summaryrefslogtreecommitdiffstats
path: root/src/plugins/map/ip4_map_t.c
diff options
context:
space:
mode:
authorOle Troan <ot@cisco.com>2018-06-22 11:32:12 +0200
committerFlorin Coras <florin.coras@gmail.com>2018-06-25 20:19:40 +0000
commit381e9a90748bb659f56081123052e3e95501a4b4 (patch)
treefbc166ecf8637b8b9ea262f3ac4181396149d534 /src/plugins/map/ip4_map_t.c
parente4d5a652a392c9f74501da5778a7fe161b183476 (diff)
MAP: Move MAP-E/T to a plugin.
Only remaining traces of MAP in the src/vnet is now in buffer.h. Awaiting a new buffer opaque API (hint, hint). Change-Id: Ie165561484731f1d7ed6e0f604b43624e06db3f0 Signed-off-by: Ole Troan <ot@cisco.com>
Diffstat (limited to 'src/plugins/map/ip4_map_t.c')
-rw-r--r--src/plugins/map/ip4_map_t.c898
1 files changed, 898 insertions, 0 deletions
diff --git a/src/plugins/map/ip4_map_t.c b/src/plugins/map/ip4_map_t.c
new file mode 100644
index 00000000000..ab4b46c916f
--- /dev/null
+++ b/src/plugins/map/ip4_map_t.c
@@ -0,0 +1,898 @@
+/*
+ * Copyright (c) 2015 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 "map.h"
+
+#include <vnet/ip/ip_frag.h>
+#include <vnet/ip/ip4_to_ip6.h>
+
+#define IP4_MAP_T_DUAL_LOOP 1
+
+typedef enum
+{
+ IP4_MAPT_NEXT_MAPT_TCP_UDP,
+ IP4_MAPT_NEXT_MAPT_ICMP,
+ IP4_MAPT_NEXT_MAPT_FRAGMENTED,
+ IP4_MAPT_NEXT_DROP,
+ IP4_MAPT_N_NEXT
+} ip4_mapt_next_t;
+
+typedef enum
+{
+ IP4_MAPT_ICMP_NEXT_IP6_LOOKUP,
+ IP4_MAPT_ICMP_NEXT_IP6_FRAG,
+ IP4_MAPT_ICMP_NEXT_DROP,
+ IP4_MAPT_ICMP_N_NEXT
+} ip4_mapt_icmp_next_t;
+
+typedef enum
+{
+ IP4_MAPT_TCP_UDP_NEXT_IP6_LOOKUP,
+ IP4_MAPT_TCP_UDP_NEXT_IP6_FRAG,
+ IP4_MAPT_TCP_UDP_NEXT_DROP,
+ IP4_MAPT_TCP_UDP_N_NEXT
+} ip4_mapt_tcp_udp_next_t;
+
+typedef enum
+{
+ IP4_MAPT_FRAGMENTED_NEXT_IP6_LOOKUP,
+ IP4_MAPT_FRAGMENTED_NEXT_IP6_FRAG,
+ IP4_MAPT_FRAGMENTED_NEXT_DROP,
+ IP4_MAPT_FRAGMENTED_N_NEXT
+} ip4_mapt_fragmented_next_t;
+
+//This is used to pass information within the buffer data.
+//Buffer structure being too small to contain big structures like this.
+/* *INDENT-OFF* */
+typedef CLIB_PACKED (struct {
+ ip6_address_t daddr;
+ ip6_address_t saddr;
+ //IPv6 header + Fragmentation header will be here
+ //sizeof(ip6) + sizeof(ip_frag) - sizeof(ip4)
+ u8 unused[28];
+}) ip4_mapt_pseudo_header_t;
+/* *INDENT-ON* */
+
+
+static_always_inline int
+ip4_map_fragment_cache (ip4_header_t * ip4, u16 port)
+{
+ u32 *ignore = NULL;
+ map_ip4_reass_lock ();
+ map_ip4_reass_t *r =
+ map_ip4_reass_get (ip4->src_address.as_u32, ip4->dst_address.as_u32,
+ ip4->fragment_id,
+ (ip4->protocol ==
+ IP_PROTOCOL_ICMP) ? IP_PROTOCOL_ICMP6 : ip4->protocol,
+ &ignore);
+ if (r)
+ r->port = port;
+
+ map_ip4_reass_unlock ();
+ return !r;
+}
+
+static_always_inline i32
+ip4_map_fragment_get_port (ip4_header_t * ip4)
+{
+ u32 *ignore = NULL;
+ map_ip4_reass_lock ();
+ map_ip4_reass_t *r =
+ map_ip4_reass_get (ip4->src_address.as_u32, ip4->dst_address.as_u32,
+ ip4->fragment_id,
+ (ip4->protocol ==
+ IP_PROTOCOL_ICMP) ? IP_PROTOCOL_ICMP6 : ip4->protocol,
+ &ignore);
+ i32 ret = r ? r->port : -1;
+ map_ip4_reass_unlock ();
+ return ret;
+}
+
+typedef struct
+{
+ map_domain_t *d;
+ u16 id;
+} icmp_to_icmp6_ctx_t;
+
+static int
+ip4_to_ip6_set_icmp_cb (ip4_header_t * ip4, ip6_header_t * ip6, void *arg)
+{
+ icmp_to_icmp6_ctx_t *ctx = arg;
+ map_main_t *mm = &map_main;
+
+ if (mm->is_ce)
+ {
+ ip6->src_address.as_u64[0] =
+ map_get_pfx_net (ctx->d, ip4->src_address.as_u32, ctx->id);
+ ip6->src_address.as_u64[1] =
+ map_get_sfx_net (ctx->d, ip4->src_address.as_u32, ctx->id);
+ ip4_map_t_embedded_address (ctx->d, &ip6->dst_address,
+ &ip4->dst_address);
+ }
+ else
+ {
+ ip4_map_t_embedded_address (ctx->d, &ip6->src_address,
+ &ip4->src_address);
+ ip6->dst_address.as_u64[0] =
+ map_get_pfx_net (ctx->d, ip4->dst_address.as_u32, ctx->id);
+ ip6->dst_address.as_u64[1] =
+ map_get_sfx_net (ctx->d, ip4->dst_address.as_u32, ctx->id);
+ }
+
+ return 0;
+}
+
+static int
+ip4_to_ip6_set_inner_icmp_cb (ip4_header_t * ip4, ip6_header_t * ip6,
+ void *arg)
+{
+ icmp_to_icmp6_ctx_t *ctx = arg;
+ map_main_t *mm = &map_main;
+
+ if (mm->is_ce)
+ {
+ //Note that the destination address is within the domain
+ //while the source address is the one outside the domain
+ ip4_map_t_embedded_address (ctx->d, &ip6->src_address,
+ &ip4->src_address);
+ ip6->dst_address.as_u64[0] =
+ map_get_pfx_net (ctx->d, ip4->dst_address.as_u32, ctx->id);
+ ip6->dst_address.as_u64[1] =
+ map_get_sfx_net (ctx->d, ip4->dst_address.as_u32, ctx->id);
+ }
+ else
+ {
+ //Note that the source address is within the domain
+ //while the destination address is the one outside the domain
+ ip4_map_t_embedded_address (ctx->d, &ip6->dst_address,
+ &ip4->dst_address);
+ ip6->src_address.as_u64[0] =
+ map_get_pfx_net (ctx->d, ip4->src_address.as_u32, ctx->id);
+ ip6->src_address.as_u64[1] =
+ map_get_sfx_net (ctx->d, ip4->src_address.as_u32, ctx->id);
+ }
+
+ return 0;
+}
+
+static uword
+ip4_map_t_icmp (vlib_main_t * vm,
+ vlib_node_runtime_t * node, vlib_frame_t * frame)
+{
+ u32 n_left_from, *from, next_index, *to_next, n_left_to_next;
+ vlib_node_runtime_t *error_node =
+ vlib_node_get_runtime (vm, ip4_map_t_icmp_node.index);
+ from = vlib_frame_vector_args (frame);
+ n_left_from = frame->n_vectors;
+ next_index = node->cached_next_index;
+ vlib_combined_counter_main_t *cm = map_main.domain_counters;
+ u32 thread_index = vlib_get_thread_index ();
+
+ while (n_left_from > 0)
+ {
+ 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 pi0;
+ vlib_buffer_t *p0;
+ ip4_mapt_icmp_next_t next0;
+ u8 error0;
+ map_domain_t *d0;
+ u16 len0;
+ icmp_to_icmp6_ctx_t ctx0;
+ ip4_header_t *ip40;
+ icmp46_header_t *icmp0;
+
+ next0 = IP4_MAPT_ICMP_NEXT_IP6_LOOKUP;
+ pi0 = to_next[0] = from[0];
+ from += 1;
+ n_left_from -= 1;
+ to_next += 1;
+ n_left_to_next -= 1;
+ error0 = MAP_ERROR_NONE;
+
+ p0 = vlib_get_buffer (vm, pi0);
+ vlib_buffer_advance (p0, sizeof (ip4_mapt_pseudo_header_t)); //The pseudo-header is not used
+ len0 =
+ clib_net_to_host_u16 (((ip4_header_t *)
+ vlib_buffer_get_current (p0))->length);
+ d0 =
+ pool_elt_at_index (map_main.domains,
+ vnet_buffer (p0)->map_t.map_domain_index);
+
+ ip40 = vlib_buffer_get_current (p0);
+ icmp0 = (icmp46_header_t *) (ip40 + 1);
+
+ ctx0.id = ip4_get_port (ip40, icmp0->type == ICMP6_echo_request);
+ ctx0.d = d0;
+ if (ctx0.id == 0)
+ {
+ // In case of 1:1 mapping, we don't care about the port
+ if (!(d0->ea_bits_len == 0 && d0->rules))
+ {
+ error0 = MAP_ERROR_ICMP;
+ goto err0;
+ }
+ }
+
+ if (icmp_to_icmp6
+ (p0, ip4_to_ip6_set_icmp_cb, &ctx0,
+ ip4_to_ip6_set_inner_icmp_cb, &ctx0))
+ {
+ error0 = MAP_ERROR_ICMP;
+ goto err0;
+ }
+
+ if (vnet_buffer (p0)->map_t.mtu < p0->current_length)
+ {
+ vnet_buffer (p0)->ip_frag.header_offset = 0;
+ vnet_buffer (p0)->ip_frag.mtu = vnet_buffer (p0)->map_t.mtu;
+ vnet_buffer (p0)->ip_frag.next_index = IP6_FRAG_NEXT_IP6_LOOKUP;
+ next0 = IP4_MAPT_ICMP_NEXT_IP6_FRAG;
+ }
+ err0:
+ if (PREDICT_TRUE (error0 == MAP_ERROR_NONE))
+ {
+ vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_TX,
+ thread_index,
+ vnet_buffer (p0)->
+ map_t.map_domain_index, 1,
+ len0);
+ }
+ else
+ {
+ next0 = IP4_MAPT_ICMP_NEXT_DROP;
+ }
+ p0->error = error_node->errors[error0];
+ vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
+ to_next, n_left_to_next, pi0,
+ next0);
+ }
+ vlib_put_next_frame (vm, node, next_index, n_left_to_next);
+ }
+ return frame->n_vectors;
+}
+
+static int
+ip4_to_ip6_set_cb (ip4_header_t * ip4, ip6_header_t * ip6, void *ctx)
+{
+ ip4_mapt_pseudo_header_t *pheader = ctx;
+
+ ip6->dst_address.as_u64[0] = pheader->daddr.as_u64[0];
+ ip6->dst_address.as_u64[1] = pheader->daddr.as_u64[1];
+ ip6->src_address.as_u64[0] = pheader->saddr.as_u64[0];
+ ip6->src_address.as_u64[1] = pheader->saddr.as_u64[1];
+
+ return 0;
+}
+
+static uword
+ip4_map_t_fragmented (vlib_main_t * vm,
+ vlib_node_runtime_t * node, vlib_frame_t * frame)
+{
+ u32 n_left_from, *from, next_index, *to_next, n_left_to_next;
+ from = vlib_frame_vector_args (frame);
+ n_left_from = frame->n_vectors;
+ next_index = node->cached_next_index;
+ vlib_node_runtime_t *error_node =
+ vlib_node_get_runtime (vm, ip4_map_t_fragmented_node.index);
+
+ while (n_left_from > 0)
+ {
+ 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 pi0;
+ vlib_buffer_t *p0;
+ ip4_mapt_pseudo_header_t *pheader0;
+ ip4_mapt_fragmented_next_t next0;
+
+ next0 = IP4_MAPT_FRAGMENTED_NEXT_IP6_LOOKUP;
+ pi0 = to_next[0] = from[0];
+ from += 1;
+ n_left_from -= 1;
+ to_next += 1;
+ n_left_to_next -= 1;
+
+ p0 = vlib_get_buffer (vm, pi0);
+
+ //Accessing pseudo header
+ pheader0 = vlib_buffer_get_current (p0);
+ vlib_buffer_advance (p0, sizeof (*pheader0));
+
+ if (ip4_to_ip6_fragmented (p0, ip4_to_ip6_set_cb, pheader0))
+ {
+ p0->error = error_node->errors[MAP_ERROR_FRAGMENT_DROPPED];
+ next0 = IP4_MAPT_FRAGMENTED_NEXT_DROP;
+ }
+ else
+ {
+ if (vnet_buffer (p0)->map_t.mtu < p0->current_length)
+ {
+ vnet_buffer (p0)->ip_frag.header_offset = 0;
+ vnet_buffer (p0)->ip_frag.mtu = vnet_buffer (p0)->map_t.mtu;
+ vnet_buffer (p0)->ip_frag.next_index =
+ IP6_FRAG_NEXT_IP6_LOOKUP;
+ next0 = IP4_MAPT_FRAGMENTED_NEXT_IP6_FRAG;
+ }
+ }
+
+ vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
+ to_next, n_left_to_next, pi0,
+ next0);
+ }
+ vlib_put_next_frame (vm, node, next_index, n_left_to_next);
+ }
+ return frame->n_vectors;
+}
+
+static uword
+ip4_map_t_tcp_udp (vlib_main_t * vm,
+ vlib_node_runtime_t * node, vlib_frame_t * frame)
+{
+ u32 n_left_from, *from, next_index, *to_next, n_left_to_next;
+ from = vlib_frame_vector_args (frame);
+ n_left_from = frame->n_vectors;
+ next_index = node->cached_next_index;
+ vlib_node_runtime_t *error_node =
+ vlib_node_get_runtime (vm, ip4_map_t_tcp_udp_node.index);
+
+
+ while (n_left_from > 0)
+ {
+ vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next);
+
+#ifdef IP4_MAP_T_DUAL_LOOP
+ while (n_left_from >= 4 && n_left_to_next >= 2)
+ {
+ u32 pi0, pi1;
+ vlib_buffer_t *p0, *p1;
+ ip4_mapt_pseudo_header_t *pheader0, *pheader1;
+ ip4_mapt_tcp_udp_next_t next0, next1;
+
+ pi0 = to_next[0] = from[0];
+ pi1 = to_next[1] = from[1];
+ from += 2;
+ n_left_from -= 2;
+ to_next += 2;
+ n_left_to_next -= 2;
+
+ next0 = IP4_MAPT_TCP_UDP_NEXT_IP6_LOOKUP;
+ next1 = IP4_MAPT_TCP_UDP_NEXT_IP6_LOOKUP;
+ p0 = vlib_get_buffer (vm, pi0);
+ p1 = vlib_get_buffer (vm, pi1);
+
+ //Accessing pseudo header
+ pheader0 = vlib_buffer_get_current (p0);
+ pheader1 = vlib_buffer_get_current (p1);
+ vlib_buffer_advance (p0, sizeof (*pheader0));
+ vlib_buffer_advance (p1, sizeof (*pheader1));
+
+ if (ip4_to_ip6_tcp_udp (p0, ip4_to_ip6_set_cb, pheader0))
+ {
+ p0->error = error_node->errors[MAP_ERROR_UNKNOWN];
+ next0 = IP4_MAPT_TCP_UDP_NEXT_DROP;
+ }
+ else
+ {
+ if (vnet_buffer (p0)->map_t.mtu < p0->current_length)
+ {
+ //Send to fragmentation node if necessary
+ vnet_buffer (p0)->ip_frag.header_offset = 0;
+ vnet_buffer (p0)->ip_frag.mtu = vnet_buffer (p0)->map_t.mtu;
+ vnet_buffer (p0)->ip_frag.next_index =
+ IP6_FRAG_NEXT_IP6_LOOKUP;
+ next0 = IP4_MAPT_TCP_UDP_NEXT_IP6_FRAG;
+ }
+ }
+
+ if (ip4_to_ip6_tcp_udp (p1, ip4_to_ip6_set_cb, pheader1))
+ {
+ p1->error = error_node->errors[MAP_ERROR_UNKNOWN];
+ next1 = IP4_MAPT_TCP_UDP_NEXT_DROP;
+ }
+ else
+ {
+ if (vnet_buffer (p1)->map_t.mtu < p1->current_length)
+ {
+ //Send to fragmentation node if necessary
+ vnet_buffer (p1)->ip_frag.header_offset = 0;
+ vnet_buffer (p1)->ip_frag.mtu = vnet_buffer (p1)->map_t.mtu;
+ vnet_buffer (p1)->ip_frag.next_index =
+ IP6_FRAG_NEXT_IP6_LOOKUP;
+ next1 = IP4_MAPT_TCP_UDP_NEXT_IP6_FRAG;
+ }
+ }
+
+ vlib_validate_buffer_enqueue_x2 (vm, node, next_index,
+ to_next, n_left_to_next, pi0, pi1,
+ next0, next1);
+ }
+#endif
+
+ while (n_left_from > 0 && n_left_to_next > 0)
+ {
+ u32 pi0;
+ vlib_buffer_t *p0;
+ ip4_mapt_pseudo_header_t *pheader0;
+ ip4_mapt_tcp_udp_next_t next0;
+
+ pi0 = to_next[0] = from[0];
+ from += 1;
+ n_left_from -= 1;
+ to_next += 1;
+ n_left_to_next -= 1;
+
+ next0 = IP4_MAPT_TCP_UDP_NEXT_IP6_LOOKUP;
+ p0 = vlib_get_buffer (vm, pi0);
+
+ //Accessing pseudo header
+ pheader0 = vlib_buffer_get_current (p0);
+ vlib_buffer_advance (p0, sizeof (*pheader0));
+
+ if (ip4_to_ip6_tcp_udp (p0, ip4_to_ip6_set_cb, pheader0))
+ {
+ p0->error = error_node->errors[MAP_ERROR_UNKNOWN];
+ next0 = IP4_MAPT_TCP_UDP_NEXT_DROP;
+ }
+ else
+ {
+ if (vnet_buffer (p0)->map_t.mtu < p0->current_length)
+ {
+ //Send to fragmentation node if necessary
+ vnet_buffer (p0)->ip_frag.header_offset = 0;
+ vnet_buffer (p0)->ip_frag.mtu = vnet_buffer (p0)->map_t.mtu;
+ vnet_buffer (p0)->ip_frag.next_index =
+ IP6_FRAG_NEXT_IP6_LOOKUP;
+ next0 = IP4_MAPT_TCP_UDP_NEXT_IP6_FRAG;
+ }
+ }
+ vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
+ to_next, n_left_to_next, pi0,
+ next0);
+ }
+ vlib_put_next_frame (vm, node, next_index, n_left_to_next);
+ }
+
+ return frame->n_vectors;
+}
+
+static_always_inline void
+ip4_map_t_classify (vlib_buffer_t * p0, map_domain_t * d0,
+ ip4_header_t * ip40, u16 ip4_len0, i32 * dst_port0,
+ u8 * error0, ip4_mapt_next_t * next0)
+{
+ map_main_t *mm = &map_main;
+ u32 port_offset;
+
+ if (mm->is_ce)
+ port_offset = 0;
+ else
+ port_offset = 2;
+
+ if (PREDICT_FALSE (ip4_get_fragment_offset (ip40)))
+ {
+ *next0 = IP4_MAPT_NEXT_MAPT_FRAGMENTED;
+ if (d0->ea_bits_len == 0 && d0->rules)
+ {
+ *dst_port0 = 0;
+ }
+ else
+ {
+ *dst_port0 = ip4_map_fragment_get_port (ip40);
+ *error0 = (*dst_port0 == -1) ? MAP_ERROR_FRAGMENT_MEMORY : *error0;
+ }
+ }
+ else if (PREDICT_TRUE (ip40->protocol == IP_PROTOCOL_TCP))
+ {
+ vnet_buffer (p0)->map_t.checksum_offset = 36;
+ *next0 = IP4_MAPT_NEXT_MAPT_TCP_UDP;
+ *error0 = ip4_len0 < 40 ? MAP_ERROR_MALFORMED : *error0;
+ *dst_port0 =
+ (i32) * ((u16 *) u8_ptr_add (ip40, sizeof (*ip40) + port_offset));
+ }
+ else if (PREDICT_TRUE (ip40->protocol == IP_PROTOCOL_UDP))
+ {
+ vnet_buffer (p0)->map_t.checksum_offset = 26;
+ *next0 = IP4_MAPT_NEXT_MAPT_TCP_UDP;
+ *error0 = ip4_len0 < 28 ? MAP_ERROR_MALFORMED : *error0;
+ *dst_port0 =
+ (i32) * ((u16 *) u8_ptr_add (ip40, sizeof (*ip40) + port_offset));
+ }
+ else if (ip40->protocol == IP_PROTOCOL_ICMP)
+ {
+ *next0 = IP4_MAPT_NEXT_MAPT_ICMP;
+ if (d0->ea_bits_len == 0 && d0->rules)
+ *dst_port0 = 0;
+ else if (((icmp46_header_t *) u8_ptr_add (ip40, sizeof (*ip40)))->code
+ == ICMP4_echo_reply
+ || ((icmp46_header_t *)
+ u8_ptr_add (ip40,
+ sizeof (*ip40)))->code == ICMP4_echo_request)
+ *dst_port0 = (i32) * ((u16 *) u8_ptr_add (ip40, sizeof (*ip40) + 6));
+ }
+ else
+ {
+ *error0 = MAP_ERROR_BAD_PROTOCOL;
+ }
+}
+
+static uword
+ip4_map_t (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame)
+{
+ u32 n_left_from, *from, next_index, *to_next, n_left_to_next;
+ vlib_node_runtime_t *error_node =
+ vlib_node_get_runtime (vm, ip4_map_t_node.index);
+ from = vlib_frame_vector_args (frame);
+ n_left_from = frame->n_vectors;
+ next_index = node->cached_next_index;
+ map_main_t *mm = &map_main;
+ vlib_combined_counter_main_t *cm = map_main.domain_counters;
+ u32 thread_index = vlib_get_thread_index ();
+
+ while (n_left_from > 0)
+ {
+ vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next);
+
+#ifdef IP4_MAP_T_DUAL_LOOP
+ while (n_left_from >= 4 && n_left_to_next >= 2)
+ {
+ u32 pi0, pi1;
+ vlib_buffer_t *p0, *p1;
+ ip4_header_t *ip40, *ip41;
+ map_domain_t *d0, *d1;
+ ip4_mapt_next_t next0 = 0, next1 = 0;
+ u16 ip4_len0, ip4_len1;
+ u8 error0, error1;
+ i32 map_port0, map_port1;
+ ip4_mapt_pseudo_header_t *pheader0, *pheader1;
+
+ pi0 = to_next[0] = from[0];
+ pi1 = to_next[1] = from[1];
+ from += 2;
+ n_left_from -= 2;
+ to_next += 2;
+ n_left_to_next -= 2;
+ error0 = MAP_ERROR_NONE;
+ error1 = MAP_ERROR_NONE;
+
+ p0 = vlib_get_buffer (vm, pi0);
+ p1 = vlib_get_buffer (vm, pi1);
+ ip40 = vlib_buffer_get_current (p0);
+ ip41 = vlib_buffer_get_current (p1);
+ ip4_len0 = clib_host_to_net_u16 (ip40->length);
+ ip4_len1 = clib_host_to_net_u16 (ip41->length);
+
+ if (PREDICT_FALSE (p0->current_length < ip4_len0 ||
+ ip40->ip_version_and_header_length != 0x45))
+ {
+ error0 = MAP_ERROR_UNKNOWN;
+ next0 = IP4_MAPT_NEXT_DROP;
+ }
+
+ if (PREDICT_FALSE (p1->current_length < ip4_len1 ||
+ ip41->ip_version_and_header_length != 0x45))
+ {
+ error1 = MAP_ERROR_UNKNOWN;
+ next1 = IP4_MAPT_NEXT_DROP;
+ }
+
+ vnet_buffer (p0)->map_t.map_domain_index =
+ vnet_buffer (p0)->ip.adj_index[VLIB_TX];
+ d0 = ip4_map_get_domain (vnet_buffer (p0)->map_t.map_domain_index);
+ vnet_buffer (p1)->map_t.map_domain_index =
+ vnet_buffer (p1)->ip.adj_index[VLIB_TX];
+ d1 = ip4_map_get_domain (vnet_buffer (p1)->map_t.map_domain_index);
+
+ vnet_buffer (p0)->map_t.mtu = d0->mtu ? d0->mtu : ~0;
+ vnet_buffer (p1)->map_t.mtu = d1->mtu ? d1->mtu : ~0;
+
+ map_port0 = -1;
+ map_port1 = -1;
+
+ ip4_map_t_classify (p0, d0, ip40, ip4_len0, &map_port0, &error0,
+ &next0);
+ ip4_map_t_classify (p1, d1, ip41, ip4_len1, &map_port1, &error1,
+ &next1);
+
+ //Add MAP-T pseudo header in front of the packet
+ vlib_buffer_advance (p0, -sizeof (*pheader0));
+ vlib_buffer_advance (p1, -sizeof (*pheader1));
+ pheader0 = vlib_buffer_get_current (p0);
+ pheader1 = vlib_buffer_get_current (p1);
+
+ //Save addresses within the packet
+ if (mm->is_ce)
+ {
+ ip4_map_t_embedded_address (d0, &pheader0->daddr,
+ &ip40->dst_address);
+ ip4_map_t_embedded_address (d1, &pheader1->daddr,
+ &ip41->dst_address);
+ pheader0->saddr.as_u64[0] =
+ map_get_pfx_net (d0, ip40->src_address.as_u32,
+ (u16) map_port0);
+ pheader0->saddr.as_u64[1] =
+ map_get_sfx_net (d0, ip40->src_address.as_u32,
+ (u16) map_port0);
+ pheader1->saddr.as_u64[0] =
+ map_get_pfx_net (d1, ip41->src_address.as_u32,
+ (u16) map_port1);
+ pheader1->saddr.as_u64[1] =
+ map_get_sfx_net (d1, ip41->src_address.as_u32,
+ (u16) map_port1);
+ }
+ else
+ {
+ ip4_map_t_embedded_address (d0, &pheader0->saddr,
+ &ip40->src_address);
+ ip4_map_t_embedded_address (d1, &pheader1->saddr,
+ &ip41->src_address);
+ pheader0->daddr.as_u64[0] =
+ map_get_pfx_net (d0, ip40->dst_address.as_u32,
+ (u16) map_port0);
+ pheader0->daddr.as_u64[1] =
+ map_get_sfx_net (d0, ip40->dst_address.as_u32,
+ (u16) map_port0);
+ pheader1->daddr.as_u64[0] =
+ map_get_pfx_net (d1, ip41->dst_address.as_u32,
+ (u16) map_port1);
+ pheader1->daddr.as_u64[1] =
+ map_get_sfx_net (d1, ip41->dst_address.as_u32,
+ (u16) map_port1);
+ }
+
+ if (PREDICT_FALSE
+ (ip4_is_first_fragment (ip40) && (map_port0 != -1)
+ && (d0->ea_bits_len != 0 || !d0->rules)
+ && ip4_map_fragment_cache (ip40, map_port0)))
+ {
+ error0 = MAP_ERROR_FRAGMENT_MEMORY;
+ }
+
+ if (PREDICT_FALSE
+ (ip4_is_first_fragment (ip41) && (map_port1 != -1)
+ && (d1->ea_bits_len != 0 || !d1->rules)
+ && ip4_map_fragment_cache (ip41, map_port1)))
+ {
+ error1 = MAP_ERROR_FRAGMENT_MEMORY;
+ }
+
+ if (PREDICT_TRUE
+ (error0 == MAP_ERROR_NONE && next0 != IP4_MAPT_NEXT_MAPT_ICMP))
+ {
+ vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_TX,
+ thread_index,
+ vnet_buffer (p0)->
+ map_t.map_domain_index, 1,
+ clib_net_to_host_u16
+ (ip40->length));
+ }
+
+ if (PREDICT_TRUE
+ (error1 == MAP_ERROR_NONE && next1 != IP4_MAPT_NEXT_MAPT_ICMP))
+ {
+ vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_TX,
+ thread_index,
+ vnet_buffer (p1)->
+ map_t.map_domain_index, 1,
+ clib_net_to_host_u16
+ (ip41->length));
+ }
+
+ next0 = (error0 != MAP_ERROR_NONE) ? IP4_MAPT_NEXT_DROP : next0;
+ next1 = (error1 != MAP_ERROR_NONE) ? IP4_MAPT_NEXT_DROP : next1;
+ p0->error = error_node->errors[error0];
+ p1->error = error_node->errors[error1];
+ vlib_validate_buffer_enqueue_x2 (vm, node, next_index, to_next,
+ n_left_to_next, pi0, pi1, next0,
+ next1);
+ }
+#endif
+
+ while (n_left_from > 0 && n_left_to_next > 0)
+ {
+ u32 pi0;
+ vlib_buffer_t *p0;
+ ip4_header_t *ip40;
+ map_domain_t *d0;
+ ip4_mapt_next_t next0;
+ u16 ip4_len0;
+ u8 error0;
+ i32 map_port0;
+ ip4_mapt_pseudo_header_t *pheader0;
+
+ pi0 = to_next[0] = from[0];
+ from += 1;
+ n_left_from -= 1;
+ to_next += 1;
+ n_left_to_next -= 1;
+ error0 = MAP_ERROR_NONE;
+
+ p0 = vlib_get_buffer (vm, pi0);
+ ip40 = vlib_buffer_get_current (p0);
+ ip4_len0 = clib_host_to_net_u16 (ip40->length);
+ if (PREDICT_FALSE (p0->current_length < ip4_len0 ||
+ ip40->ip_version_and_header_length != 0x45))
+ {
+ error0 = MAP_ERROR_UNKNOWN;
+ next0 = IP4_MAPT_NEXT_DROP;
+ }
+
+ vnet_buffer (p0)->map_t.map_domain_index =
+ vnet_buffer (p0)->ip.adj_index[VLIB_TX];
+ d0 = ip4_map_get_domain (vnet_buffer (p0)->map_t.map_domain_index);
+
+ vnet_buffer (p0)->map_t.mtu = d0->mtu ? d0->mtu : ~0;
+
+ map_port0 = -1;
+ ip4_map_t_classify (p0, d0, ip40, ip4_len0, &map_port0, &error0,
+ &next0);
+
+ //Add MAP-T pseudo header in front of the packet
+ vlib_buffer_advance (p0, -sizeof (*pheader0));
+ pheader0 = vlib_buffer_get_current (p0);
+
+ //Save addresses within the packet
+ if (mm->is_ce)
+ {
+ ip4_map_t_embedded_address (d0, &pheader0->daddr,
+ &ip40->dst_address);
+ pheader0->saddr.as_u64[0] =
+ map_get_pfx_net (d0, ip40->src_address.as_u32,
+ (u16) map_port0);
+ pheader0->saddr.as_u64[1] =
+ map_get_sfx_net (d0, ip40->src_address.as_u32,
+ (u16) map_port0);
+ }
+ else
+ {
+ ip4_map_t_embedded_address (d0, &pheader0->saddr,
+ &ip40->src_address);
+ pheader0->daddr.as_u64[0] =
+ map_get_pfx_net (d0, ip40->dst_address.as_u32,
+ (u16) map_port0);
+ pheader0->daddr.as_u64[1] =
+ map_get_sfx_net (d0, ip40->dst_address.as_u32,
+ (u16) map_port0);
+ }
+
+ //It is important to cache at this stage because the result might be necessary
+ //for packets within the same vector.
+ //Actually, this approach even provides some limited out-of-order fragments support
+ if (PREDICT_FALSE
+ (ip4_is_first_fragment (ip40) && (map_port0 != -1)
+ && (d0->ea_bits_len != 0 || !d0->rules)
+ && ip4_map_fragment_cache (ip40, map_port0)))
+ {
+ error0 = MAP_ERROR_UNKNOWN;
+ }
+
+ if (PREDICT_TRUE
+ (error0 == MAP_ERROR_NONE && next0 != IP4_MAPT_NEXT_MAPT_ICMP))
+ {
+ vlib_increment_combined_counter (cm + MAP_DOMAIN_COUNTER_TX,
+ thread_index,
+ vnet_buffer (p0)->
+ map_t.map_domain_index, 1,
+ clib_net_to_host_u16
+ (ip40->length));
+ }
+
+ next0 = (error0 != MAP_ERROR_NONE) ? IP4_MAPT_NEXT_DROP : next0;
+ p0->error = error_node->errors[error0];
+ vlib_validate_buffer_enqueue_x1 (vm, node, next_index,
+ to_next, n_left_to_next, pi0,
+ next0);
+ }
+ vlib_put_next_frame (vm, node, next_index, n_left_to_next);
+ }
+ return frame->n_vectors;
+}
+
+static char *map_t_error_strings[] = {
+#define _(sym,string) string,
+ foreach_map_error
+#undef _
+};
+
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE(ip4_map_t_fragmented_node) = {
+ .function = ip4_map_t_fragmented,
+ .name = "ip4-map-t-fragmented",
+ .vector_size = sizeof(u32),
+ .format_trace = format_map_trace,
+ .type = VLIB_NODE_TYPE_INTERNAL,
+
+ .n_errors = MAP_N_ERROR,
+ .error_strings = map_t_error_strings,
+
+ .n_next_nodes = IP4_MAPT_FRAGMENTED_N_NEXT,
+ .next_nodes = {
+ [IP4_MAPT_FRAGMENTED_NEXT_IP6_LOOKUP] = "ip6-lookup",
+ [IP4_MAPT_FRAGMENTED_NEXT_IP6_FRAG] = IP6_FRAG_NODE_NAME,
+ [IP4_MAPT_FRAGMENTED_NEXT_DROP] = "error-drop",
+ },
+};
+/* *INDENT-ON* */
+
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE(ip4_map_t_icmp_node) = {
+ .function = ip4_map_t_icmp,
+ .name = "ip4-map-t-icmp",
+ .vector_size = sizeof(u32),
+ .format_trace = format_map_trace,
+ .type = VLIB_NODE_TYPE_INTERNAL,
+
+ .n_errors = MAP_N_ERROR,
+ .error_strings = map_t_error_strings,
+
+ .n_next_nodes = IP4_MAPT_ICMP_N_NEXT,
+ .next_nodes = {
+ [IP4_MAPT_ICMP_NEXT_IP6_LOOKUP] = "ip6-lookup",
+ [IP4_MAPT_ICMP_NEXT_IP6_FRAG] = IP6_FRAG_NODE_NAME,
+ [IP4_MAPT_ICMP_NEXT_DROP] = "error-drop",
+ },
+};
+/* *INDENT-ON* */
+
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE(ip4_map_t_tcp_udp_node) = {
+ .function = ip4_map_t_tcp_udp,
+ .name = "ip4-map-t-tcp-udp",
+ .vector_size = sizeof(u32),
+ .format_trace = format_map_trace,
+ .type = VLIB_NODE_TYPE_INTERNAL,
+
+ .n_errors = MAP_N_ERROR,
+ .error_strings = map_t_error_strings,
+
+ .n_next_nodes = IP4_MAPT_TCP_UDP_N_NEXT,
+ .next_nodes = {
+ [IP4_MAPT_TCP_UDP_NEXT_IP6_LOOKUP] = "ip6-lookup",
+ [IP4_MAPT_TCP_UDP_NEXT_IP6_FRAG] = IP6_FRAG_NODE_NAME,
+ [IP4_MAPT_TCP_UDP_NEXT_DROP] = "error-drop",
+ },
+};
+/* *INDENT-ON* */
+
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE(ip4_map_t_node) = {
+ .function = ip4_map_t,
+ .name = "ip4-map-t",
+ .vector_size = sizeof(u32),
+ .format_trace = format_map_trace,
+ .type = VLIB_NODE_TYPE_INTERNAL,
+
+ .n_errors = MAP_N_ERROR,
+ .error_strings = map_t_error_strings,
+
+ .n_next_nodes = IP4_MAPT_N_NEXT,
+ .next_nodes = {
+ [IP4_MAPT_NEXT_MAPT_TCP_UDP] = "ip4-map-t-tcp-udp",
+ [IP4_MAPT_NEXT_MAPT_ICMP] = "ip4-map-t-icmp",
+ [IP4_MAPT_NEXT_MAPT_FRAGMENTED] = "ip4-map-t-fragmented",
+ [IP4_MAPT_NEXT_DROP] = "error-drop",
+ },
+};
+/* *INDENT-ON* */
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */