/* * 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. */ /* * ip/ip4_forward.c: IP v4 forwarding * * Copyright (c) 2008 Eliot Dresselhaus * * Permission is hereby granted, free of charge, to any person obtaining * a copy of this software and associated documentation files (the * "Software"), to deal in the Software without restriction, including * without limitation the rights to use, copy, modify, merge, publish, * distribute, sublicense, and/or sell copies of the Software, and to * permit persons to whom the Software is furnished to do so, subject to * the following conditions: * * The above copyright notice and this permission notice shall be * included in all copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. */ #include #include #include #include /* for ethernet_header_t */ #include /* for ethernet_arp_header_t */ #include #include /* for srp_hw_interface_class */ #include /* for API error numbers */ #include /* for FIB table and entry creation */ #include /* for FIB table and entry creation */ #include /* for FIB uRPF check */ #include #include #include #include #include /* for mFIB table and entry creation */ #include #include /** @brief IPv4 lookup node. @node ip4-lookup This is the main IPv4 lookup dispatch node. @param vm vlib_main_t corresponding to the current thread @param node vlib_node_runtime_t @param frame vlib_frame_t whose contents should be dispatched @par Graph mechanics: buffer metadata, next index usage @em Uses: - vnet_buffer(b)->sw_if_index[VLIB_RX] - Indicates the @c sw_if_index value of the interface that the packet was received on. - vnet_buffer(b)->sw_if_index[VLIB_TX] - When the value is @c ~0 then the node performs a longest prefix match (LPM) for the packet destination address in the FIB attached to the receive interface. - Otherwise perform LPM for the packet destination address in the indicated FIB. In this case [VLIB_TX] is a FIB index value (0, 1, ...) and not a VRF id. @em Sets: - vnet_buffer(b)->ip.adj_index[VLIB_TX] - The lookup result adjacency index. Next Index: - Dispatches the packet to the node index found in ip_adjacency_t @c adj->lookup_next_index (where @c adj is the lookup result adjacency). */ VLIB_NODE_FN (ip4_lookup_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { return ip4_lookup_inline (vm, node, frame, /* lookup_for_responses_to_locally_received_packets */ 0); } static u8 *format_ip4_lookup_trace (u8 * s, va_list * args); /* *INDENT-OFF* */ VLIB_REGISTER_NODE (ip4_lookup_node) = { .name = "ip4-lookup", .vector_size = sizeof (u32), .format_trace = format_ip4_lookup_trace, .n_next_nodes = IP_LOOKUP_N_NEXT, .next_nodes = IP4_LOOKUP_NEXT_NODES, }; /* *INDENT-ON* */ VLIB_NODE_FN (ip4_load_balance_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { vlib_combined_counter_main_t *cm = &load_balance_main.lbm_via_counters; u32 n_left, *from; u32 thread_index = vm->thread_index; vlib_buffer_t *bufs[VLIB_FRAME_SIZE], **b = bufs; u16 nexts[VLIB_FRAME_SIZE], *next; from = vlib_frame_vector_args (frame); n_left = frame->n_vectors; next = nexts; vlib_get_buffers (vm, from, bufs, n_left); while (n_left >= 4) { const load_balance_t *lb0, *lb1; const ip4_header_t *ip0, *ip1; u32 lbi0, hc0, lbi1, hc1; const dpo_id_t *dpo0, *dpo1; /* Prefetch next iteration. */ { vlib_prefetch_buffer_header (b[2], LOAD); vlib_prefetch_buffer_header (b[3], LOAD); CLIB_PREFETCH (b[2]->data, sizeof (ip0[0]), LOAD); CLIB_PREFETCH (b[3]->data, sizeof (ip0[0]), LOAD); } ip0 = vlib_buffer_get_current (b[0]); ip1 = vlib_buffer_get_current (b[1]); lbi0 = vnet_buffer (b[0])->ip.adj_index[VLIB_TX]; lbi1 = vnet_buffer (b[1])->ip.adj_index[VLIB_TX]; lb0 = load_balance_get (lbi0); lb1 = load_balance_get (lbi1); /* * this node is for via FIBs we can re-use the hash value from the * to node if present. * We don't want to use the same hash value at each level in the recursion * graph as that would lead to polarisation */ hc0 = hc1 = 0; if (PREDICT_FALSE (lb0->lb_n_buckets > 1)) { if (PREDICT_TRUE (vnet_buffer (b[0])->ip.flow_hash)) { hc0 = vnet_buffer (b[0])->ip.flow_hash = vnet_buffer (b[0])->ip.flow_hash >> 1; } else { hc0 = vnet_buffer (b[0])->ip.flow_hash = ip4_compute_flow_hash (ip0, lb0->lb_hash_config); } dpo0 = load_balance_get_fwd_bucket (lb0, (hc0 & (lb0->lb_n_buckets_minus_1))); } else { dpo0 = load_balance_get_bucket_i (lb0, 0); } if (PREDICT_FALSE (lb1->lb_n_buckets > 1)) { if (PREDICT_TRUE (vnet_buffer (b[1])->ip.flow_hash)) { hc1 = vnet_buffer (b[1])->ip.flow_hash = vnet_buffer (b[1])->ip.flow_hash >> 1; } else { hc1 = vnet_buffer (b[1])->ip.flow_hash = ip4_compute_flow_hash (ip1, lb1->lb_hash_config); } dpo1 = load_balance_get_fwd_bucket (lb1, (hc1 & (lb1->lb_n_buckets_minus_1))); } else { dpo1 = load_balance_get_bucket_i (lb1, 0); } next[0] = dpo0->dpoi_next_node; next[1] = dpo1->dpoi_next_node; vnet_buffer (b[0])->ip.adj_index[VLIB_TX] = dpo0->dpoi_index; vnet_buffer (b[1])->ip.adj_index[VLIB_TX] = dpo1->dpoi_index; vlib_increment_combined_counter (cm, thread_index, lbi0, 1, vlib_buffer_length_in_chain (vm, b[0])); vlib_increment_combined_counter (cm, thread_index, lbi1, 1, vlib_buffer_length_in_chain (vm, b[1])); b += 2; next += 2; n_left -= 2; } while (n_left > 0) { const load_balance_t *lb0; const ip4_header_t *ip0; const dpo_id_t *dpo0; u32 lbi0, hc0; ip0 = vlib_buffer_get_current (b[0]); lbi0 = vnet_buffer (b[0])->ip.adj_index[VLIB_TX]; lb0 = load_balance_get (lbi0); hc0 = 0; if (PREDICT_FALSE (lb0->lb_n_buckets > 1)) { if (PREDICT_TRUE (vnet_buffer (b[0])->ip.flow_hash)) { hc0 = vnet_buffer (b[0])->ip.flow_hash = vnet_buffer (b[0])->ip.flow_hash >> 1; } else { hc0 = vnet_buffer (b[0])->ip.flow_hash = ip4_compute_flow_hash (ip0, lb0->lb_hash_config); } dpo0 = load_balance_get_fwd_bucket (lb0, (hc0 & (lb0->lb_n_buckets_minus_1))); } else { dpo0 = load_balance_get_bucket_i (lb0, 0); } next[0] = dpo0->dpoi_next_node; vnet_buffer (b[0])->ip.adj_index[VLIB_TX] = dpo0->dpoi_index; vlib_increment_combined_counter (cm, thread_index, lbi0, 1, vlib_buffer_length_in_chain (vm, b[0])); b += 1; next += 1; n_left -= 1; } vlib_buffer_enqueue_to_next (vm, node, from, nexts, frame->n_vectors); if (node->flags & VLIB_NODE_FLAG_TRACE) ip4_forward_next_trace (vm, node, frame, VLIB_TX); return frame->n_vectors; } /* *INDENT-OFF* */ VLIB_REGISTER_NODE (ip4_load_balance_node) = { .name = "ip4-load-balance", .vector_size = sizeof (u32), .sibling_of = "ip4-lookup", .format_trace = format_ip4_lookup_trace, }; /* *INDENT-ON* */ #ifndef CLIB_MARCH_VARIANT /* get first interface address */ ip4_address_t * ip4_interface_first_address (ip4_main_t * im, u32 sw_if_index, ip_interface_address_t ** result_ia) { ip_lookup_main_t *lm = &im->lookup_main; ip_interface_address_t *ia = 0; ip4_address_t *result = 0; /* *INDENT-OFF* */ foreach_ip_interface_address (lm, ia, sw_if_index, 1 /* honor unnumbered */ , ({ ip4_address_t * a = ip_interface_address_get_address (lm, ia); result = a; break; })); /* *INDENT-OFF* */ if (result_ia) *result_ia = result ? ia : 0; return result; } static void ip4_add_subnet_bcast_route (u32 fib_index, fib_prefix_t *pfx, u32 sw_if_index) { vnet_sw_interface_flags_t iflags; iflags = vnet_sw_interface_get_flags(vnet_get_main(), sw_if_index); fib_table_entry_special_remove(fib_index, pfx, FIB_SOURCE_INTERFACE); if (iflags & VNET_SW_INTERFACE_FLAG_DIRECTED_BCAST) { fib_table_entry_update_one_path (fib_index, pfx, FIB_SOURCE_INTERFACE, FIB_ENTRY_FLAG_NONE, DPO_PROTO_IP4, /* No next-hop address */ &ADJ_BCAST_ADDR, sw_if_index, // invalid FIB index ~0, 1, // no out-label stack NULL, FIB_ROUTE_PATH_FLAG_NONE); } else { fib_table_entry_special_add(fib_index, pfx, FIB_SOURCE_INTERFACE, (FIB_ENTRY_FLAG_DROP | FIB_ENTRY_FLAG_LOOSE_URPF_EXEMPT)); } } static void ip4_add_interface_routes (u32 sw_if_index, ip4_main_t * im, u32 fib_index, ip_interface_address_t * a) { ip_lookup_main_t *lm = &im->lookup_main; ip4_address_t *address = ip_interface_address_get_address (lm, a); fib_prefix_t pfx = { .fp_len = a->address_length, .fp_proto = FIB_PROTOCOL_IP4, .fp_addr.ip4 = *address, }; if (pfx.fp_len <= 30) { /* a /30 or shorter - add a glean for the network address */ fib_table_entry_update_one_path (fib_index, &pfx, FIB_SOURCE_INTERFACE, (FIB_ENTRY_FLAG_CONNECTED | FIB_ENTRY_FLAG_ATTACHED), DPO_PROTO_IP4, /* No next-hop address */ NULL, sw_if_index, // invalid FIB index ~0, 1, // no out-label stack NULL, FIB_ROUTE_PATH_FLAG_NONE); /* Add the two broadcast addresses as drop */ fib_prefix_t net_pfx = { .fp_len = 32, .fp_proto = FIB_PROTOCOL_IP4, .fp_addr.ip4.as_u32 = address->as_u32 & im->fib_masks[pfx.fp_len], }; if (net_pfx.fp_addr.ip4.as_u32 != pfx.fp_addr.ip4.as_u32) fib_table_entry_special_add(fib_index, &net_pfx, FIB_SOURCE_INTERFACE, (FIB_ENTRY_FLAG_DROP | FIB_ENTRY_FLAG_LOOSE_URPF_EXEMPT)); net_pfx.fp_addr.ip4.as_u32 |= ~im->fib_masks[pfx.fp_len]; if (net_pfx.fp_addr.ip4.as_u32 != pfx.fp_addr.ip4.as_u32) ip4_add_subnet_bcast_route(fib_index, &net_pfx, sw_if_index); } else if (pfx.fp_len == 31) { u32 mask = clib_host_to_net_u32(1); fib_prefix_t net_pfx = pfx; net_pfx.fp_len = 32; net_pfx.fp_addr.ip4.as_u32 ^= mask; /* a /31 - add the other end as an attached host */ fib_table_entry_update_one_path (fib_index, &net_pfx, FIB_SOURCE_INTERFACE, (FIB_ENTRY_FLAG_ATTACHED), DPO_PROTO_IP4, &net_pfx.fp_addr, sw_if_index, // invalid FIB index ~0, 1, NULL, FIB_ROUTE_PATH_FLAG_NONE); } pfx.fp_len = 32; if (sw_if_index < vec_len (lm->classify_table_index_by_sw_if_index)) { u32 classify_table_index = lm->classify_table_index_by_sw_if_index[sw_if_index]; if (classify_table_index != (u32) ~ 0) { dpo_id_t dpo = DPO_INVALID; dpo_set (&dpo, DPO_CLASSIFY, DPO_PROTO_IP4, classify_dpo_create (DPO_PROTO_IP4, classify_table_index)); fib_table_entry_special_dpo_add (fib_index, &pfx, FIB_SOURCE_CLASSIFY, FIB_ENTRY_FLAG_NONE, &dpo); dpo_reset (&dpo); } } fib_table_entry_update_one_path (fib_index, &pfx, FIB_SOURCE_INTERFACE, (FIB_ENTRY_FLAG_CONNECTED | FIB_ENTRY_FLAG_LOCAL), DPO_PROTO_IP4, &pfx.fp_addr, sw_if_index, // invalid FIB index ~0, 1, NULL, FIB_ROUTE_PATH_FLAG_NONE); } static void ip4_del_interface_routes (ip4_main_t * im, u32 fib_index, ip4_address_t * address, u32 address_length) { fib_prefix_t pfx = { .fp_len = address_length, .fp_proto = FIB_PROTOCOL_IP4, .fp_addr.ip4 = *address, }; if (pfx.fp_len <= 30) { fib_prefix_t net_pfx = { .fp_len = 32, .fp_proto = FIB_PROTOCOL_IP4, .fp_addr.ip4.as_u32 = address->as_u32 & im->fib_masks[pfx.fp_len], }; if (net_pfx.fp_addr.ip4.as_u32 != pfx.fp_addr.ip4.as_u32) fib_table_entry_special_remove(fib_index, &net_pfx, FIB_SOURCE_INTERFACE); net_pfx.fp_addr.ip4.as_u32 |= ~im->fib_masks[pfx.fp_len]; if (net_pfx.fp_addr.ip4.as_u32 != pfx.fp_addr.ip4.as_u32) fib_table_entry_special_remove(fib_index, &net_pfx, FIB_SOURCE_INTERFACE); fib_table_entry_delete (fib_index, &pfx, FIB_SOURCE_INTERFACE); } else if (pfx.fp_len == 31) { u32 mask = clib_host_to_net_u32(1); fib_prefix_t net_pfx = pfx; net_pfx.fp_len = 32; net_pfx.fp_addr.ip4.as_u32 ^= mask; fib_table_entry_delete (fib_index, &net_pfx, FIB_SOURCE_INTERFACE); } pfx.fp_len = 32; fib_table_entry_delete (fib_index, &pfx, FIB_SOURCE_INTERFACE); } void ip4_sw_interface_enable_disable (u32 sw_if_index, u32 is_enable) { ip4_main_t *im = &ip4_main; vec_validate_init_empty (im->ip_enabled_by_sw_if_index, sw_if_index, 0); /* * enable/disable only on the 1<->0 transition */ if (is_enable) { if (1 != ++im->ip_enabled_by_sw_if_index[sw_if_index]) return; } else { ASSERT (im->ip_enabled_by_sw_if_index[sw_if_index] > 0); if (0 != --im->ip_enabled_by_sw_if_index[sw_if_index]) return; } vnet_feature_enable_disable ("ip4-unicast", "ip4-not-enabled", sw_if_index, !is_enable, 0, 0); vnet_feature_enable_disable ("ip4-multicast", "ip4-not-enabled", sw_if_index, !is_enable, 0, 0); { ip4_enable_disable_interface_callback_t *cb; vec_foreach (cb, im->enable_disable_interface_callbacks) cb->function (im, cb->function_opaque, sw_if_index, is_enable); } } static clib_error_t * ip4_add_del_interface_address_internal (vlib_main_t * vm, u32 sw_if_index, ip4_address_t * address, u32 address_length, u32 is_del) { vnet_main_t *vnm = vnet_get_main (); ip4_main_t *im = &ip4_main; ip_lookup_main_t *lm = &im->lookup_main; clib_error_t *error = 0; u32 if_address_index, elts_before; ip4_address_fib_t ip4_af, *addr_fib = 0; /* local0 interface doesn't support IP addressing */ if (sw_if_index == 0) { return clib_error_create ("local0 interface doesn't support IP addressing"); } vec_validate (im->fib_index_by_sw_if_index, sw_if_index); ip4_addr_fib_init (&ip4_af, address, vec_elt (im->fib_index_by_sw_if_index, sw_if_index)); vec_add1 (addr_fib, ip4_af); /* * there is no support for adj-fib handling in the presence of overlapping * subnets on interfaces. Easy fix - disallow overlapping subnets, like * most routers do. */ /* *INDENT-OFF* */ if (!is_del) { /* When adding an address check that it does not conflict with an existing address on any interface in this table. */ ip_interface_address_t *ia; vnet_sw_interface_t *sif; pool_foreach(sif, vnm->interface_main.sw_interfaces, ({ if (im->fib_index_by_sw_if_index[sw_if_index] == im->fib_index_by_sw_if_index[sif->sw_if_index]) { foreach_ip_interface_address (&im->lookup_main, ia, sif->sw_if_index, 0 /* honor unnumbered */ , ({ ip4_address_t * x = ip_interface_address_get_address (&im->lookup_main, ia); if (ip4_destination_matches_route (im, address, x, ia->address_length) || ip4_destination_matches_route (im, x, address, address_length)) { vnm->api_errno = VNET_API_ERROR_DUPLICATE_IF_ADDRESS; return clib_error_create ("failed to add %U which conflicts with %U for interface %U", format_ip4_address_and_length, address, address_length, format_ip4_address_and_length, x, ia->address_length, format_vnet_sw_if_index_name, vnm, sif->sw_if_index); } })); } })); } /* *INDENT-ON* */ elts_before = pool_elts (lm->if_address_pool); error = ip_interface_address_add_del (lm, sw_if_index, addr_fib, address_length, is_del, &if_address_index); if (error) goto done; ip4_sw_interface_enable_disable (sw_if_index, !is_del); if (is_del) ip4_del_interface_routes (im, ip4_af.fib_index, address, address_length); else ip4_add_interface_routes (sw_if_index, im, ip4_af.fib_index, pool_elt_at_index (lm->if_address_pool, if_address_index)); /* If pool did not grow/shrink: add duplicate address. */ if (elts_before != pool_elts (lm->if_address_pool)) { ip4_add_del_interface_address_callback_t *cb; vec_foreach (cb, im->add_del_interface_address_callbacks) cb->function (im, cb->function_opaque, sw_if_index, address, address_length, if_address_index, is_del); } done: vec_free (addr_fib); return error; } clib_error_t * ip4_add_del_interface_address (vlib_main_t * vm, u32 sw_if_index, ip4_address_t * address, u32 address_length, u32 is_del) { return ip4_add_del_interface_address_internal (vm, sw_if_index, address, address_length, is_del); } void ip4_directed_broadcast (u32 sw_if_index, u8 enable) { ip_interface_address_t *ia; ip4_main_t *im; im = &ip4_main; /* * when directed broadcast is enabled, the subnet braodcast route will forward * packets using an adjacency with a broadcast MAC. otherwise it drops */ /* *INDENT-OFF* */ foreach_ip_interface_address(&im->lookup_main, ia, sw_if_index, 0, ({ if (ia->address_length <= 30) { ip4_address_t *ipa; ipa = ip_interface_address_get_address (&im->lookup_main, ia); fib_prefix_t pfx = { .fp_len = 32, .fp_proto = FIB_PROTOCOL_IP4, .fp_addr = { .ip4.as_u32 = (ipa->as_u32 | ~im->fib_masks[ia->address_length]), }, }; ip4_add_subnet_bcast_route (fib_table_get_index_for_sw_if_index(FIB_PROTOCOL_IP4, sw_if_index), &pfx, sw_if_index); } })); /* *INDENT-ON* */ } #endif /* Built-in ip4 unicast rx feature path definition */ /* *INDENT-OFF* */ VNET_FEATURE_ARC_INIT (ip4_unicast, static) = { .arc_name = "ip4-unicast", .start_nodes = VNET_FEATURES ("ip4-input", "ip4-input-no-checksum"), .last_in_arc = "ip4-lookup", .arc_index_ptr = &ip4_main.lookup_main.ucast_feature_arc_index, }; VNET_FEATURE_INIT (ip4_flow_classify, static) = { .arc_name = "ip4-unicast", .node_name = "ip4-flow-classify", .runs_before = VNET_FEATURES ("ip4-inacl"), }; VNET_FEATURE_INIT (ip4_inacl, static) = { .arc_name = "ip4-unicast", .node_name = "ip4-inacl", .runs_before = VNET_FEATURES ("ip4-source-check-via-rx"), }; VNET_FEATURE_INIT (ip4_source_check_1, static) = { .arc_name = "ip4-unicast", .node_name = "ip4-source-check-via-rx", .runs_before = VNET_FEATURES ("ip4-source-check-via-any"), }; VNET_FEATURE_INIT (ip4_source_check_2, static) = { .arc_name = "ip4-unicast", .node_name = "ip4-source-check-via-any", .runs_before = VNET_FEATURES ("ip4-policer-classify"), }; VNET_FEATURE_INIT (ip4_source_and_port_range_check_rx, static) = { .arc_name = "ip4-unicast", .node_name = "ip4-source-and-port-range-check-rx", .runs_before = VNET_FEATURES ("ip4-policer-classify"), }; VNET_FEATURE_INIT (ip4_policer_classify, static) = { .arc_name = "ip4-unicast", .node_name = "ip4-policer-classify", .runs_before = VNET_FEATURES ("ipsec4-input-feature"), }; VNET_FEATURE_INIT (ip4_ipsec, static) = { .arc_name = "ip4-unicast", .node_name = "ipsec4-input-feature", .runs_before = VNET_FEATURES ("vpath-input-ip4"), }; VNET_FEATURE_INIT (ip4_vpath, static) = { .arc_name = "ip4-unicast", .node_name = "vpath-input-ip4", .runs_before = VNET_FEATURES ("ip4-vxlan-bypass"), }; VNET_FEATURE_INIT (ip4_vxlan_bypass, static) = { .arc_name = "ip4-unicast", .node_name = "ip4-vxlan-bypass", .runs_before = VNET_FEATURES ("ip4-lookup"), }; VNET_FEATURE_INIT (ip4_not_enabled, static) = { .arc_name = "ip4-unicast", .node_name = "ip4-not-enabled", .runs_before = VNET_FEATURES ("ip4-lookup"), }; VNET_FEATURE_INIT (ip4_lookup, static) = { .arc_name = "ip4-unicast", .node_name = "ip4-lookup", .runs_before = 0, /* not before any other features */ }; /* Built-in ip4 multicast rx feature path definition */ VNET_FEATURE_ARC_INIT (ip4_multicast, static) = { .arc_name = "ip4-multicast", .start_nodes = VNET_FEATURES ("ip4-input", "ip4-input-no-checksum"), .last_in_arc = "ip4-mfib-forward-lookup", .arc_index_ptr = &ip4_main.lookup_main.mcast_feature_arc_index, }; VNET_FEATURE_INIT (ip4_vpath_mc, static) = { .arc_name = "ip4-multicast", .node_name = "vpath-input-ip4", .runs_before = VNET_FEATURES ("ip4-mfib-forward-lookup"), }; VNET_FEATURE_INIT (ip4_mc_not_enabled, static) = { .arc_name = "ip4-multicast", .node_name = "ip4-not-enabled", .runs_before = VNET_FEATURES ("ip4-mfib-forward-lookup"), }; VNET_FEATURE_INIT (ip4_lookup_mc, static) = { .arc_name = "ip4-multicast", .node_name = "ip4-mfib-forward-lookup", .runs_before = 0, /* last feature */ }; /* Source and port-range check ip4 tx feature path definition */ VNET_FEATURE_ARC_INIT (ip4_output, static) = { .arc_name = "ip4-output", .start_nodes = VNET_FEATURES ("ip4-rewrite", "ip4-midchain", "ip4-dvr-dpo"), .last_in_arc = "interface-output", .arc_index_ptr = &ip4_main.lookup_main.output_feature_arc_index, }; VNET_FEATURE_INIT (ip4_source_and_port_range_check_tx, static) = { .arc_name = "ip4-output", .node_name = "ip4-source-and-port-range-check-tx", .runs_before = VNET_FEATURES ("ip4-outacl"), }; VNET_FEATURE_INIT (ip4_outacl, static) = { .arc_name = "ip4-output", .node_name = "ip4-outacl", .runs_before = VNET_FEATURES ("ipsec4-output-feature"), }; VNET_FEATURE_INIT (ip4_ipsec_output, static) = { .arc_name = "ip4-output", .node_name = "ipsec4-output-feature", .runs_before = VNET_FEATURES ("interface-output"), }; /* Built-in ip4 tx feature path definition */ VNET_FEATURE_INIT (ip4_interface_output, static) = { .arc_name = "ip4-output", .node_name = "interface-output", .runs_before = 0, /* not before any other features */ }; /* *INDENT-ON* */ static clib_error_t * ip4_sw_interface_add_del (vnet_main_t * vnm, u32 sw_if_index, u32 is_add) { ip4_main_t *im = &ip4_main; /* Fill in lookup tables with default table (0). */ vec_validate (im->fib_index_by_sw_if_index, sw_if_index); vec_validate (im->mfib_index_by_sw_if_index, sw_if_index); if (!is_add) { ip4_main_t *im4 = &ip4_main; ip_lookup_main_t *lm4 = &im4->lookup_main; ip_interface_address_t *ia = 0; ip4_address_t *address; vlib_main_t *vm = vlib_get_main (); vnet_sw_interface_update_unnumbered (sw_if_index, ~0, 0); /* *INDENT-OFF* */ foreach_ip_interface_address (lm4, ia, sw_if_index, 0, ({ address = ip_interface_address_get_address (lm4, ia); ip4_add_del_interface_address(vm, sw_if_index, address, ia->address_length, 1); })); /* *INDENT-ON* */ } vnet_feature_enable_disable ("ip4-unicast", "ip4-not-enabled", sw_if_index, is_add, 0, 0); vnet_feature_enable_disable ("ip4-multicast", "ip4-not-enabled", sw_if_index, is_add, 0, 0); return /* no error */ 0; } VNET_SW_INTERFACE_ADD_DEL_FUNCTION (ip4_sw_interface_add_del); /* Global IP4 main. */ #ifndef CLIB_MARCH_VARIANT ip4_main_t ip4_main; #endif /* CLIB_MARCH_VARIANT */ static clib_error_t * ip4_lookup_init (vlib_main_t * vm) { ip4_main_t *im = &ip4_main; clib_error_t *error; uword i; if ((error = vlib_call_init_function (vm, vnet_feature_init))) return error; if ((error = vlib_call_init_function (vm, ip4_mtrie_module_init))) return (error); if ((error = vlib_call_init_function (vm, fib_module_init))) return error; if ((error = vlib_call_init_function (vm, mfib_module_init))) return error; for (i = 0; i < ARRAY_LEN (im->fib_masks); i++) { u32 m; if (i < 32) m = pow2_mask (i) << (32 - i); else m = ~0; im->fib_masks[i] = clib_host_to_net_u32 (m); } ip_lookup_init (&im->lookup_main, /* is_ip6 */ 0); /* Create FIB with index 0 and table id of 0. */ fib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, 0, FIB_SOURCE_DEFAULT_ROUTE); mfib_table_find_or_create_and_lock (FIB_PROTOCOL_IP4, 0, MFIB_SOURCE_DEFAULT_ROUTE); { pg_node_t *pn; pn = pg_get_node (ip4_lookup_node.index); pn->unformat_edit = unformat_pg_ip4_header; } { ethernet_arp_header_t h; clib_memset (&h, 0, sizeof (h)); #define _16(f,v) h.f = clib_host_to_net_u16 (v); #define _8(f,v) h.f = v; _16 (l2_type, ETHERNET_ARP_HARDWARE_TYPE_ethernet); _16 (l3_type, ETHERNET_TYPE_IP4); _8 (n_l2_address_bytes, 6); _8 (n_l3_address_bytes, 4); _16 (opcode, ETHERNET_ARP_OPCODE_request); #undef _16 #undef _8 vlib_packet_template_init (vm, &im->ip4_arp_request_packet_template, /* data */ &h, sizeof (h), /* alloc chunk size */ 8, "ip4 arp"); } return error; } VLIB_INIT_FUNCTION (ip4_lookup_init); typedef struct { /* Adjacency taken. */ u32 dpo_index; u32 flow_hash; u32 fib_index; /* Packet data, possibly *after* rewrite. */ u8 packet_data[64 - 1 * sizeof (u32)]; } ip4_forward_next_trace_t; #ifndef CLIB_MARCH_VARIANT u8 * format_ip4_forward_next_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 *); ip4_forward_next_trace_t *t = va_arg (*args, ip4_forward_next_trace_t *); u32 indent = format_get_indent (s); s = format (s, "%U%U", format_white_space, indent, format_ip4_header, t->packet_data, sizeof (t->packet_data)); return s; } #endif static u8 * format_ip4_lookup_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 *); ip4_forward_next_trace_t *t = va_arg (*args, ip4_forward_next_trace_t *); u32 indent = format_get_indent (s); s = format (s, "fib %d dpo-idx %d flow hash: 0x%08x", t->fib_index, t->dpo_index, t->flow_hash); s = format (s, "\n%U%U", format_white_space, indent, format_ip4_header, t->packet_data, sizeof (t->packet_data)); return s; } static u8 * format_ip4_rewrite_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 *); ip4_forward_next_trace_t *t = va_arg (*args, ip4_forward_next_trace_t *); u32 indent = format_get_indent (s); s = format (s, "tx_sw_if_index %d dpo-idx %d : %U flow hash: 0x%08x", t->fib_index, t->dpo_index, format_ip_adjacency, t->dpo_index, FORMAT_IP_ADJACENCY_NONE, t->flow_hash); s = format (s, "\n%U%U", format_white_space, indent, format_ip_adjacency_packet_data, t->dpo_index, t->packet_data, sizeof (t->packet_data)); return s; } #ifndef CLIB_MARCH_VARIANT /* Common trace function for all ip4-forward next nodes. */ void ip4_forward_next_trace (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame, vlib_rx_or_tx_t which_adj_index) { u32 *from, n_left; ip4_main_t *im = &ip4_main; n_left = frame->n_vectors; from = vlib_frame_vector_args (frame); while (n_left >= 4) { u32 bi0, bi1; vlib_buffer_t *b0, *b1; ip4_forward_next_trace_t *t0, *t1; /* Prefetch next iteration. */ vlib_prefetch_buffer_with_index (vm, from[2], LOAD); vlib_prefetch_buffer_with_index (vm, from[3], LOAD); bi0 = from[0]; bi1 = from[1]; b0 = vlib_get_buffer (vm, bi0); b1 = vlib_get_buffer (vm, bi1); if (b0->flags & VLIB_BUFFER_IS_TRACED) { t0 = vlib_add_trace (vm, node, b0, sizeof (t0[0])); t0->dpo_index = vnet_buffer (b0)->ip.adj_index[which_adj_index]; t0->flow_hash = vnet_buffer (b0)->ip.flow_hash; t0->fib_index = (vnet_buffer (b0)->sw_if_index[VLIB_TX] != (u32) ~ 0) ? vnet_buffer (b0)->sw_if_index[VLIB_TX] : vec_elt (im->fib_index_by_sw_if_index, vnet_buffer (b0)->sw_if_index[VLIB_RX]); clib_memcpy_fast (t0->packet_data, vlib_buffer_get_current (b0), sizeof (t0->packet_data)); } if (b1->flags & VLIB_BUFFER_IS_TRACED) { t1 = vlib_add_trace (vm, node, b1, sizeof (t1[0])); t1->dpo_index = vnet_buffer (b1)->ip.adj_index[which_adj_index]; t1->flow_hash = vnet_buffer (b1)->ip.flow_hash; t1->fib_index = (vnet_buffer (b1)->sw_if_index[VLIB_TX] != (u32) ~ 0) ? vnet_buffer (b1)->sw_if_index[VLIB_TX] : vec_elt (im->fib_index_by_sw_if_index, vnet_buffer (b1)->sw_if_index[VLIB_RX]); clib_memcpy_fast (t1->packet_data, vlib_buffer_get_current (b1), sizeof (t1->packet_data)); } from += 2; n_left -= 2; } while (n_left >= 1) { u32 bi0; vlib_buffer_t *b0; ip4_forward_next_trace_t *t0; bi0 = from[0]; b0 = vlib_get_buffer (vm, bi0); if (b0->flags & VLIB_BUFFER_IS_TRACED) { t0 = vlib_add_trace (vm, node, b0, sizeof (t0[0])); t0->dpo_index = vnet_buffer (b0)->ip.adj_index[which_adj_index]; t0->flow_hash = vnet_buffer (b0)->ip.flow_hash; t0->fib_index = (vnet_buffer (b0)->sw_if_index[VLIB_TX] != (u32) ~ 0) ? vnet_buffer (b0)->sw_if_index[VLIB_TX] : vec_elt (im->fib_index_by_sw_if_index, vnet_buffer (b0)->sw_if_index[VLIB_RX]); clib_memcpy_fast (t0->packet_data, vlib_buffer_get_current (b0), sizeof (t0->packet_data)); } from += 1; n_left -= 1; } } /* Compute TCP/UDP/ICMP4 checksum in software. */ u16 ip4_tcp_udp_compute_checksum (vlib_main_t * vm, vlib_buffer_t * p0, ip4_header_t * ip0) { ip_csum_t sum0; u32 ip_header_length, payload_length_host_byte_order; u32 n_this_buffer, n_bytes_left, n_ip_bytes_this_buffer; u16 sum16; void *data_this_buffer; /* Initialize checksum with ip header. */ ip_header_length = ip4_header_bytes (ip0); payload_length_host_byte_order = clib_net_to_host_u16 (ip0->length) - ip_header_length; sum0 = clib_host_to_net_u32 (payload_length_host_byte_order + (ip0->protocol << 16)); if (BITS (uword) == 32) { sum0 = ip_csum_with_carry (sum0, clib_mem_unaligned (&ip0->src_address, u32)); sum0 = ip_csum_with_carry (sum0, clib_mem_unaligned (&ip0->dst_address, u32)); } else sum0 = ip_csum_with_carry (sum0, clib_mem_unaligned (&ip0->src_address, u64)); n_bytes_left = n_this_buffer = payload_length_host_byte_order; data_this_buffer = (void *) ip0 + ip_header_length; n_ip_bytes_this_buffer = p0->current_length - (((u8 *) ip0 - p0->data) - p0->current_data); if (n_this_buffer + ip_header_length > n_ip_bytes_this_buffer) { n_this_buffer = n_ip_bytes_this_buffer > ip_header_length ? n_ip_bytes_this_buffer - ip_header_length : 0; } while (1) { sum0 = ip_incremental_checksum (sum0, data_this_buffer, n_this_buffer); n_bytes_left -= n_this_buffer; if (n_bytes_left == 0) break; ASSERT (p0->flags & VLIB_BUFFER_NEXT_PRESENT); p0 = vlib_get_buffer (vm, p0->next_buffer); data_this_buffer = vlib_buffer_get_current (p0); n_this_buffer = clib_min (p0->current_length, n_bytes_left); } sum16 = ~ip_csum_fold (sum0); return sum16; } u32 ip4_tcp_udp_validate_checksum (vlib_main_t * vm, vlib_buffer_t * p0) { ip4_header_t *ip0 = vlib_buffer_get_current (p0); udp_header_t *udp0; u16 sum16; ASSERT (ip0->protocol == IP_PROTOCOL_TCP || ip0->protocol == IP_PROTOCOL_UDP); udp0 = (void *) (ip0 + 1); if (ip0->protocol == IP_PROTOCOL_UDP && udp0->checksum == 0) { p0->flags |= (VNET_BUFFER_F_L4_CHECKSUM_COMPUTED | VNET_BUFFER_F_L4_CHECKSUM_CORRECT); return p0->flags; } sum16 = ip4_tcp_udp_compute_checksum (vm, p0, ip0); p0->flags |= (VNET_BUFFER_F_L4_CHECKSUM_COMPUTED | ((sum16 == 0) << VNET_BUFFER_F_LOG2_L4_CHECKSUM_CORRECT)); return p0->flags; } #endif /* *INDENT-OFF* */ VNET_FEATURE_ARC_INIT (ip4_local) = { .arc_name = "ip4-local", .start_nodes = VNET_FEATURES ("ip4-local"), .last_in_arc = "ip4-local-end-of-arc", }; /* *INDENT-ON* */ static inline void ip4_local_l4_csum_validate (vlib_main_t * vm, vlib_buffer_t * p, ip4_header_t * ip, u8 is_udp, u8 * error, u8 * good_tcp_udp) { u32 flags0; flags0 = ip4_tcp_udp_validate_checksum (vm, p); *good_tcp_udp = (flags0 & VNET_BUFFER_F_L4_CHECKSUM_CORRECT) != 0; if (is_udp) { udp_header_t *udp; u32 ip_len, udp_len; i32 len_diff; udp = ip4_next_header (ip); /* Verify UDP length. */ ip_len = clib_net_to_host_u16 (ip->length); udp_len = clib_net_to_host_u16 (udp->length); len_diff = ip_len - udp_len; *good_tcp_udp &= len_diff >= 0; *error = len_diff < 0 ? IP4_ERROR_UDP_LENGTH : *error; } } #define ip4_local_csum_is_offloaded(_b) \ _b->flags & VNET_BUFFER_F_OFFLOAD_TCP_CKSUM \ || _b->flags & VNET_BUFFER_F_OFFLOAD_UDP_CKSUM #define ip4_local_need_csum_check(is_tcp_udp, _b) \ (is_tcp_udp && !(_b->flags & VNET_BUFFER_F_L4_CHECKSUM_COMPUTED \ || ip4_local_csum_is_offloaded (_b))) #define ip4_local_csum_is_valid(_b) \ (_b->flags & VNET_BUFFER_F_L4_CHECKSUM_CORRECT \ || (ip4_local_csum_is_offloaded (_b))) != 0 static inline void ip4_local_check_l4_csum (vlib_main_t * vm, vlib_buffer_t * b, ip4_header_t * ih, u8 * error) { u8 is_udp, is_tcp_udp, good_tcp_udp; is_udp = ih->protocol == IP_PROTOCOL_UDP; is_tcp_udp = is_udp || ih->protocol == IP_PROTOCOL_TCP; if (PREDICT_FALSE (ip4_local_need_csum_check (is_tcp_udp, b))) ip4_local_l4_csum_validate (vm, b, ih, is_udp, error, &good_tcp_udp); else good_tcp_udp = ip4_local_csum_is_valid (b); ASSERT (IP4_ERROR_TCP_CHECKSUM + 1 == IP4_ERROR_UDP_CHECKSUM); *error = (is_tcp_udp && !good_tcp_udp ? IP4_ERROR_TCP_CHECKSUM + is_udp : *error); } static inline void ip4_local_check_l4_csum_x2 (vlib_main_t * vm, vlib_buffer_t ** b, ip4_header_t ** ih, u8 * error) { u8 is_udp[2], is_tcp_udp[2], good_tcp_udp[2]; is_udp[0] = ih[0]->protocol == IP_PROTOCOL_UDP; is_udp[1] = ih[1]->protocol == IP_PROTOCOL_UDP; is_tcp_udp[0] = is_udp[0] || ih[0]->protocol == IP_PROTOCOL_TCP; is_tcp_udp[1] = is_udp[1] || ih[1]->protocol == IP_PROTOCOL_TCP; good_tcp_udp[0] = ip4_local_csum_is_valid (b[0]); good_tcp_udp[1] = ip4_local_csum_is_valid (b[1]); if (PREDICT_FALSE (ip4_local_need_csum_check (is_tcp_udp[0], b[0]) || ip4_local_need_csum_check (is_tcp_udp[1], b[1]))) { if (is_tcp_udp[0]) ip4_local_l4_csum_validate (vm, b[0], ih[0], is_udp[0], &error[0], &good_tcp_udp[0]); if (is_tcp_udp[1]) ip4_local_l4_csum_validate (vm, b[1], ih[1], is_udp[1], &error[1], &good_tcp_udp[1]); } error[0] = (is_tcp_udp[0] && !good_tcp_udp[0] ? IP4_ERROR_TCP_CHECKSUM + is_udp[0] : error[0]); error[1] = (is_tcp_udp[1] && !good_tcp_udp[1] ? IP4_ERROR_TCP_CHECKSUM + is_udp[1] : error[1]); } static inline void ip4_local_set_next_and_error (vlib_node_runtime_t * error_node, vlib_buffer_t * b, u16 * next, u8 error, u8 head_of_feature_arc) { u8 arc_index = vnet_feat_arc_ip4_local.feature_arc_index; u32 next_index; *next = error != IP4_ERROR_UNKNOWN_PROTOCOL ? IP_LOCAL_NEXT_DROP : *next; b->error = error ? error_node->errors[error] : 0; if (head_of_feature_arc) { next_index = *next; if (PREDICT_TRUE (error == (u8) IP4_ERROR_UNKNOWN_PROTOCOL)) { vnet_feature_arc_start (arc_index, vnet_buffer (b)->sw_if_index[VLIB_RX], &next_index, b); *next = next_index; } } } typedef struct { ip4_address_t src; u32 lbi; u8 error; u8 first; } ip4_local_last_check_t; static inline void ip4_local_check_src (vlib_buffer_t * b, ip4_header_t * ip0, ip4_local_last_check_t * last_check, u8 * error0) { ip4_fib_mtrie_leaf_t leaf0; ip4_fib_mtrie_t *mtrie0; const dpo_id_t *dpo0; load_balance_t *lb0; u32 lbi0; vnet_buffer (b)->ip.fib_index = vnet_buffer (b)->sw_if_index[VLIB_TX] != ~0 ? vnet_buffer (b)->sw_if_index[VLIB_TX] : vnet_buffer (b)->ip.fib_index; if (PREDICT_FALSE (last_check->first || (last_check->src.as_u32 != ip0->src_address.as_u32))) { mtrie0 = &ip4_fib_get (vnet_buffer (b)->ip.fib_index)->mtrie; leaf0 = ip4_fib_mtrie_lookup_step_one (mtrie0, &ip0->src_address); leaf0 = ip4_fib_mtrie_lookup_step (mtrie0, leaf0, &ip0->src_address, 2); leaf0 = ip4_fib_mtrie_lookup_step (mtrie0, leaf0, &ip0->src_address, 3); lbi0 = ip4_fib_mtrie_leaf_get_adj_index (leaf0); vnet_buffer (b)->ip.adj_index[VLIB_TX] = lbi0; vnet_buffer (b)->ip.adj_index[VLIB_RX] = lbi0; lb0 = load_balance_get (lbi0); dpo0 = load_balance_get_bucket_i (lb0, 0); /* * Must have a route to source otherwise we drop the packet. * ip4 broadcasts are accepted, e.g. to make dhcp client work * * The checks are: * - the source is a recieve => it's from us => bogus, do this * first since it sets a different error code. * - uRPF check for any route to source - accept if passes. * - allow packets destined to the broadcast address from unknown sources */ *error0 = ((*error0 == IP4_ERROR_UNKNOWN_PROTOCOL && dpo0->dpoi_type == DPO_RECEIVE) ? IP4_ERROR_SPOOFED_LOCAL_PACKETS : *error0); *error0 = ((*error0 == IP4_ERROR_UNKNOWN_PROTOCOL && !fib_urpf_check_size (lb0->lb_urpf) && ip0->dst_address.as_u32 != 0xFFFFFFFF) ? IP4_ERROR_SRC_LOOKUP_MISS : *error0); last_check->src.as_u32 = ip0->src_address.as_u32; last_check->lbi = lbi0; last_check->error = *error0; } else { vnet_buffer (b)->ip.adj_index[VLIB_TX] = last_check->lbi; vnet_buffer (b)->ip.adj_index[VLIB_RX] = last_check->lbi; *error0 = last_check->error; last_check->first = 0; } } static inline void ip4_local_check_src_x2 (vlib_buffer_t ** b, ip4_header_t ** ip, ip4_local_last_check_t * last_check, u8 * error) { ip4_fib_mtrie_leaf_t leaf[2]; ip4_fib_mtrie_t *mtrie[2]; const dpo_id_t *dpo[2]; load_balance_t *lb[2]; u32 not_last_hit; u32 lbi[2]; not_last_hit = last_check->first; not_last_hit |= ip[0]->src_address.as_u32 ^ last_check->src.as_u32; not_last_hit |= ip[1]->src_address.as_u32 ^ last_check->src.as_u32; vnet_buffer (b[0])->ip.fib_index = vnet_buffer (b[0])->sw_if_index[VLIB_TX] != ~0 ? vnet_buffer (b[0])->sw_if_index[VLIB_TX] : vnet_buffer (b[0])->ip.fib_index; vnet_buffer (b[1])->ip.fib_index = vnet_buffer (b[1])->sw_if_index[VLIB_TX] != ~0 ? vnet_buffer (b[1])->sw_if_index[VLIB_TX] : vnet_buffer (b[1])->ip.fib_index; if (PREDICT_FALSE (not_last_hit)) { mtrie[0] = &ip4_fib_get (vnet_buffer (b[0])->ip.fib_index)->mtrie; mtrie[1] = &ip4_fib_get (vnet_buffer (b[1])->ip.fib_index)->mtrie; leaf[0] = ip4_fib_mtrie_lookup_step_one (mtrie[0], &ip[0]->src_address); leaf[1] = ip4_fib_mtrie_lookup_step_one (mtrie[1], &ip[1]->src_address); leaf[0] = ip4_fib_mtrie_lookup_step (mtrie[0], leaf[0], &ip[0]->src_address, 2); leaf[1] = ip4_fib_mtrie_lookup_step (mtrie[1], leaf[1], &ip[1]->src_address, 2); leaf[0] = ip4_fib_mtrie_lookup_step (mtrie[0], leaf[0], &ip[0]->src_address, 3); leaf[1] = ip4_fib_mtrie_lookup_step (mtrie[1], leaf[1], &ip[1]->src_address, 3); lbi[0] = ip4_fib_mtrie_leaf_get_adj_index (leaf[0]); lbi[1] = ip4_fib_mtrie_leaf_get_adj_index (leaf[1]); vnet_buffer (b[0])->ip.adj_index[VLIB_TX] = lbi[0]; vnet_buffer (b[0])->ip.adj_index[VLIB_RX] = lbi[0]; vnet_buffer (b[1])->ip.adj_index[VLIB_TX] = lbi[1]; vnet_buffer (b[1])->ip.adj_index[VLIB_RX] = lbi[1]; lb[0] = load_balance_get (lbi[0]); lb[1] = load_balance_get (lbi[1]); dpo[0] = load_balance_get_bucket_i (lb[0], 0); dpo[1] = load_balance_get_bucket_i (lb[1], 0); error[0] = ((error[0] == IP4_ERROR_UNKNOWN_PROTOCOL && dpo[0]->dpoi_type == DPO_RECEIVE) ? IP4_ERROR_SPOOFED_LOCAL_PACKETS : error[0]); error[0] = ((error[0] == IP4_ERROR_UNKNOWN_PROTOCOL && !fib_urpf_check_size (lb[0]->lb_urpf) && ip[0]->dst_address.as_u32 != 0xFFFFFFFF) ? IP4_ERROR_SRC_LOOKUP_MISS : error[0]); error[1] = ((error[1] == IP4_ERROR_UNKNOWN_PROTOCOL && dpo[1]->dpoi_type == DPO_RECEIVE) ? IP4_ERROR_SPOOFED_LOCAL_PACKETS : error[1]); error[1] = ((error[1] == IP4_ERROR_UNKNOWN_PROTOCOL && !fib_urpf_check_size (lb[1]->lb_urpf) && ip[1]->dst_address.as_u32 != 0xFFFFFFFF) ? IP4_ERROR_SRC_LOOKUP_MISS : error[1]); last_check->src.as_u32 = ip[1]->src_address.as_u32; last_check->lbi = lbi[1]; last_check->error = error[1]; } else { vnet_buffer (b[0])->ip.adj_index[VLIB_TX] = last_check->lbi; vnet_buffer (b[0])->ip.adj_index[VLIB_RX] = last_check->lbi; vnet_buffer (b[1])->ip.adj_index[VLIB_TX] = last_check->lbi; vnet_buffer (b[1])->ip.adj_index[VLIB_RX] = last_check->lbi; error[0] = last_check->error; error[1] = last_check->error; last_check->first = 0; } } enum ip_local_packet_type_e { IP_LOCAL_PACKET_TYPE_L4, IP_LOCAL_PACKET_TYPE_NAT, IP_LOCAL_PACKET_TYPE_FRAG, }; /** * Determine packet type and next node. * * The expectation is that all packets that are not L4 will skip * checksums and source checks. */ always_inline u8 ip4_local_classify (vlib_buffer_t * b, ip4_header_t * ip, u16 * next) { ip_lookup_main_t *lm = &ip4_main.lookup_main; if (PREDICT_FALSE (ip4_is_fragment (ip))) { *next = IP_LOCAL_NEXT_REASSEMBLY; return IP_LOCAL_PACKET_TYPE_FRAG; } if (PREDICT_FALSE (b->flags & VNET_BUFFER_F_IS_NATED)) { *next = lm->local_next_by_ip_protocol[ip->protocol]; return IP_LOCAL_PACKET_TYPE_NAT; } *next = lm->local_next_by_ip_protocol[ip->protocol]; return IP_LOCAL_PACKET_TYPE_L4; } static inline uword ip4_local_inline (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame, int head_of_feature_arc) { u32 *from, n_left_from; vlib_node_runtime_t *error_node = vlib_node_get_runtime (vm, ip4_input_node.index); u16 nexts[VLIB_FRAME_SIZE], *next; vlib_buffer_t *bufs[VLIB_FRAME_SIZE], **b; ip4_header_t *ip[2]; u8 error[2], pt[2]; ip4_local_last_check_t last_check = { /* * 0.0.0.0 can appear as the source address of an IP packet, * as can any other address, hence the need to use the 'first' * member to make sure the .lbi is initialised for the first * packet. */ .src = {.as_u32 = 0}, .lbi = ~0, .error = IP4_ERROR_UNKNOWN_PROTOCOL, .first = 1, }; from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; if (node->flags & VLIB_NODE_FLAG_TRACE) ip4_forward_next_trace (vm, node, frame, VLIB_TX); vlib_get_buffers (vm, from, bufs, n_left_from); b = bufs; next = nexts; while (n_left_from >= 6) { u8 not_batch = 0; /* Prefetch next iteration. */ { vlib_prefetch_buffer_header (b[4], LOAD); vlib_prefetch_buffer_header (b[5], LOAD); CLIB_PREFETCH (b[4]->data, CLIB_CACHE_LINE_BYTES, LOAD); CLIB_PREFETCH (b[5]->data, CLIB_CACHE_LINE_BYTES, LOAD); } error[0] = error[1] = IP4_ERROR_UNKNOWN_PROTOCOL; ip[0] = vlib_buffer_get_current (b[0]); ip[1] = vlib_buffer_get_current (b[1]); vnet_buffer (b[0])->l3_hdr_offset = b[0]->current_data; vnet_buffer (b[1])->l3_hdr_offset = b[1]->current_data; pt[0] = ip4_local_classify (b[0], ip[0], &next[0]); pt[1] = ip4_local_classify (b[1], ip[1], &next[1]); not_batch = pt[0] ^ pt[1]; if (head_of_feature_arc == 0 || (pt[0] && not_batch == 0)) goto skip_checks; if (PREDICT_TRUE (not_batch == 0)) { ip4_local_check_l4_csum_x2 (vm, b, ip, error); ip4_local_check_src_x2 (b, ip, &last_check, error); } else { if (!pt[0]) { ip4_local_check_l4_csum (vm, b[0], ip[0], &error[0]); ip4_local_check_src (b[0], ip[0], &last_check, &error[0]); } if (!pt[1]) { ip4_local_check_l4_csum (vm, b[1], ip[1], &error[1]); ip4_local_check_src (b[1], ip[1], &last_check, &error[1]); } } skip_checks: ip4_local_set_next_and_error (error_node, b[0], &next[0], error[0], head_of_feature_arc); ip4_local_set_next_and_error (error_node, b[1], &next[1], error[1], head_of_feature_arc); b += 2; next += 2; n_left_from -= 2; } while (n_left_from > 0) { error[0] = IP4_ERROR_UNKNOWN_PROTOCOL; ip[0] = vlib_buffer_get_current (b[0]); vnet_buffer (b[0])->l3_hdr_offset = b[0]->current_data; pt[0] = ip4_local_classify (b[0], ip[0], &next[0]); if (head_of_feature_arc == 0 || pt[0]) goto skip_check; ip4_local_check_l4_csum (vm, b[0], ip[0], &error[0]); ip4_local_check_src (b[0], ip[0], &last_check, &error[0]); skip_check: ip4_local_set_next_and_error (error_node, b[0], &next[0], error[0], head_of_feature_arc); b += 1; next += 1; n_left_from -= 1; } vlib_buffer_enqueue_to_next (vm, node, from, nexts, frame->n_vectors); return frame->n_vectors; } VLIB_NODE_FN (ip4_local_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { return ip4_local_inline (vm, node, frame, 1 /* head of feature arc */ ); } /* *INDENT-OFF* */ VLIB_REGISTER_NODE (ip4_local_node) = { .name = "ip4-local", .vector_size = sizeof (u32), .format_trace = format_ip4_forward_next_trace, .n_next_nodes = IP_LOCAL_N_NEXT, .next_nodes = { [IP_LOCAL_NEXT_DROP] = "ip4-drop", [IP_LOCAL_NEXT_PUNT] = "ip4-punt", [IP_LOCAL_NEXT_UDP_LOOKUP] = "ip4-udp-lookup", [IP_LOCAL_NEXT_ICMP] = "ip4-icmp-input", [IP_LOCAL_NEXT_REASSEMBLY] = "ip4-reassembly", }, }; /* *INDENT-ON* */ VLIB_NODE_FN (ip4_local_end_of_arc_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { return ip4_local_inline (vm, node, frame, 0 /* head of feature arc */ ); } /* *INDENT-OFF* */ VLIB_REGISTER_NODE (ip4_local_end_of_arc_node) = { .name = "ip4-local-end-of-arc", .vector_size = sizeof (u32), .format_trace = format_ip4_forward_next_trace, .sibling_of = "ip4-local", }; VNET_FEATURE_INIT (ip4_local_end_of_arc, static) = { .arc_name = "ip4-local", .node_name = "ip4-local-end-of-arc", .runs_before = 0, /* not before any other features */ }; /* *INDENT-ON* */ #ifndef CLIB_MARCH_VARIANT void ip4_register_protocol (u32 protocol, u32 node_index) { vlib_main_t *vm = vlib_get_main (); ip4_main_t *im = &ip4_main; ip_lookup_main_t *lm = &im->lookup_main; ASSERT (protocol < ARRAY_LEN (lm->local_next_by_ip_protocol)); lm->local_next_by_ip_protocol[protocol] = vlib_node_add_next (vm, ip4_local_node.index, node_index); } void ip4_unregister_protocol (u32 protocol) { ip4_main_t *im = &ip4_main; ip_lookup_main_t *lm = &im->lookup_main; ASSERT (protocol < ARRAY_LEN (lm->local_next_by_ip_protocol)); lm->local_next_by_ip_protocol[protocol] = IP_LOCAL_NEXT_PUNT; } #endif static clib_error_t * show_ip_local_command_fn (vlib_main_t * vm, unformat_input_t * input, vlib_cli_command_t * cmd) { ip4_main_t *im = &ip4_main; ip_lookup_main_t *lm = &im->lookup_main; int i; vlib_cli_output (vm, "Protocols handled by ip4_local"); for (i = 0; i < ARRAY_LEN (lm->local_next_by_ip_protocol); i++) { if (lm->local_next_by_ip_protocol[i] != IP_LOCAL_NEXT_PUNT) { u32 node_index = vlib_get_node (vm, ip4_local_node.index)-> next_nodes[lm->local_next_by_ip_protocol[i]]; vlib_cli_output (vm, "%U: %U", format_ip_protocol, i, format_vlib_node_name, vm, node_index); } } return 0; } /*? * Display the set of protocols handled by the local IPv4 stack. * * @cliexpar * Example of how to display local protocol table: * @cliexstart{show ip local} * Protocols handled by ip4_local * 1 * 17 * 47 * @cliexend ?*/ /* *INDENT-OFF* */ VLIB_CLI_COMMAND (show_ip_local, static) = { .path = "show ip local", .function = show_ip_local_command_fn, .short_help = "show ip local", }; /* *INDENT-ON* */ always_inline uword ip4_arp_inline (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame, int is_glean) { vnet_main_t *vnm = vnet_get_main (); ip4_main_t *im = &ip4_main; ip_lookup_main_t *lm = &im->lookup_main; u32 *from, *to_next_drop; uword n_left_from, n_left_to_next_drop, next_index; u32 thread_index = vm->thread_index; u64 seed; if (node->flags & VLIB_NODE_FLAG_TRACE) ip4_forward_next_trace (vm, node, frame, VLIB_TX); seed = throttle_seed (&im->arp_throttle, thread_index, vlib_time_now (vm)); from = vlib_frame_vector_args (frame); n_left_from = frame->n_vectors; next_index = node->cached_next_index; if (next_index == IP4_ARP_NEXT_DROP) next_index = IP4_ARP_N_NEXT; /* point to first interface */ while (n_left_from > 0) { vlib_get_next_frame (vm, node, IP4_ARP_NEXT_DROP, to_next_drop, n_left_to_next_drop); while (n_left_from > 0 && n_left_to_next_drop > 0) { u32 pi0, bi0, adj_index0, sw_if_index0; ip_adjacency_t *adj0; vlib_buffer_t *p0, *b0; ip4_address_t resolve0; ethernet_arp_header_t *h0; vnet_hw_interface_t *hw_if0; u64 r0; pi0 = from[0]; p0 = vlib_get_buffer (vm, pi0); from += 1; n_left_from -= 1; to_next_drop[0] = pi0; to_next_drop += 1; n_left_to_next_drop -= 1; adj_index0 = vnet_buffer (p0)->ip.adj_index[VLIB_TX]; adj0 = adj_get (adj_index0); if (is_glean) { /* resolve the packet's destination */ ip4_header_t *ip0 = vlib_buffer_get_current (p0); resolve0 = ip0->dst_address; } else { /* resolve the incomplete adj */ resolve0 = adj0->sub_type.nbr.next_hop.ip4; } /* combine the address and interface for the hash key */ sw_if_index0 = adj0->rewrite_header.sw_if_index; r0 = (u64) resolve0.data_u32 << 32; r0 |= sw_if_index0; if (throttle_check (&im->arp_throttle, thread_index, r0, seed)) { p0->error = node->errors[IP4_ARP_ERROR_THROTTLED]; continue; } /* * the adj has been updated to a rewrite but the node the DPO that got * us here hasn't - yet. no big deal. we'll drop while we wait. */ if (IP_LOOKUP_NEXT_REWRITE == adj0->lookup_next_index) { p0->error = node->errors[IP4_ARP_ERROR_RESOLVED]; continue; } /* * Can happen if the control-plane is programming tables * with traffic flowing; at least that's today's lame excuse. */ if ((is_glean && adj0->lookup_next_index != IP_LOOKUP_NEXT_GLEAN) || (!is_glean && adj0->lookup_next_index != IP_LOOKUP_NEXT_ARP)) { p0->error = node->errors[IP4_ARP_ERROR_NON_ARP_ADJ]; continue; } /* Send ARP request. */ h0 = vlib_packet_template_get_packet (vm, &im->ip4_arp_request_packet_template, &bi0); /* Seems we're out of buffers */ if (PREDICT_FALSE (!h0)) { p0->error = node->errors[IP4_ARP_ERROR_NO_BUFFERS]; continue; } b0 = vlib_get_buffer (vm, bi0); /* copy the persistent fields from the original */ clib_memcpy_fast (b0->opaque2, p0->opaque2, sizeof (p0->opaque2)); /* Add rewrite/encap string for ARP packet. */ vnet_rewrite_one_header (adj0[0], h0, sizeof (ethernet_header_t)); hw_if0 = vnet_get_sup_hw_interface (vnm, sw_if_index0); /* Src ethernet address in ARP header. */ mac_address_from_bytes (&h0->ip4_over_ethernet[0].mac, hw_if0->hw_address); if (is_glean) { /* The interface's source address is stashed in the Glean Adj */ h0->ip4_over_ethernet[0].ip4 = adj0->sub_type.glean.receive_addr.ip4; } else { /* Src IP address in ARP header. */ if (ip4_src_address_for_packet (lm, sw_if_index0, &h0->ip4_over_ethernet[0].ip4)) { /* No source address available */ p0->error = node->errors[IP4_ARP_ERROR_NO_SOURCE_ADDRESS]; vlib_buffer_free (vm, &bi0, 1); continue; } } h0->ip4_over_ethernet[1].ip4 = resolve0; p0->error = node->errors[IP4_ARP_ERROR_REQUEST_SENT]; vlib_buffer_copy_trace_flag (vm, p0, bi0); VLIB_BUFFER_TRACE_TRAJECTORY_INIT (b0); vnet_buffer (b0)->sw_if_index[VLIB_TX] = sw_if_index0; vlib_buffer_advance (b0, -adj0->rewrite_header.data_bytes); vlib_set_next_frame_buffer (vm, node, adj0->rewrite_header.next_index, bi0); } vlib_put_next_frame (vm, node, IP4_ARP_NEXT_DROP, n_left_to_next_drop); } return frame->n_vectors; } VLIB_NODE_FN (ip4_arp_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { return (ip4_arp_inline (vm, node, frame, 0)); } VLIB_NODE_FN (ip4_glean_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { return (ip4_arp_inline (vm, node, frame, 1)); } static char *ip4_arp_error_strings[] = { [IP4_ARP_ERROR_THROTTLED] = "ARP requests throttled", [IP4_ARP_ERROR_RESOLVED] = "ARP requests resolved", [IP4_ARP_ERROR_NO_BUFFERS] = "ARP requests out of buffer", [IP4_ARP_ERROR_REQUEST_SENT] = "ARP requests sent", [IP4_ARP_ERROR_NON_ARP_ADJ] = "ARPs to non-ARP adjacencies", [IP4_ARP_ERROR_NO_SOURCE_ADDRESS] = "no source address for ARP request", }; /* *INDENT-OFF* */ VLIB_REGISTER_NODE (ip4_arp_node) = { .name = "ip4-arp", .vector_size = sizeof (u32), .format_trace = format_ip4_forward_next_trace, .n_errors = ARRAY_LEN (ip4_arp_error_strings), .error_strings = ip4_arp_error_strings, .n_next_nodes = IP4_ARP_N_NEXT, .next_nodes = { [IP4_ARP_NEXT_DROP] = "error-drop", }, }; VLIB_REGISTER_NODE (ip4_glean_node) = { .name = "ip4-glean", .vector_size = sizeof (u32), .format_trace = format_ip4_forward_next_trace, .n_errors = ARRAY_LEN (ip4_arp_error_strings), .error_strings = ip4_arp_error_strings, .n_next_nodes = IP4_ARP_N_NEXT, .next_nodes = { [IP4_ARP_NEXT_DROP] = "error-drop", }, }; /* *INDENT-ON* */ #define foreach_notrace_ip4_arp_error \ _(THROTTLED) \ _(RESOLVED) \ _(NO_BUFFERS) \ _(REQUEST_SENT) \ _(NON_ARP_ADJ) \ _(NO_SOURCE_ADDRESS) static clib_error_t * arp_notrace_init (vlib_main_t * vm) { vlib_node_runtime_t *rt = vlib_node_get_runtime (vm, ip4_arp_node.index); /* don't trace ARP request packets */ #define _(a) \ vnet_pcap_drop_trace_filter_add_del \ (rt->errors[IP4_ARP_ERROR_##a], \ 1 /* is_add */); foreach_notrace_ip4_arp_error; #undef _ return 0; } VLIB_INIT_FUNCTION (arp_notrace_init); #ifndef CLIB_MARCH_VARIANT /* Send an ARP request to see if given destination is reachable on given interface. */ clib_error_t * ip4_probe_neighbor (vlib_main_t * vm, ip4_address_t * dst, u32 sw_if_index, u8 refresh) { vnet_main_t *vnm = vnet_get_main (); ip4_main_t *im = &ip4_main; ethernet_arp_header_t *h; ip4_address_t *src; ip_interface_address_t *ia; ip_adjacency_t *adj; vnet_hw_interface_t *hi; vnet_sw_interface_t *si; vlib_buffer_t *b; adj_index_t ai; u32 bi = 0; u8 unicast_rewrite = 0; si = vnet_get_sw_interface (vnm, sw_if_index); if (!(si->flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP)) { return clib_error_return (0, "%U: interface %U down", format_ip4_address, dst, format_vnet_sw_if_index_name, vnm, sw_if_index); } src = ip4_interface_address_matching_destination (im, dst, sw_if_index, &ia); if (!src) { vnm->api_errno = VNET_API_ERROR_NO_MATCHING_INTERFACE; return clib_error_return (0, "no matching interface address for destination %U (interface %U)", format_ip4_address, dst, format_vnet_sw_if_index_name, vnm, sw_if_index); } h = vlib_packet_template_get_packet (vm, &im->ip4_arp_request_packet_template, &bi); if (!h) return clib_error_return (0, "ARP request packet allocation failed"); hi = vnet_get_sup_hw_interface (vnm, sw_if_index); if (PREDICT_FALSE (!hi->hw_address)) { return clib_error_return (0, "%U: interface %U do not support ip probe", format_ip4_address, dst, format_vnet_sw_if_index_name, vnm, sw_if_index); } mac_address_from_bytes (&h->ip4_over_ethernet[0].mac, hi->hw_address); h->ip4_over_ethernet[0].ip4 = src[0]; h->ip4_over_ethernet[1].ip4 = dst[0]; b = vlib_get_buffer (vm, bi); vnet_buffer (b)->sw_if_index[VLIB_RX] = vnet_buffer (b)->sw_if_index[VLIB_TX] = sw_if_index; ip46_address_t nh = { .ip4 = *dst, }; ai = adj_nbr_add_or_lock (FIB_PROTOCOL_IP4, VNET_LINK_IP4, &nh, sw_if_index); adj = adj_get (ai); /* Peer has been previously resolved, retrieve glean adj instead */ if (adj->lookup_next_index == IP_LOOKUP_NEXT_REWRITE) { if (refresh) unicast_rewrite = 1; else { adj_unlock (ai); ai = adj_glean_add_or_lock (FIB_PROTOCOL_IP4, VNET_LINK_IP4, sw_if_index, &nh); adj = adj_get (ai); } } /* Add encapsulation string for software interface (e.g. ethernet header). */ vnet_rewrite_one_header (adj[0], h, sizeof (ethernet_header_t)); if (unicast_rewrite) { u16 *etype = vlib_buffer_get_current (b) - 2; etype[0] = clib_host_to_net_u16 (ETHERNET_TYPE_ARP); } vlib_buffer_advance (b, -adj->rewrite_header.data_bytes); { vlib_frame_t *f = vlib_get_frame_to_node (vm, hi->output_node_index); u32 *to_next = vlib_frame_vector_args (f); to_next[0] = bi; f->n_vectors = 1; vlib_put_frame_to_node (vm, hi->output_node_index, f); } adj_unlock (ai); return /* no error */ 0; } #endif typedef enum { IP4_REWRITE_NEXT_DROP, IP4_REWRITE_NEXT_ICMP_ERROR, IP4_REWRITE_NEXT_FRAGMENT, IP4_REWRITE_N_NEXT /* Last */ } ip4_rewrite_next_t; /** * This bits of an IPv4 address to mask to construct a multicast * MAC address */ #if CLIB_ARCH_IS_BIG_ENDIAN #define IP4_MCAST_ADDR_MASK 0x007fffff #else #define IP4_MCAST_ADDR_MASK 0xffff7f00 #endif always_inline void ip4_mtu_check (vlib_buffer_t * b, u16 packet_len, u16 adj_packet_bytes, bool df, u16 * next, u32 * error) { if (packet_len > adj_packet_bytes) { *error = IP4_ERROR_MTU_EXCEEDED; if (df) { icmp4_error_set_vnet_buffer (b, ICMP4_destination_unreachable, ICMP4_destination_unreachable_fragmentation_needed_and_dont_fragment_set, adj_packet_bytes); *next = IP4_REWRITE_NEXT_ICMP_ERROR; } else { /* IP fragmentation */ ip_frag_set_vnet_buffer (b, adj_packet_bytes, IP4_FRAG_NEXT_IP4_REWRITE, 0); *next = IP4_REWRITE_NEXT_FRAGMENT; } } } /* Decrement TTL & update checksum. Works either endian, so no need for byte swap. */ static_always_inline void ip4_ttl_and_checksum_check (vlib_buffer_t * b, ip4_header_t * ip, u16 * next, u32 * error) { i32 ttl; u32 checksum; if (PREDICT_FALSE (b->flags & VNET_BUFFER_F_LOCALLY_ORIGINATED)) { b->flags &= ~VNET_BUFFER_F_LOCALLY_ORIGINATED; return; } ttl = ip->ttl; /* Input node should have reject packets with ttl 0. */ ASSERT (ip->ttl > 0); checksum = ip->checksum + clib_host_to_net_u16 (0x0100); checksum += checksum >= 0xffff; ip->checksum = checksum; ttl -= 1; ip->ttl = ttl; /* * If the ttl drops below 1 when forwarding, generate * an ICMP response. */ if (PREDICT_FALSE (ttl <= 0)) { *error = IP4_ERROR_TIME_EXPIRED; vnet_buffer (b)->sw_if_index[VLIB_TX] = (u32) ~ 0; icmp4_error_set_vnet_buffer (b, ICMP4_time_exceeded, ICMP4_time_exceeded_ttl_exceeded_in_transit, 0); *next = IP4_REWRITE_NEXT_ICMP_ERROR; } /* Verify checksum. */ ASSERT ((ip->checksum == ip4_header_checksum (ip)) || (b->flags & VNET_BUFFER_F_OFFLOAD_IP_CKSUM)); } always_inline uword ip4_rewrite_inline_with_gso (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame, int do_counters, int is_midchain, int is_mcast, int do_gso) { ip_lookup_main_t *lm = &ip4_main.lookup_main; u32 *from = vlib_frame_vector_args (frame); vlib_buffer_t *bufs[VLIB_FRAME_SIZE], **b; u16 nexts[VLIB_FRAME_SIZE], *next; u32 n_left_from; vlib_node_runtime_t *error_node = vlib_node_get_runtime (vm, ip4_input_node.index); n_left_from = frame->n_vectors; u32 thread_index = vm->thread_index; vlib_get_buffers (vm, from, bufs, n_left_from); clib_memset_u16 (nexts, IP4_REWRITE_NEXT_DROP, n_left_from); if (n_left_from >= 6) { int i; for (i = 2; i < 6; i++) vlib_prefetch_buffer_header (bufs[i], LOAD); } next = nexts; b = bufs; while (n_left_from >= 8) { ip_adjacency_t *adj0, *adj1; ip4_header_t *ip0, *ip1; u32 rw_len0, error0, adj_index0; u32 rw_len1, error1, adj_index1; u32 tx_sw_if_index0, tx_sw_if_index1; u8 *p; vlib_prefetch_buffer_header (b[6], LOAD); vlib_prefetch_buffer_header (b[7], LOAD); adj_index0 = vnet_buffer (b[0])->ip.adj_index[VLIB_TX]; adj_index1 = vnet_buffer (b[1])->ip.adj_index[VLIB_TX]; /* * pre-fetch the per-adjacency counters */ if (do_counters) { vlib_prefetch_combined_counter (&adjacency_counters, thread_index, adj_index0); vlib_prefetch_combined_counter (&adjacency_counters, thread_index, adj_index1); } ip0 = vlib_buffer_get_current (b[0]); ip1 = vlib_buffer_get_current (b[1]); error0 = error1 = IP4_ERROR_NONE; ip4_ttl_and_checksum_check (b[0], ip0, next + 0, &error0); ip4_ttl_and_checksum_check (b[1], ip1, next + 1, &error1); /* Rewrite packet header and updates lengths. */ adj0 = adj_get (adj_index0); adj1 = adj_get (adj_index1); /* Worth pipelining. No guarantee that adj0,1 are hot... */ rw_len0 = adj0[0].rewrite_header.data_bytes; rw_len1 = adj1[0].rewrite_header.data_bytes; vnet_buffer (b[0])->ip.save_rewrite_length = rw_len0; vnet_buffer (b[1])->ip.save_rewrite_length = rw_len1; p = vlib_buffer_get_current (b[2]); CLIB_PREFETCH (p - CLIB_CACHE_LINE_BYTES, CLIB_CACHE_LINE_BYTES, STORE); CLIB_PREFETCH (p, CLIB_CACHE_LINE_BYTES, LOAD); p = vlib_buffer_get_current (b[3]); CLIB_PREFETCH (p - CLIB_CACHE_LINE_BYTES, CLIB_CACHE_LINE_BYTES, STORE); CLIB_PREFETCH (p, CLIB_CACHE_LINE_BYTES, LOAD); /* Check MTU of outgoing interface. */ u16 ip0_len = clib_net_to_host_u16 (ip0->length); u16 ip1_len = clib_net_to_host_u16 (ip1->length); if (do_gso && (b[0]->flags & VNET_BUFFER_F_GSO)) ip0_len = gso_mtu_sz (b[0]); if (do_gso && (b[1]->flags & VNET_BUFFER_F_GSO)) ip1_len = gso_mtu_sz (b[1]); ip4_mtu_check (b[0], ip0_len, adj0[0].rewrite_header.max_l3_packet_bytes, ip0->flags_and_fragment_offset & clib_host_to_net_u16 (IP4_HEADER_FLAG_DONT_FRAGMENT), next + 0, &error0); ip4_mtu_check (b[1], ip1_len, adj1[0].rewrite_header.max_l3_packet_bytes, ip1->flags_and_fragment_offset & clib_host_to_net_u16 (IP4_HEADER_FLAG_DONT_FRAGMENT), next + 1, &error1); if (is_mcast) { error0 = ((adj0[0].rewrite_header.sw_if_index == vnet_buffer (b[0])->sw_if_index[VLIB_RX]) ? IP4_ERROR_SAME_INTERFACE : error0); error1 = ((adj1[0].rewrite_header.sw_if_index == vnet_buffer (b[1])->sw_if_index[VLIB_RX]) ? IP4_ERROR_SAME_INTERFACE : error1); } /* Don't adjust the buffer for ttl issue; icmp-error node wants * to see the IP header */ if (PREDICT_TRUE (error0 == IP4_ERROR_NONE)) { u32 next_index = adj0[0].rewrite_header.next_index; b[0]->current_data -= rw_len0; b[0]->current_length += rw_len0; tx_sw_if_index0 = adj0[0].rewrite_header.sw_if_index; vnet_buffer (b[0])->sw_if_index[VLIB_TX] = tx_sw_if_index0; if (PREDICT_FALSE (adj0[0].rewrite_header.flags & VNET_REWRITE_HAS_FEATURES)) vnet_feature_arc_start (lm->output_feature_arc_index, tx_sw_if_index0, &next_index, b[0]); next[0] = next_index; } else { b[0]->error = error_node->errors[error0]; } if (PREDICT_TRUE (error1 == IP4_ERROR_NONE)) { u32 next_index = adj1[0].rewrite_header.next_index; b[1]->current_data -= rw_len1; b[1]->current_length += rw_len1; tx_sw_if_index1 = adj1[0].rewrite_header.sw_if_index; vnet_buffer (b[1])->sw_if_index[VLIB_TX] = tx_sw_if_index1; if (PREDICT_FALSE (adj1[0].rewrite_header.flags & VNET_REWRITE_HAS_FEATURES)) vnet_feature_arc_start (lm->output_feature_arc_index, tx_sw_if_index1, &next_index, b[1]); next[1] = next_index; } else { b[1]->error = error_node->errors[error1]; } if (is_midchain) { calc_checksums (vm, b[0]); calc_checksums (vm, b[1]); } /* Guess we are only writing on simple Ethernet header. */ vnet_rewrite_two_headers (adj0[0], adj1[0], ip0, ip1, sizeof (ethernet_header_t)); /* * Bump the per-adjacency counters */ if (do_counters) { vlib_increment_combined_counter (&adjacency_counters, thread_index, adj_index0, 1, vlib_buffer_length_in_chain (vm, b[0]) + rw_len0); vlib_increment_combined_counter (&adjacency_counters, thread_index, adj_index1, 1, vlib_buffer_length_in_chain (vm, b[1]) + rw_len1); } if (is_midchain) { if (adj0->sub_type.midchain.fixup_func) adj0->sub_type.midchain.fixup_func (vm, adj0, b[0], adj0->sub_type.midchain.fixup_data); if (adj1->sub_type.midchain.fixup_func) adj1->sub_type.midchain.fixup_func (vm, adj1, b[1], adj1->sub_type.midchain.fixup_data); } if (is_mcast) { /* * copy bytes from the IP address into the MAC rewrite */ vnet_ip_mcast_fixup_header (IP4_MCAST_ADDR_MASK, adj0->rewrite_header.dst_mcast_offset, &ip0->dst_address.as_u32, (u8 *) ip0); vnet_ip_mcast_fixup_header (IP4_MCAST_ADDR_MASK, adj1->rewrite_header.dst_mcast_offset, &ip1->dst_address.as_u32, (u8 *) ip1); } next += 2; b += 2; n_left_from -= 2; } while (n_left_from > 0) { ip_adjacency_t *adj0; ip4_header_t *ip0; u32 rw_len0, adj_index0, error0; u32 tx_sw_if_index0; adj_index0 = vnet_buffer (b[0])->ip.adj_index[VLIB_TX]; adj0 = adj_get (adj_index0); if (do_counters) vlib_prefetch_combined_counter (&adjacency_counters, thread_index, adj_index0); ip0 = vlib_buffer_get_current (b[0]); error0 = IP4_ERROR_NONE; ip4_ttl_and_checksum_check (b[0], ip0, next + 0, &error0); /* Update packet buffer attributes/set output interface. */ rw_len0 = adj0[0].rewrite_header.data_bytes; vnet_buffer (b[0])->ip.save_rewrite_length = rw_len0; /* Check MTU of outgoing interface. */ u16 ip0_len = clib_net_to_host_u16 (ip0->length); if (do_gso && (b[0]->flags & VNET_BUFFER_F_GSO)) ip0_len = gso_mtu_sz (b[0]); ip4_mtu_check (b[0], ip0_len, adj0[0].rewrite_header.max_l3_packet_bytes, ip0->flags_and_fragment_offset & clib_host_to_net_u16 (IP4_HEADER_FLAG_DONT_FRAGMENT), next + 0, &error0); if (is_mcast) { error0 = ((adj0[0].rewrite_header.sw_if_index == vnet_buffer (b[0])->sw_if_index[VLIB_RX]) ? IP4_ERROR_SAME_INTERFACE : error0); } /* Don't adjust the buffer for ttl issue; icmp-error node wants * to see the IP header */ if (PREDICT_TRUE (error0 == IP4_ERROR_NONE)) { u32 next_index = adj0[0].rewrite_header.next_index; b[0]->current_data -= rw_len0; b[0]->current_length += rw_len0; tx_sw_if_index0 = adj0[0].rewrite_header.sw_if_index; vnet_buffer (b[0])->sw_if_index[VLIB_TX] = tx_sw_if_index0; if (PREDICT_FALSE (adj0[0].rewrite_header.flags & VNET_REWRITE_HAS_FEATURES)) vnet_feature_arc_start (lm->output_feature_arc_index, tx_sw_if_index0, &next_index, b[0]); next[0] = next_index; } else { b[0]->error = error_node->errors[error0]; } if (is_midchain) { calc_checksums (vm, b[0]); } /* Guess we are only writing on simple Ethernet header. */ vnet_rewrite_one_header (adj0[0], ip0, sizeof (ethernet_header_t)); if (do_counters) vlib_increment_combined_counter (&adjacency_counters, thread_index, adj_index0, 1, vlib_buffer_length_in_chain (vm, b[0]) + rw_len0); if (is_midchain) { if (adj0->sub_type.midchain.fixup_func) adj0->sub_type.midchain.fixup_func (vm, adj0, b[0], adj0->sub_type.midchain.fixup_data); } if (is_mcast) { /* * copy bytes from the IP address into the MAC rewrite */ vnet_ip_mcast_fixup_header (IP4_MCAST_ADDR_MASK, adj0->rewrite_header.dst_mcast_offset, &ip0->dst_address.as_u32, (u8 *) ip0); } next += 1; b += 1; n_left_from -= 1; } /* Need to do trace after rewrites to pick up new packet data. */ if (node->flags & VLIB_NODE_FLAG_TRACE) ip4_forward_next_trace (vm, node, frame, VLIB_TX); vlib_buffer_enqueue_to_next (vm, node, from, nexts, frame->n_vectors); return frame->n_vectors; } always_inline uword ip4_rewrite_inline (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame, int do_counters, int is_midchain, int is_mcast) { vnet_main_t *vnm = vnet_get_main (); if (PREDICT_FALSE (vnm->interface_main.gso_interface_count > 0)) return ip4_rewrite_inline_with_gso (vm, node, frame, do_counters, is_midchain, is_mcast, 1 /* do_gso */ ); else return ip4_rewrite_inline_with_gso (vm, node, frame, do_counters, is_midchain, is_mcast, 0 /* no do_gso */ ); } /** @brief IPv4 rewrite node. @node ip4-rewrite This is the IPv4 transit-rewrite node: decrement TTL, fix the ipv4 header checksum, fetch the ip adjacency, check the outbound mtu, apply the adjacency rewrite, and send pkts to the adjacency rewrite header's rewrite_next_index. @param vm vlib_main_t corresponding to the current thread @param node vlib_node_runtime_t @param frame vlib_frame_t whose contents should be dispatched @par Graph mechanics: buffer metadata, next index usage @em Uses: - vnet_buffer(b)->ip.adj_index[VLIB_TX] - the rewrite adjacency index - adj->lookup_next_index - Must be IP_LOOKUP_NEXT_REWRITE or IP_LOOKUP_NEXT_ARP, otherwise the packet will be dropped. - adj->rewrite_header - Rewrite string length, rewrite string, next_index @em Sets: - b->current_data, b->current_length - Updated net of applying the rewrite string Next Indices: - adj->rewrite_header.next_index or @c ip4-drop */ VLIB_NODE_FN (ip4_rewrite_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { if (adj_are_counters_enabled ()) return ip4_rewrite_inline (vm, node, frame, 1, 0, 0); else return ip4_rewrite_inline (vm, node, frame, 0, 0, 0); } VLIB_NODE_FN (ip4_rewrite_bcast_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { if (adj_are_counters_enabled ()) return ip4_rewrite_inline (vm, node, frame, 1, 0, 0); else return ip4_rewrite_inline (vm, node, frame, 0, 0, 0); } VLIB_NODE_FN (ip4_midchain_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { if (adj_are_counters_enabled ()) return ip4_rewrite_inline (vm, node, frame, 1, 1, 0); else return ip4_rewrite_inline (vm, node, frame, 0, 1, 0); } VLIB_NODE_FN (ip4_rewrite_mcast_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { if (adj_are_counters_enabled ()) return ip4_rewrite_inline (vm, node, frame, 1, 0, 1); else return ip4_rewrite_inline (vm, node, frame, 0, 0, 1); } VLIB_NODE_FN (ip4_mcast_midchain_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) { if (adj_are_counters_enabled ()) return ip4_rewrite_inline (vm, node, frame, 1, 1, 1); else return ip4_rewrite_inline (vm, node, frame, 0, 1, 1); } /* *INDENT-OFF* */ VLIB_REGISTER_NODE (ip4_rewrite_node) = { .name = "ip4-rewrite", .vector_size = sizeof (u32), .format_trace = format_ip4_rewrite_trace, .n_next_nodes = IP4_REWRITE_N_NEXT, .next_nodes = { [IP4_REWRITE_NEXT_DROP] = "ip4-drop", [IP4_REWRITE_NEXT_ICMP_ERROR] = "ip4-icmp-error", [IP4_REWRITE_NEXT_FRAGMENT] = "ip4-frag", }, }; VLIB_REGISTER_NODE (ip4_rewrite_bcast_node) = { .name = "ip4-rewrite-bcast", .vector_size = sizeof (u32), .format_trace = format_ip4_rewrite_trace, .sibling_of = "ip4-rewrite", }; VLIB_REGISTER_NODE (ip4_rewrite_mcast_node) = { .name = "ip4-rewrite-mcast", .vector_size = sizeof (u32), .format_trace = format_ip4_rewrite_trace, .sibling_of = "ip4-rewrite", }; VLIB_REGISTER_NODE (ip4_mcast_midchain_node) = { .name = "ip4-mcast-midchain", .vector_size = sizeof (u32), .format_trace = format_ip4_rewrite_trace, .sibling_of = "ip4-rewrite", }; VLIB_REGISTER_NODE (ip4_midchain_node) = { .name = "ip4-midchain", .vector_size = sizeof (u32), .format_trace = format_ip4_forward_next_trace, .sibling_of = "ip4-rewrite", }; /* *INDENT-ON */ static int ip4_lookup_validate (ip4_address_t * a, u32 fib_index0) { ip4_fib_mtrie_t *mtrie0; ip4_fib_mtrie_leaf_t leaf0; u32 lbi0; mtrie0 = &ip4_fib_get (fib_index0)->mtrie; leaf0 = ip4_fib_mtrie_lookup_step_one (mtrie0, a); leaf0 = ip4_fib_mtrie_lookup_step (mtrie0, leaf0, a, 2); leaf0 = ip4_fib_mtrie_lookup_step (mtrie0, leaf0, a, 3); lbi0 = ip4_fib_mtrie_leaf_get_adj_index (leaf0); return lbi0 == ip4_fib_table_lookup_lb (ip4_fib_get (fib_index0), a); } static clib_error_t * test_lookup_command_fn (vlib_main_t * vm, unformat_input_t * input, vlib_cli_command_t * cmd) { ip4_fib_t *fib; u32 table_id = 0; f64 count = 1; u32 n; int i; ip4_address_t ip4_base_address; u64 errors = 0; while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT) { if (unformat (input, "table %d", &table_id)) { /* Make sure the entry exists. */ fib = ip4_fib_get (table_id); if ((fib) && (fib->index != table_id)) return clib_error_return (0, " %d does not exist", table_id); } else if (unformat (input, "count %f", &count)) ; else if (unformat (input, "%U", unformat_ip4_address, &ip4_base_address)) ; else return clib_error_return (0, "unknown input `%U'", format_unformat_error, input); } n = count; for (i = 0; i < n; i++) { if (!ip4_lookup_validate (&ip4_base_address, table_id)) errors++; ip4_base_address.as_u32 = clib_host_to_net_u32 (1 + clib_net_to_host_u32 (ip4_base_address.as_u32)); } if (errors) vlib_cli_output (vm, "%llu errors out of %d lookups\n", errors, n); else vlib_cli_output (vm, "No errors in %d lookups\n", n); return 0; } /*? * Perform a lookup of an IPv4 Address (or range of addresses) in the * given FIB table to determine if there is a conflict with the * adjacency table. The fib-id can be determined by using the * 'show ip fib' command. If fib-id is not entered, default value * of 0 is used. * * @todo This command uses fib-id, other commands use table-id (not * just a name, they are different indexes). Would like to change this * to table-id for consistency. * * @cliexpar * Example of how to run the test lookup command: * @cliexstart{test lookup 172.16.1.1 table 1 count 2} * No errors in 2 lookups * @cliexend ?*/ /* *INDENT-OFF* */ VLIB_CLI_COMMAND (lookup_test_command, static) = { .path = "test lookup", .short_help = "test lookup [table ] [count ]", .function = test_lookup_command_fn, }; /* *INDENT-ON* */ #ifndef CLIB_MARCH_VARIANT int vnet_set_ip4_flow_hash (u32 table_id, u32 flow_hash_config) { u32 fib_index; fib_index = fib_table_find (FIB_PROTOCOL_IP4, table_id); if (~0 == fib_index) return VNET_API_ERROR_NO_SUCH_FIB; fib_table_set_flow_hash_config (fib_index, FIB_PROTOCOL_IP4, flow_hash_config); return 0; } #endif static clib_error_t * set_ip_flow_hash_command_fn (vlib_main_t * vm, unformat_input_t * input, vlib_cli_command_t * cmd) { int matched = 0; u32 table_id = 0; u32 flow_hash_config = 0; int rv; while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT) { if (unformat (input, "table %d", &table_id)) matched = 1; #define _(a,v) \ else if (unformat (input, #a)) { flow_hash_config |= v; matched=1;} foreach_flow_hash_bit #undef _ else break; } if (matched == 0) return clib_error_return (0, "unknown input `%U'", format_unformat_error, input); rv = vnet_set_ip4_flow_hash (table_id, flow_hash_config); switch (rv) { case 0: break; case VNET_API_ERROR_NO_SUCH_FIB: return clib_error_return (0, "no such FIB table %d", table_id); default: clib_warning ("BUG: illegal flow hash config 0x%x", flow_hash_config); break; } return 0; } /*? * Configure the set of IPv4 fields used by the flow hash. * * @cliexpar * Example of how to set the flow hash on a given table: * @cliexcmd{set ip flow-hash table 7 dst sport dport proto} * Example of display the configured flow hash: * @cliexstart{show ip fib} * ipv4-VRF:0, fib_index 0, flow hash: src dst sport dport proto * 0.0.0.0/0 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:0 buckets:1 uRPF:0 to:[0:0]] * [0] [@0]: dpo-drop ip6 * 0.0.0.0/32 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:1 buckets:1 uRPF:1 to:[0:0]] * [0] [@0]: dpo-drop ip6 * 224.0.0.0/8 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:3 buckets:1 uRPF:3 to:[0:0]] * [0] [@0]: dpo-drop ip6 * 6.0.1.2/32 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:30 buckets:1 uRPF:29 to:[0:0]] * [0] [@3]: arp-ipv4: via 6.0.0.1 af_packet0 * 7.0.0.1/32 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:31 buckets:4 uRPF:30 to:[0:0]] * [0] [@3]: arp-ipv4: via 6.0.0.2 af_packet0 * [1] [@3]: arp-ipv4: via 6.0.0.2 af_packet0 * [2] [@3]: arp-ipv4: via 6.0.0.2 af_packet0 * [3] [@3]: arp-ipv4: via 6.0.0.1 af_packet0 * 240.0.0.0/8 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:2 buckets:1 uRPF:2 to:[0:0]] * [0] [@0]: dpo-drop ip6 * 255.255.255.255/32 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:4 buckets:1 uRPF:4 to:[0:0]] * [0] [@0]: dpo-drop ip6 * ipv4-VRF:7, fib_index 1, flow hash: dst sport dport proto * 0.0.0.0/0 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:12 buckets:1 uRPF:11 to:[0:0]] * [0] [@0]: dpo-drop ip6 * 0.0.0.0/32 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:13 buckets:1 uRPF:12 to:[0:0]] * [0] [@0]: dpo-drop ip6 * 172.16.1.0/24 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:17 buckets:1 uRPF:16 to:[0:0]] * [0] [@4]: ipv4-glean: af_packet0 * 172.16.1.1/32 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:18 buckets:1 uRPF:17 to:[1:84]] * [0] [@2]: dpo-receive: 172.16.1.1 on af_packet0 * 172.16.1.2/32 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:21 buckets:1 uRPF:20 to:[0:0]] * [0] [@5]: ipv4 via 172.16.1.2 af_packet0: IP4: 02:fe:9e:70:7a:2b -> 26:a5:f6:9c:3a:36 * 172.16.2.0/24 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:19 buckets:1 uRPF:18 to:[0:0]] * [0] [@4]: ipv4-glean: af_packet1 * 172.16.2.1/32 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:20 buckets:1 uRPF:19 to:[0:0]] * [0] [@2]: dpo-receive: 172.16.2.1 on af_packet1 * 224.0.0.0/8 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:15 buckets:1 uRPF:14 to:[0:0]] * [0] [@0]: dpo-drop ip6 * 240.0.0.0/8 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:14 buckets:1 uRPF:13 to:[0:0]] * [0] [@0]: dpo-drop ip6 * 255.255.255.255/32 * unicast-ip4-chain * [@0]: dpo-load-balance: [index:16 buckets:1 uRPF:15 to:[0:0]] * [0] [@0]: dpo-drop ip6 * @cliexend ?*/ /* *INDENT-OFF* */ VLIB_CLI_COMMAND (set_ip_flow_hash_command, static) = { .path = "set ip flow-hash", .short_help = "set ip flow-hash table [src] [dst] [sport] [dport] [proto] [reverse]", .function = set_ip_flow_hash_command_fn, }; /* *INDENT-ON* */ #ifndef CLIB_MARCH_VARIANT int vnet_set_ip4_classify_intfc (vlib_main_t * vm, u32 sw_if_index, u32 table_index) { vnet_main_t *vnm = vnet_get_main (); vnet_interface_main_t *im = &vnm->interface_main; ip4_main_t *ipm = &ip4_main; ip_lookup_main_t *lm = &ipm->lookup_main; vnet_classify_main_t *cm = &vnet_classify_main; ip4_address_t *if_addr; if (pool_is_free_index (im->sw_interfaces, sw_if_index)) return VNET_API_ERROR_NO_MATCHING_INTERFACE; if (table_index != ~0 && pool_is_free_index (cm->tables, table_index)) return VNET_API_ERROR_NO_SUCH_ENTRY; vec_validate (lm->classify_table_index_by_sw_if_index, sw_if_index); lm->classify_table_index_by_sw_if_index[sw_if_index] = table_index; if_addr = ip4_interface_first_address (ipm, sw_if_index, NULL); if (NULL != if_addr) { fib_prefix_t pfx = { .fp_len = 32, .fp_proto = FIB_PROTOCOL_IP4, .fp_addr.ip4 = *if_addr, }; u32 fib_index; fib_index = fib_table_get_index_for_sw_if_index (FIB_PROTOCOL_IP4, sw_if_index); if (table_index != (u32) ~ 0) { dpo_id_t dpo = DPO_INVALID; dpo_set (&dpo, DPO_CLASSIFY, DPO_PROTO_IP4, classify_dpo_create (DPO_PROTO_IP4, table_index)); fib_table_entry_special_dpo_add (fib_index, &pfx, FIB_SOURCE_CLASSIFY, FIB_ENTRY_FLAG_NONE, &dpo); dpo_reset (&dpo); } else { fib_table_entry_special_remove (fib_index, &pfx, FIB_SOURCE_CLASSIFY); } } return 0; } #endif static clib_error_t * set_ip_classify_command_fn (vlib_main_t * vm, unformat_input_t * input, vlib_cli_command_t * cmd) { u32 table_index = ~0; int table_index_set = 0; u32 sw_if_index = ~0; int rv; while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT) { if (unformat (input, "table-index %d", &table_index)) table_index_set = 1; else if (unformat (input, "intfc %U", unformat_vnet_sw_interface, vnet_get_main (), &sw_if_index)) ; else break; } if (table_index_set == 0) return clib_error_return (0, "classify table-index must be specified"); if (sw_if_index == ~0) return clib_error_return (0, "interface / subif must be specified"); rv = vnet_set_ip4_classify_intfc (vm, sw_if_index, table_index); switch (rv) { case 0: break; case VNET_API_ERROR_NO_MATCHING_INTERFACE: return clib_error_return (0, "No such interface"); case VNET_API_ERROR_NO_SUCH_ENTRY: return clib_error_return (0, "No such classifier table"); } return 0; } /*? * Assign a classification table to an interface. The classification * table is created using the 'classify table' and 'classify session' * commands. Once the table is create, use this command to filter packets * on an interface. * * @cliexpar * Example of how to assign a classification table to an interface: * @cliexcmd{set ip classify intfc GigabitEthernet2/0/0 table-index 1} ?*/ /* *INDENT-OFF* */ VLIB_CLI_COMMAND (set_ip_classify_command, static) = { .path = "set ip classify", .short_help = "set ip classify intfc table-index ", .function = set_ip_classify_command_fn, }; /* *INDENT-ON* */ static clib_error_t * ip4_config (vlib_main_t * vm, unformat_input_t * input) { ip4_main_t *im = &ip4_main; uword heapsize = 0; while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT) { if (unformat (input, "heap-size %U", unformat_memory_size, &heapsize)) ; else return clib_error_return (0, "invalid heap-size parameter `%U'", format_unformat_error, input); } im->mtrie_heap_size = heapsize; return 0; } VLIB_EARLY_CONFIG_FUNCTION (ip4_config, "ip"); /* * fd.io coding-style-patch-verification: ON * * Local Variables: * eval: (c-set-style "gnu") * End: */