->n_vectors, 1);
+
+ if (n_enq < frame->n_vectors)
+ vlib_node_increment_counter (vm, node->node_index,
+ NAT64_OUT2IN_HANDOFF_ERROR_CONGESTION_DROP,
+ frame->n_vectors - n_enq);
+ vlib_node_increment_counter (vm, node->node_index,
+ NAT64_OUT2IN_HANDOFF_ERROR_SAME_WORKER,
+ same_worker);
+ vlib_node_increment_counter (vm, node->node_index,
+ NAT64_OUT2IN_HANDOFF_ERROR_DO_HANDOFF,
+ do_handoff);
+
+ return frame->n_vectors;
+}
+
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE (nat64_out2in_handoff_node) = {
+ .name = "nat64-out2in-handoff",
+ .vector_size = sizeof (u32),
+ .format_trace = format_nat64_out2in_handoff_trace,
+ .type = VLIB_NODE_TYPE_INTERNAL,
+ .n_errors = ARRAY_LEN(nat64_out2in_handoff_error_strings),
+ .error_strings = nat64_out2in_handoff_error_strings,
+
+ .n_next_nodes = 1,
+
+ .next_nodes = {
+ [0] = "error-drop",
+ },
+};
+/* *INDENT-ON* */
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */