diff options
Diffstat (limited to 'src/vnet')
84 files changed, 12715 insertions, 11986 deletions
diff --git a/src/vnet/CMakeLists.txt b/src/vnet/CMakeLists.txt index dcba3c725db..9867b7cc8c0 100644 --- a/src/vnet/CMakeLists.txt +++ b/src/vnet/CMakeLists.txt @@ -128,6 +128,7 @@ list(APPEND VNET_API_FILES cop/cop.api) # Layer 2 protocol: Ethernet ############################################################################## list(APPEND VNET_SOURCES + ethernet/arp_packet.c ethernet/ethernet_types_api.c ethernet/format.c ethernet/init.c @@ -154,6 +155,7 @@ list(APPEND VNET_HEADERS ethernet/types.def ethernet/sfp.h ethernet/p2p_ethernet.h + ethernet/arp_packet.h ) list(APPEND VNET_API_FILES @@ -166,6 +168,7 @@ list(APPEND VNET_API_FILES list(APPEND VNET_SOURCES l2/feat_bitmap.c l2/l2_api.c + l2/l2_arp_term.c l2/l2_bd.c l2/l2_bvi.c l2/l2_bvi_node.c @@ -422,6 +425,7 @@ list(APPEND VNET_SOURCES ip/icmp6.c ip/ip46_cli.c ip/ip_types_api.c + ip/ip46_address.c ip/ip4_format.c ip/ip4_forward.c ip/ip4_punt_drop.c @@ -440,12 +444,10 @@ list(APPEND VNET_SOURCES ip/ip6_punt_drop.c ip/ip6_hop_by_hop.c ip/ip6_input.c - ip/ip6_neighbor.c + ip/ip6_link.c ip/ip6_pg.c ip/reass/ip6_full_reass.c ip/reass/ip6_sv_reass.c - ip/rd_cp.c - ip/ip_neighbor.c ip/ip_api.c ip/ip_checksum.c ip/ip_frag.c @@ -486,16 +488,15 @@ list(APPEND VNET_HEADERS ip/ip4.h ip/ip4_mtrie.h ip/ip4_packet.h + ip/ip46_address.h ip/ip6_error.h ip/ip6.h ip/ip6_hop_by_hop.h ip/ip6_hop_by_hop_packet.h ip/ip6_packet.h - ip/ip6_neighbor.h ip/ip.h ip/ip_packet.h ip/ip_source_and_port_range_check.h - ip/ip_neighbor.h ip/ip_types.h ip/lookup.h ip/ports.def @@ -506,7 +507,6 @@ list(APPEND VNET_HEADERS list(APPEND VNET_API_FILES ip/ip.api - ip/rd_cp.api ip/punt.api ) @@ -517,15 +517,25 @@ list(APPEND VNET_MULTIARCH_SOURCES ) ############################################################################## -# Layer 2/3 ARP +# Layer 3 neighbours ############################################################################## list(APPEND VNET_SOURCES - ethernet/arp.c + ip-neighbor/ip_neighbor.c + ip-neighbor/ip_neighbor_api.c + ip-neighbor/ip_neighbor_dp.c + ip-neighbor/ip_neighbor_types.c + ip-neighbor/ip_neighbor_watch.c + ip-neighbor/ip4_neighbor.c + ip-neighbor/ip6_neighbor.c ) list(APPEND VNET_HEADERS - ethernet/arp_packet.h - ethernet/arp.h + ip-neighbor/ip_neighbor.h + ip-neighbor/ip_neighbor_types.h +) + +list(APPEND VNET_API_FILES + ip-neighbor/ip_neighbor.api ) ############################################################################## @@ -1513,6 +1523,41 @@ list(APPEND VNET_HEADERS list(APPEND VNET_API_FILES nhrp/nhrp.api) ############################################################################## +# ARP/ND +############################################################################## + +list (APPEND VNET_SOURCES + arp/arp_api.c + arp/arp.c + arp/arp_proxy.c +) + +list(APPEND VNET_HEADERS + arp/arp.h +) + +list(APPEND VNET_API_FILES arp/arp.api) + +list (APPEND VNET_SOURCES + ip6-nd/ip6_mld.c + ip6-nd/ip6_nd.c + ip6-nd/ip6_nd_api.c + ip6-nd/ip6_nd_proxy.c + ip6-nd/ip6_ra.c + ip6-nd/rd_cp.c + ip6-nd/rd_cp_api.c +) + +list(APPEND VNET_HEADERS + ip6-nd/ip6_nd.h +) + +list(APPEND VNET_API_FILES + ip6-nd/ip6_nd.api + ip6-nd/rd_cp.api +) + +############################################################################## # VNET Library ############################################################################## diff --git a/src/vnet/adj/adj.h b/src/vnet/adj/adj.h index cd6e8382ba5..1a841ed2892 100644 --- a/src/vnet/adj/adj.h +++ b/src/vnet/adj/adj.h @@ -326,6 +326,10 @@ typedef struct ip_adjacency_t_ */ struct adj_delegate_t_ *ia_delegates; + /** + * The VLIB node in which this adj is used to forward packets + */ + u32 ia_node_index; } ip_adjacency_t; STATIC_ASSERT ((STRUCT_OFFSET_OF (ip_adjacency_t, cacheline0) == 0), diff --git a/src/vnet/adj/adj_glean.c b/src/vnet/adj/adj_glean.c index a8a422d30ba..338d7320d53 100644 --- a/src/vnet/adj/adj_glean.c +++ b/src/vnet/adj/adj_glean.c @@ -24,19 +24,19 @@ */ static adj_index_t *adj_gleans[FIB_PROTOCOL_MAX]; -static inline vlib_node_registration_t* +static inline u32 adj_get_glean_node (fib_protocol_t proto) { switch (proto) { case FIB_PROTOCOL_IP4: - return (&ip4_glean_node); + return (ip4_glean_node.index); case FIB_PROTOCOL_IP6: - return (&ip6_glean_node); + return (ip6_glean_node.index); case FIB_PROTOCOL_MPLS: break; } ASSERT(0); - return (NULL); + return (~0); } /* @@ -63,6 +63,7 @@ adj_glean_add_or_lock (fib_protocol_t proto, adj->lookup_next_index = IP_LOOKUP_NEXT_GLEAN; adj->ia_nh_proto = proto; adj->ia_link = linkt; + adj->ia_node_index = adj_get_glean_node(proto); adj_gleans[proto][sw_if_index] = adj_get_index(adj); if (NULL != nh_addr) @@ -111,7 +112,7 @@ adj_glean_update_rewrite (adj_index_t adj_index) vnet_rewrite_for_sw_interface(vnet_get_main(), adj_fib_proto_2_nd(adj->ia_nh_proto), adj->rewrite_header.sw_if_index, - adj_get_glean_node(adj->ia_nh_proto)->index, + adj->ia_node_index, VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST, &adj->rewrite_header, sizeof (adj->rewrite_data)); diff --git a/src/vnet/adj/adj_internal.h b/src/vnet/adj/adj_internal.h index 2053c69342a..adb7a32375d 100644 --- a/src/vnet/adj/adj_internal.h +++ b/src/vnet/adj/adj_internal.h @@ -98,7 +98,7 @@ adj_proto_to_46 (fib_protocol_t proto) * Get a pointer to an adjacency object from its index */ static inline adj_index_t -adj_get_index (ip_adjacency_t *adj) +adj_get_index (const ip_adjacency_t *adj) { return (adj - adj_pool); } diff --git a/src/vnet/adj/adj_mcast.c b/src/vnet/adj/adj_mcast.c index 4d7172cdb58..590652244e6 100644 --- a/src/vnet/adj/adj_mcast.c +++ b/src/vnet/adj/adj_mcast.c @@ -66,11 +66,12 @@ adj_mcast_add_or_lock (fib_protocol_t proto, adj->lookup_next_index = IP_LOOKUP_NEXT_MCAST; adj->ia_nh_proto = proto; adj->ia_link = link_type; + adj->ia_node_index = adj_get_mcast_node(proto); adj_mcasts[proto][sw_if_index] = adj_get_index(adj); adj_lock(adj_get_index(adj)); vnet_rewrite_init(vnm, sw_if_index, link_type, - adj_get_mcast_node(proto), + adj->ia_node_index, vnet_tx_node_index_for_sw_interface(vnm, sw_if_index), &adj->rewrite_header); diff --git a/src/vnet/adj/adj_nbr.c b/src/vnet/adj/adj_nbr.c index 28ee53e7d49..f769c56d8ec 100644 --- a/src/vnet/adj/adj_nbr.c +++ b/src/vnet/adj/adj_nbr.c @@ -464,6 +464,7 @@ adj_nbr_update_rewrite_internal (ip_adjacency_t *adj, vlib_worker_thread_barrier_sync(vm); adj->lookup_next_index = adj_next_index; + adj->ia_node_index = this_node; if (NULL != rewrite) { @@ -666,15 +667,17 @@ adj_nbr_walk_nh (u32 sw_if_index, if (!ADJ_NBR_ITF_OK(adj_nh_proto, sw_if_index)) return; - vnet_link_t linkt; - adj_index_t ai; - - FOR_EACH_VNET_LINK(linkt) + switch (adj_nh_proto) { - ai = adj_nbr_find (adj_nh_proto, linkt, nh, sw_if_index); - - if (INDEX_INVALID != ai) - cb(ai, ctx); + case FIB_PROTOCOL_IP4: + adj_nbr_walk_nh4(sw_if_index, &nh->ip4, cb, ctx); + break; + case FIB_PROTOCOL_IP6: + adj_nbr_walk_nh6(sw_if_index, &nh->ip6, cb, ctx); + break; + case FIB_PROTOCOL_MPLS: + ASSERT(0); + break; } } diff --git a/src/vnet/api_errno.h b/src/vnet/api_errno.h index 2fbedf2eed6..9e261d73763 100644 --- a/src/vnet/api_errno.h +++ b/src/vnet/api_errno.h @@ -152,7 +152,8 @@ _(KEY_LENGTH, -156, "invalid Key Length") \ _(FIB_PATH_UNSUPPORTED_NH_PROTO, -157, "Unsupported FIB Path protocol") \ _(API_ENDIAN_FAILED, -159, "Endian mismatch detected") \ _(NO_CHANGE, -160, "No change in table") \ -_(MISSING_CERT_KEY, -161, "Missing certifcate or key") +_(MISSING_CERT_KEY, -161, "Missing certifcate or key") \ +_(LIMIT_EXCEEDED, -162, "limit exceeded") typedef enum { diff --git a/src/vnet/arp/arp.api b/src/vnet/arp/arp.api new file mode 100644 index 00000000000..27bfa3b65c6 --- /dev/null +++ b/src/vnet/arp/arp.api @@ -0,0 +1,100 @@ +/* Hey Emacs use -*- mode: C -*- */ +/* + * Copyright (c) 2019 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. + */ + +/** \file + This file defines the vpp control-plane API messages + used to control the ABF plugin +*/ + +option version = "1.0.0"; + +import "vnet/ip/ip_types.api"; +import "vnet/ethernet/ethernet_types.api"; +import "vnet/interface_types.api"; + +/** \brief Proxy ARP configuration type + @param table_id - VRF / Fib table ID + @param low - Low address of the Proxy ARP range + @param hi - High address of the Proxy ARP range +*/ +typedef proxy_arp +{ + u32 table_id; + vl_api_ip4_address_t low; + vl_api_ip4_address_t hi; +}; + +/** \brief Proxy ARP add / del request + @param client_index - opaque cookie to identify the sender + @param context - sender context, to match reply w/ request + @param is_add - 1 if adding the Proxy ARP range, 0 if deleting + @param proxy - Proxy configuration +*/ +autoreply define proxy_arp_add_del +{ + u32 client_index; + u32 context; + bool is_add; + vl_api_proxy_arp_t proxy; +}; + +/** \brief Proxy ARP dump request + */ +define proxy_arp_dump +{ + u32 client_index; + u32 context; +}; + +/** \brief Proxy ARP dump details reply + * @param proxy - Same data as used to configure + */ +define proxy_arp_details +{ + u32 context; + vl_api_proxy_arp_t proxy; +}; + +/** \brief Proxy ARP add / del interface request + @param client_index - opaque cookie to identify the sender + @param context - sender context, to match reply w/ request + @param sw_if_index - Which interface to enable / disable Proxy Arp on + @param enable - 1 to enable Proxy ARP on interface, 0 to disable +*/ +autoreply define proxy_arp_intfc_enable_disable +{ + u32 client_index; + u32 context; + vl_api_interface_index_t sw_if_index; + bool enable; +}; + +/** \brief Proxy ARP interface dump request + */ +define proxy_arp_intfc_dump +{ + u32 client_index; + u32 context; +}; + +/** \brief Proxy ARP interface dump details reply + * @param sw_if_index The interface on which ARP proxy is enabled. + */ +define proxy_arp_intfc_details +{ + u32 context; + u32 sw_if_index; +}; diff --git a/src/vnet/arp/arp.c b/src/vnet/arp/arp.c new file mode 100644 index 00000000000..14a1ae97d1e --- /dev/null +++ b/src/vnet/arp/arp.c @@ -0,0 +1,934 @@ +/* + * ethernet/arp.c: IP v4 ARP node + * + * Copyright (c) 2010 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/arp/arp.h> +#include <vnet/arp/arp_packet.h> + +#include <vnet/fib/ip4_fib.h> +#include <vnet/fib/fib_entry_src.h> +#include <vnet/adj/adj_nbr.h> +#include <vnet/adj/adj_mcast.h> + +#include <vnet/ip-neighbor/ip_neighbor.h> +#include <vnet/ip-neighbor/ip_neighbor_dp.h> + +#include <vlibmemory/api.h> + +/** + * @file + * @brief IPv4 ARP. + * + * This file contains code to manage the IPv4 ARP tables (IP Address + * to MAC Address lookup). + */ + +/** + * @brief Per-interface ARP configuration and state + */ +typedef struct ethernet_arp_interface_t_ +{ + /** + * Is ARP enabled on this interface + */ + u32 enabled; +} ethernet_arp_interface_t; + +typedef struct +{ + /* Hash tables mapping name to opcode. */ + uword *opcode_by_name; + + /** Per interface state */ + ethernet_arp_interface_t *ethernet_arp_by_sw_if_index; + + /* ARP feature arc index */ + u8 feature_arc_index; +} ethernet_arp_main_t; + +static ethernet_arp_main_t ethernet_arp_main; + +static const u8 vrrp_prefix[] = { 0x00, 0x00, 0x5E, 0x00, 0x01 }; + +static uword +unformat_ethernet_arp_opcode_host_byte_order (unformat_input_t * input, + va_list * args) +{ + int *result = va_arg (*args, int *); + ethernet_arp_main_t *am = ðernet_arp_main; + int x, i; + + /* Numeric opcode. */ + if (unformat (input, "0x%x", &x) || unformat (input, "%d", &x)) + { + if (x >= (1 << 16)) + return 0; + *result = x; + return 1; + } + + /* Named type. */ + if (unformat_user (input, unformat_vlib_number_by_name, + am->opcode_by_name, &i)) + { + *result = i; + return 1; + } + + return 0; +} + +static uword +unformat_ethernet_arp_opcode_net_byte_order (unformat_input_t * input, + va_list * args) +{ + int *result = va_arg (*args, int *); + if (!unformat_user + (input, unformat_ethernet_arp_opcode_host_byte_order, result)) + return 0; + + *result = clib_host_to_net_u16 ((u16) * result); + return 1; +} + +typedef struct +{ + u8 packet_data[64]; +} ethernet_arp_input_trace_t; + +static u8 * +format_ethernet_arp_input_trace (u8 * s, va_list * va) +{ + CLIB_UNUSED (vlib_main_t * vm) = va_arg (*va, vlib_main_t *); + CLIB_UNUSED (vlib_node_t * node) = va_arg (*va, vlib_node_t *); + ethernet_arp_input_trace_t *t = va_arg (*va, ethernet_arp_input_trace_t *); + + s = format (s, "%U", + format_ethernet_arp_header, + t->packet_data, sizeof (t->packet_data)); + + return s; +} + +static int +arp_is_enabled (ethernet_arp_main_t * am, u32 sw_if_index) +{ + if (vec_len (am->ethernet_arp_by_sw_if_index) <= sw_if_index) + return 0; + + return (am->ethernet_arp_by_sw_if_index[sw_if_index].enabled); +} + +static void +arp_enable (ethernet_arp_main_t * am, u32 sw_if_index) +{ + if (arp_is_enabled (am, sw_if_index)) + return; + + vec_validate (am->ethernet_arp_by_sw_if_index, sw_if_index); + + am->ethernet_arp_by_sw_if_index[sw_if_index].enabled = 1; + + vnet_feature_enable_disable ("arp", "arp-reply", sw_if_index, 1, NULL, 0); + vnet_feature_enable_disable ("arp", "arp-disabled", sw_if_index, 0, NULL, + 0); +} + +static void +arp_disable (ethernet_arp_main_t * am, u32 sw_if_index) +{ + if (!arp_is_enabled (am, sw_if_index)) + return; + + vnet_feature_enable_disable ("arp", "arp-disabled", sw_if_index, 1, NULL, + 0); + vnet_feature_enable_disable ("arp", "arp-reply", sw_if_index, 0, NULL, 0); + + am->ethernet_arp_by_sw_if_index[sw_if_index].enabled = 0; +} + +static int +arp_unnumbered (vlib_buffer_t * p0, + u32 input_sw_if_index, u32 conn_sw_if_index) +{ + vnet_main_t *vnm = vnet_get_main (); + vnet_interface_main_t *vim = &vnm->interface_main; + vnet_sw_interface_t *si; + + /* verify that the input interface is unnumbered to the connected. + * the connected interface is the interface on which the subnet is + * configured */ + si = &vim->sw_interfaces[input_sw_if_index]; + + if (!(si->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED && + (si->unnumbered_sw_if_index == conn_sw_if_index))) + { + /* the input interface is not unnumbered to the interface on which + * the sub-net is configured that covers the ARP request. + * So this is not the case for unnumbered.. */ + return 0; + } + + return !0; +} + +always_inline u32 +arp_learn (u32 sw_if_index, + const ethernet_arp_ip4_over_ethernet_address_t * addr) +{ + ip_neighbor_learn_t l = { + .ip.ip4 = addr->ip4, + .type = IP46_TYPE_IP4, + .mac = addr->mac, + .sw_if_index = sw_if_index, + }; + + ip_neighbor_learn_dp (&l); + + return (ETHERNET_ARP_ERROR_l3_src_address_learned); +} + +typedef enum arp_input_next_t_ +{ + ARP_INPUT_NEXT_DROP, + ARP_INPUT_NEXT_DISABLED, + ARP_INPUT_N_NEXT, +} arp_input_next_t; + +static uword +arp_input (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + u32 n_left_from, next_index, *from, *to_next, n_left_to_next; + ethernet_arp_main_t *am = ðernet_arp_main; + + from = vlib_frame_vector_args (frame); + n_left_from = frame->n_vectors; + next_index = node->cached_next_index; + + if (node->flags & VLIB_NODE_FLAG_TRACE) + vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors, + /* stride */ 1, + sizeof (ethernet_arp_input_trace_t)); + + 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) + { + const ethernet_arp_header_t *arp0; + arp_input_next_t next0; + vlib_buffer_t *p0; + u32 pi0, error0; + + pi0 = to_next[0] = from[0]; + from += 1; + to_next += 1; + n_left_from -= 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, pi0); + arp0 = vlib_buffer_get_current (p0); + + error0 = ETHERNET_ARP_ERROR_replies_sent; + next0 = ARP_INPUT_NEXT_DROP; + + error0 = + (arp0->l2_type != + clib_net_to_host_u16 (ETHERNET_ARP_HARDWARE_TYPE_ethernet) ? + ETHERNET_ARP_ERROR_l2_type_not_ethernet : error0); + error0 = + (arp0->l3_type != + clib_net_to_host_u16 (ETHERNET_TYPE_IP4) ? + ETHERNET_ARP_ERROR_l3_type_not_ip4 : error0); + error0 = + (0 == arp0->ip4_over_ethernet[0].ip4.as_u32 ? + ETHERNET_ARP_ERROR_l3_dst_address_unset : error0); + + if (ETHERNET_ARP_ERROR_replies_sent == error0) + { + next0 = ARP_INPUT_NEXT_DISABLED; + vnet_feature_arc_start (am->feature_arc_index, + vnet_buffer (p0)->sw_if_index[VLIB_RX], + &next0, p0); + } + else + p0->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; +} + +typedef enum arp_disabled_next_t_ +{ + ARP_DISABLED_NEXT_DROP, + ARP_DISABLED_N_NEXT, +} arp_disabled_next_t; + +#define foreach_arp_disabled_error \ + _ (DISABLED, "ARP Disabled on this interface") \ + +typedef enum +{ +#define _(sym,string) ARP_DISABLED_ERROR_##sym, + foreach_arp_disabled_error +#undef _ + ARP_DISABLED_N_ERROR, +} arp_disabled_error_t; + +static char *arp_disabled_error_strings[] = { +#define _(sym,string) string, + foreach_arp_disabled_error +#undef _ +}; + +static uword +arp_disabled (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + u32 n_left_from, next_index, *from, *to_next, n_left_to_next; + + from = vlib_frame_vector_args (frame); + n_left_from = frame->n_vectors; + next_index = node->cached_next_index; + + if (node->flags & VLIB_NODE_FLAG_TRACE) + vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors, + /* stride */ 1, + sizeof (ethernet_arp_input_trace_t)); + + 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) + { + arp_disabled_next_t next0 = ARP_DISABLED_NEXT_DROP; + vlib_buffer_t *p0; + u32 pi0, error0; + + next0 = ARP_DISABLED_NEXT_DROP; + error0 = ARP_DISABLED_ERROR_DISABLED; + + pi0 = to_next[0] = from[0]; + from += 1; + to_next += 1; + n_left_from -= 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, pi0); + p0->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; +} + +enum arp_dst_fib_type +{ + ARP_DST_FIB_NONE, + ARP_DST_FIB_ADJ, + ARP_DST_FIB_CONN +}; + +/* + * we're looking for FIB sources that indicate the destination + * is attached. There may be interposed DPO prior to the one + * we are looking for + */ +static enum arp_dst_fib_type +arp_dst_fib_check (const fib_node_index_t fei, fib_entry_flag_t * flags) +{ + const fib_entry_t *entry = fib_entry_get (fei); + const fib_entry_src_t *entry_src; + fib_source_t src; + /* *INDENT-OFF* */ + FOR_EACH_SRC_ADDED(entry, entry_src, src, + ({ + *flags = fib_entry_get_flags_for_source (fei, src); + if (fib_entry_is_sourced (fei, FIB_SOURCE_ADJ)) + return ARP_DST_FIB_ADJ; + else if (FIB_ENTRY_FLAG_CONNECTED & *flags) + return ARP_DST_FIB_CONN; + })) + /* *INDENT-ON* */ + + return ARP_DST_FIB_NONE; +} + +static uword +arp_reply (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + vnet_main_t *vnm = vnet_get_main (); + u32 n_left_from, next_index, *from, *to_next; + u32 n_replies_sent = 0; + + from = vlib_frame_vector_args (frame); + n_left_from = frame->n_vectors; + next_index = node->cached_next_index; + + if (node->flags & VLIB_NODE_FLAG_TRACE) + vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors, + /* stride */ 1, + sizeof (ethernet_arp_input_trace_t)); + + while (n_left_from > 0) + { + u32 n_left_to_next; + + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); + + while (n_left_from > 0 && n_left_to_next > 0) + { + vlib_buffer_t *p0; + ethernet_arp_header_t *arp0; + ethernet_header_t *eth_rx; + const ip4_address_t *if_addr0; + u32 pi0, error0, next0, sw_if_index0, conn_sw_if_index0, fib_index0; + u8 dst_is_local0, is_vrrp_reply0; + fib_node_index_t dst_fei, src_fei; + const fib_prefix_t *pfx0; + fib_entry_flag_t src_flags, dst_flags; + + pi0 = from[0]; + to_next[0] = pi0; + from += 1; + to_next += 1; + n_left_from -= 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, pi0); + arp0 = vlib_buffer_get_current (p0); + /* Fill in ethernet header. */ + eth_rx = ethernet_buffer_get_header (p0); + + next0 = ARP_REPLY_NEXT_DROP; + error0 = ETHERNET_ARP_ERROR_replies_sent; + sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX]; + + /* Check that IP address is local and matches incoming interface. */ + fib_index0 = ip4_fib_table_get_index_for_sw_if_index (sw_if_index0); + if (~0 == fib_index0) + { + error0 = ETHERNET_ARP_ERROR_interface_no_table; + goto drop; + + } + + { + /* + * we're looking for FIB entries that indicate the source + * is attached. There may be more specific non-attached + * routes that match the source, but these do not influence + * whether we respond to an ARP request, i.e. they do not + * influence whether we are the correct way for the sender + * to reach us, they only affect how we reach the sender. + */ + fib_entry_t *src_fib_entry; + const fib_prefix_t *pfx; + fib_entry_src_t *src; + fib_source_t source; + int attached; + int mask; + + mask = 32; + attached = 0; + + do + { + src_fei = ip4_fib_table_lookup (ip4_fib_get (fib_index0), + &arp0-> + ip4_over_ethernet[0].ip4, + mask); + src_fib_entry = fib_entry_get (src_fei); + + /* + * It's possible that the source that provides the + * flags we need, or the flags we must not have, + * is not the best source, so check then all. + */ + /* *INDENT-OFF* */ + FOR_EACH_SRC_ADDED(src_fib_entry, src, source, + ({ + src_flags = fib_entry_get_flags_for_source (src_fei, source); + + /* Reject requests/replies with our local interface + address. */ + if (FIB_ENTRY_FLAG_LOCAL & src_flags) + { + error0 = ETHERNET_ARP_ERROR_l3_src_address_is_local; + /* + * When VPP has an interface whose address is also + * applied to a TAP interface on the host, then VPP's + * TAP interface will be unnumbered to the 'real' + * interface and do proxy ARP from the host. + * The curious aspect of this setup is that ARP requests + * from the host will come from the VPP's own address. + * So don't drop immediately here, instead go see if this + * is a proxy ARP case. + */ + goto next_feature; + } + /* A Source must also be local to subnet of matching + * interface address. */ + if ((FIB_ENTRY_FLAG_ATTACHED & src_flags) || + (FIB_ENTRY_FLAG_CONNECTED & src_flags)) + { + attached = 1; + break; + } + /* + * else + * The packet was sent from an address that is not + * connected nor attached i.e. it is not from an + * address that is covered by a link's sub-net, + * nor is it a already learned host resp. + */ + })); + /* *INDENT-ON* */ + + /* + * shorter mask lookup for the next iteration. + */ + pfx = fib_entry_get_prefix (src_fei); + mask = pfx->fp_len - 1; + + /* + * continue until we hit the default route or we find + * the attached we are looking for. The most likely + * outcome is we find the attached with the first source + * on the first lookup. + */ + } + while (!attached && + !fib_entry_is_sourced (src_fei, FIB_SOURCE_DEFAULT_ROUTE)); + + if (!attached) + { + /* + * the matching route is a not attached, i.e. it was + * added as a result of routing, rather than interface/ARP + * configuration. If the matching route is not a host route + * (i.e. a /32) + */ + error0 = ETHERNET_ARP_ERROR_l3_src_address_not_local; + goto drop; + } + } + + dst_fei = ip4_fib_table_lookup (ip4_fib_get (fib_index0), + &arp0->ip4_over_ethernet[1].ip4, + 32); + switch (arp_dst_fib_check (dst_fei, &dst_flags)) + { + case ARP_DST_FIB_ADJ: + /* + * We matched an adj-fib on ths source subnet (a /32 previously + * added as a result of ARP). If this request is a gratuitous + * ARP, then learn from it. + * The check for matching an adj-fib, is to prevent hosts + * from spamming us with gratuitous ARPS that might otherwise + * blow our ARP cache + */ + if (arp0->ip4_over_ethernet[0].ip4.as_u32 == + arp0->ip4_over_ethernet[1].ip4.as_u32) + error0 = + arp_learn (sw_if_index0, &arp0->ip4_over_ethernet[0]); + goto drop; + case ARP_DST_FIB_CONN: + /* destination is connected, continue to process */ + break; + case ARP_DST_FIB_NONE: + /* destination is not connected, stop here */ + error0 = ETHERNET_ARP_ERROR_l3_dst_address_not_local; + goto next_feature; + } + + dst_is_local0 = (FIB_ENTRY_FLAG_LOCAL & dst_flags); + pfx0 = fib_entry_get_prefix (dst_fei); + if_addr0 = &pfx0->fp_addr.ip4; + + is_vrrp_reply0 = + ((arp0->opcode == + clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_reply)) + && + (!memcmp + (arp0->ip4_over_ethernet[0].mac.bytes, vrrp_prefix, + sizeof (vrrp_prefix)))); + + /* Trash ARP packets whose ARP-level source addresses do not + match their L2-frame-level source addresses, unless it's + a reply from a VRRP virtual router */ + if (!ethernet_mac_address_equal + (eth_rx->src_address, + arp0->ip4_over_ethernet[0].mac.bytes) && !is_vrrp_reply0) + { + error0 = ETHERNET_ARP_ERROR_l2_address_mismatch; + goto drop; + } + + /* Learn or update sender's mapping only for replies to addresses + * that are local to the subnet */ + if (arp0->opcode == + clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_reply)) + { + if (dst_is_local0) + error0 = + arp_learn (sw_if_index0, &arp0->ip4_over_ethernet[0]); + else + /* a reply for a non-local destination could be a GARP. + * GARPs for hosts we know were handled above, so this one + * we drop */ + error0 = ETHERNET_ARP_ERROR_l3_dst_address_not_local; + + goto next_feature; + } + else if (arp0->opcode == + clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_request) && + (dst_is_local0 == 0)) + { + goto next_feature; + } + + /* Honor unnumbered interface, if any */ + conn_sw_if_index0 = fib_entry_get_resolving_interface (dst_fei); + if (sw_if_index0 != conn_sw_if_index0 || + sw_if_index0 != fib_entry_get_resolving_interface (src_fei)) + { + /* + * The interface the ARP is sent to or was received on is not the + * interface on which the covering prefix is configured. + * Maybe this is a case for unnumbered. + */ + if (!arp_unnumbered (p0, sw_if_index0, conn_sw_if_index0)) + { + error0 = ETHERNET_ARP_ERROR_unnumbered_mismatch; + goto drop; + } + } + if (arp0->ip4_over_ethernet[0].ip4.as_u32 == + arp0->ip4_over_ethernet[1].ip4.as_u32) + { + error0 = ETHERNET_ARP_ERROR_gratuitous_arp; + goto drop; + } + + next0 = arp_mk_reply (vnm, p0, sw_if_index0, + if_addr0, arp0, eth_rx); + + /* We are going to reply to this request, so, in the absence of + errors, learn the sender */ + if (!error0) + error0 = arp_learn (sw_if_index0, &arp0->ip4_over_ethernet[1]); + + n_replies_sent += 1; + goto enqueue; + + next_feature: + vnet_feature_next (&next0, p0); + goto enqueue; + + drop: + p0->error = node->errors[error0]; + + enqueue: + 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); + } + + vlib_error_count (vm, node->node_index, + ETHERNET_ARP_ERROR_replies_sent, n_replies_sent); + + return frame->n_vectors; +} + + +static char *ethernet_arp_error_strings[] = { +#define _(sym,string) string, + foreach_ethernet_arp_error +#undef _ +}; + +/* *INDENT-OFF* */ + +VLIB_REGISTER_NODE (arp_input_node, static) = +{ + .function = arp_input, + .name = "arp-input", + .vector_size = sizeof (u32), + .n_errors = ETHERNET_ARP_N_ERROR, + .error_strings = ethernet_arp_error_strings, + .n_next_nodes = ARP_INPUT_N_NEXT, + .next_nodes = { + [ARP_INPUT_NEXT_DROP] = "error-drop", + [ARP_INPUT_NEXT_DISABLED] = "arp-disabled", + }, + .format_buffer = format_ethernet_arp_header, + .format_trace = format_ethernet_arp_input_trace, +}; + +VLIB_REGISTER_NODE (arp_disabled_node, static) = +{ + .function = arp_disabled, + .name = "arp-disabled", + .vector_size = sizeof (u32), + .n_errors = ARP_DISABLED_N_ERROR, + .error_strings = arp_disabled_error_strings, + .n_next_nodes = ARP_DISABLED_N_NEXT, + .next_nodes = { + [ARP_INPUT_NEXT_DROP] = "error-drop", + }, + .format_buffer = format_ethernet_arp_header, + .format_trace = format_ethernet_arp_input_trace, +}; + +VLIB_REGISTER_NODE (arp_reply_node, static) = +{ + .function = arp_reply, + .name = "arp-reply", + .vector_size = sizeof (u32), + .n_errors = ETHERNET_ARP_N_ERROR, + .error_strings = ethernet_arp_error_strings, + .n_next_nodes = ARP_REPLY_N_NEXT, + .next_nodes = { + [ARP_REPLY_NEXT_DROP] = "error-drop", + [ARP_REPLY_NEXT_REPLY_TX] = "interface-output", + }, + .format_buffer = format_ethernet_arp_header, + .format_trace = format_ethernet_arp_input_trace, +}; + +/* Built-in ARP rx feature path definition */ +VNET_FEATURE_ARC_INIT (arp_feat, static) = +{ + .arc_name = "arp", + .start_nodes = VNET_FEATURES ("arp-input"), + .last_in_arc = "error-drop", + .arc_index_ptr = ðernet_arp_main.feature_arc_index, +}; + +VNET_FEATURE_INIT (arp_reply_feat_node, static) = +{ + .arc_name = "arp", + .node_name = "arp-reply", + .runs_before = VNET_FEATURES ("arp-disabled"), +}; + +VNET_FEATURE_INIT (arp_proxy_feat_node, static) = +{ + .arc_name = "arp", + .node_name = "arp-proxy", + .runs_after = VNET_FEATURES ("arp-reply"), + .runs_before = VNET_FEATURES ("arp-disabled"), +}; + +VNET_FEATURE_INIT (arp_disabled_feat_node, static) = +{ + .arc_name = "arp", + .node_name = "arp-disabled", + .runs_before = VNET_FEATURES ("error-drop"), +}; + +VNET_FEATURE_INIT (arp_drop_feat_node, static) = +{ + .arc_name = "arp", + .node_name = "error-drop", + .runs_before = 0, /* last feature */ +}; + +/* *INDENT-ON* */ + +typedef struct +{ + pg_edit_t l2_type, l3_type; + pg_edit_t n_l2_address_bytes, n_l3_address_bytes; + pg_edit_t opcode; + struct + { + pg_edit_t mac; + pg_edit_t ip4; + } ip4_over_ethernet[2]; +} pg_ethernet_arp_header_t; + +static inline void +pg_ethernet_arp_header_init (pg_ethernet_arp_header_t * p) +{ + /* Initialize fields that are not bit fields in the IP header. */ +#define _(f) pg_edit_init (&p->f, ethernet_arp_header_t, f); + _(l2_type); + _(l3_type); + _(n_l2_address_bytes); + _(n_l3_address_bytes); + _(opcode); + _(ip4_over_ethernet[0].mac); + _(ip4_over_ethernet[0].ip4); + _(ip4_over_ethernet[1].mac); + _(ip4_over_ethernet[1].ip4); +#undef _ +} + +uword +unformat_pg_arp_header (unformat_input_t * input, va_list * args) +{ + pg_stream_t *s = va_arg (*args, pg_stream_t *); + pg_ethernet_arp_header_t *p; + u32 group_index; + + p = pg_create_edit_group (s, sizeof (p[0]), sizeof (ethernet_arp_header_t), + &group_index); + pg_ethernet_arp_header_init (p); + + /* Defaults. */ + pg_edit_set_fixed (&p->l2_type, ETHERNET_ARP_HARDWARE_TYPE_ethernet); + pg_edit_set_fixed (&p->l3_type, ETHERNET_TYPE_IP4); + pg_edit_set_fixed (&p->n_l2_address_bytes, 6); + pg_edit_set_fixed (&p->n_l3_address_bytes, 4); + + if (!unformat (input, "%U: %U/%U -> %U/%U", + unformat_pg_edit, + unformat_ethernet_arp_opcode_net_byte_order, &p->opcode, + unformat_pg_edit, + unformat_mac_address_t, &p->ip4_over_ethernet[0].mac, + unformat_pg_edit, + unformat_ip4_address, &p->ip4_over_ethernet[0].ip4, + unformat_pg_edit, + unformat_mac_address_t, &p->ip4_over_ethernet[1].mac, + unformat_pg_edit, + unformat_ip4_address, &p->ip4_over_ethernet[1].ip4)) + { + /* Free up any edits we may have added. */ + pg_free_edit_group (s); + return 0; + } + return 1; +} + +/* + * callback when an interface address is added or deleted + */ +static void +arp_enable_disable_interface (ip4_main_t * im, + uword opaque, u32 sw_if_index, u32 is_enable) +{ + ethernet_arp_main_t *am = ðernet_arp_main; + + if (is_enable) + arp_enable (am, sw_if_index); + else + arp_disable (am, sw_if_index); +} + +/* + * Remove any arp entries associated with the specified interface + */ +static clib_error_t * +vnet_arp_add_del_sw_interface (vnet_main_t * vnm, u32 sw_if_index, u32 is_add) +{ + ethernet_arp_main_t *am = ðernet_arp_main; + + if (!is_add && sw_if_index != ~0) + { + arp_disable (am, sw_if_index); + } + else if (is_add) + { + vnet_feature_enable_disable ("arp", "arp-disabled", + sw_if_index, 1, NULL, 0); + } + + return (NULL); +} + +VNET_SW_INTERFACE_ADD_DEL_FUNCTION (vnet_arp_add_del_sw_interface); + +const static ip_neighbor_vft_t arp_vft = { + .inv_proxy4_add = arp_proxy_add, + .inv_proxy4_del = arp_proxy_del, + .inv_proxy4_enable = arp_proxy_disable, + .inv_proxy4_disable = arp_proxy_disable, +}; + +static clib_error_t * +ethernet_arp_init (vlib_main_t * vm) +{ + ethernet_arp_main_t *am = ðernet_arp_main; + ip4_main_t *im = &ip4_main; + pg_node_t *pn; + + ethernet_register_input_type (vm, ETHERNET_TYPE_ARP, arp_input_node.index); + + pn = pg_get_node (arp_input_node.index); + pn->unformat_edit = unformat_pg_arp_header; + + am->opcode_by_name = hash_create_string (0, sizeof (uword)); +#define _(o) hash_set_mem (am->opcode_by_name, #o, ETHERNET_ARP_OPCODE_##o); + foreach_ethernet_arp_opcode; +#undef _ + + /* don't trace ARP error packets */ + { + vlib_node_runtime_t *rt = + vlib_node_get_runtime (vm, arp_input_node.index); + +#define _(a,b) \ + vnet_pcap_drop_trace_filter_add_del \ + (rt->errors[ETHERNET_ARP_ERROR_##a], \ + 1 /* is_add */); + foreach_ethernet_arp_error +#undef _ + } + + { + ip4_enable_disable_interface_callback_t cb = { + .function = arp_enable_disable_interface, + }; + vec_add1 (im->enable_disable_interface_callbacks, cb); + } + + ip_neighbor_register (IP46_TYPE_IP4, &arp_vft); + + return 0; +} + +/* *INDENT-OFF* */ +VLIB_INIT_FUNCTION (ethernet_arp_init) = +{ + .runs_after = VLIB_INITS("ethernet_init", + "ip_neighbor_init"), +}; +/* *INDENT-ON* */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/arp/arp.h b/src/vnet/arp/arp.h new file mode 100644 index 00000000000..7446564b0cf --- /dev/null +++ b/src/vnet/arp/arp.h @@ -0,0 +1,85 @@ +/* + * 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. + */ + +#ifndef __ARP_H__ +#define __ARP_H__ + +#include <vnet/ethernet/ethernet.h> +#include <vnet/ip/ip.h> +#include <vnet/ethernet/arp_packet.h> + +#define foreach_ethernet_arp_error \ + _ (replies_sent, "ARP replies sent") \ + _ (l2_type_not_ethernet, "L2 type not ethernet") \ + _ (l3_type_not_ip4, "L3 type not IP4") \ + _ (l3_src_address_not_local, "IP4 source address not local to subnet") \ + _ (l3_dst_address_not_local, "IP4 destination address not local to subnet") \ + _ (l3_dst_address_unset, "IP4 destination address is unset") \ + _ (l3_src_address_is_local, "IP4 source address matches local interface") \ + _ (l3_src_address_learned, "ARP request IP4 source address learned") \ + _ (replies_received, "ARP replies received") \ + _ (opcode_not_request, "ARP opcode not request") \ + _ (proxy_arp_replies_sent, "Proxy ARP replies sent") \ + _ (l2_address_mismatch, "ARP hw addr does not match L2 frame src addr") \ + _ (gratuitous_arp, "ARP probe or announcement dropped") \ + _ (interface_no_table, "Interface is not mapped to an IP table") \ + _ (interface_not_ip_enabled, "Interface is not IP enabled") \ + _ (unnumbered_mismatch, "RX interface is unnumbered to different subnet") \ + +typedef enum +{ +#define _(sym,string) ETHERNET_ARP_ERROR_##sym, + foreach_ethernet_arp_error +#undef _ + ETHERNET_ARP_N_ERROR, +} ethernet_arp_reply_error_t; + +extern int arp_proxy_add (u32 fib_index, + const ip4_address_t * lo_addr, + const ip4_address_t * hi_addr); +extern int arp_proxy_del (u32 fib_index, + const ip4_address_t * lo_addr, + const ip4_address_t * hi_addr); + +extern int arp_proxy_enable (u32 sw_if_index); +extern int arp_proxy_disable (u32 sw_if_index); + +/** + * call back function when walking the DB of proxy ARPs + * @return 0 to stop the walk !0 to continue + */ +typedef walk_rc_t (proxy_arp_walk_t) (const ip4_address_t * lo_addr, + const ip4_address_t * hi_addr, + u32 fib_index, void *dat); + +extern void proxy_arp_walk (proxy_arp_walk_t cb, void *data); + +/** + * call back function when walking the DB of proxy ARP interface + * @return 0 to stop the walk !0 to continue + */ +typedef walk_rc_t (proxy_arp_intf_walk_t) (u32 sw_if_index, void *data); + +extern void proxy_arp_intfc_walk (proxy_arp_intf_walk_t cb, void *data); + +#endif + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/arp/arp_api.c b/src/vnet/arp/arp_api.c new file mode 100644 index 00000000000..170bace2b0a --- /dev/null +++ b/src/vnet/arp/arp_api.c @@ -0,0 +1,190 @@ +/* + * Copyright (c) 2016 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 <stddef.h> + +#include <vnet/arp/arp.h> + +#include <vnet/fib/fib_table.h> +#include <vnet/ip/ip_types_api.h> + +#include <vpp/app/version.h> + +#include <vlibapi/api.h> +#include <vlibmemory/api.h> + +/* define message IDs */ +#include <vnet/format_fns.h> +#include <vnet/arp/arp.api_enum.h> +#include <vnet/arp/arp.api_types.h> + +/** + * Base message ID fot the plugin + */ +static u32 arp_base_msg_id; +#define REPLY_MSG_ID_BASE arp_base_msg_id + +#include <vlibapi/api_helper_macros.h> + +static void +vl_api_proxy_arp_add_del_t_handler (vl_api_proxy_arp_add_del_t * mp) +{ + vl_api_proxy_arp_add_del_reply_t *rmp; + ip4_address_t lo, hi; + u32 fib_index; + int rv; + + fib_index = fib_table_find (FIB_PROTOCOL_IP4, ntohl (mp->proxy.table_id)); + + if (~0 == fib_index) + { + rv = VNET_API_ERROR_NO_SUCH_FIB; + goto out; + } + + ip4_address_decode (mp->proxy.low, &lo); + ip4_address_decode (mp->proxy.hi, &hi); + + if (mp->is_add) + rv = arp_proxy_add (fib_index, &lo, &hi); + else + rv = arp_proxy_del (fib_index, &lo, &hi); + +out: + REPLY_MACRO (VL_API_PROXY_ARP_ADD_DEL_REPLY); +} + +typedef struct proxy_arp_walk_ctx_t_ +{ + vl_api_registration_t *reg; + u32 context; +} proxy_arp_walk_ctx_t; + +static walk_rc_t +send_proxy_arp_details (const ip4_address_t * lo_addr, + const ip4_address_t * hi_addr, + u32 fib_index, void *data) +{ + vl_api_proxy_arp_details_t *mp; + proxy_arp_walk_ctx_t *ctx; + + ctx = data; + + mp = vl_msg_api_alloc (sizeof (*mp)); + clib_memset (mp, 0, sizeof (*mp)); + mp->_vl_msg_id = ntohs (VL_API_PROXY_ARP_DETAILS + REPLY_MSG_ID_BASE); + mp->context = ctx->context; + mp->proxy.table_id = htonl (fib_index); + + ip4_address_encode (lo_addr, mp->proxy.low); + ip4_address_encode (hi_addr, mp->proxy.hi); + + vl_api_send_msg (ctx->reg, (u8 *) mp); + + return (WALK_CONTINUE); +} + +static void +vl_api_proxy_arp_dump_t_handler (vl_api_proxy_arp_dump_t * mp) +{ + vl_api_registration_t *reg; + + reg = vl_api_client_index_to_registration (mp->client_index); + if (!reg) + return; + + proxy_arp_walk_ctx_t wctx = { + .reg = reg, + .context = mp->context, + }; + + proxy_arp_walk (send_proxy_arp_details, &wctx); +} + +static walk_rc_t +send_proxy_arp_intfc_details (u32 sw_if_index, void *data) +{ + vl_api_proxy_arp_intfc_details_t *mp; + proxy_arp_walk_ctx_t *ctx; + + ctx = data; + + mp = vl_msg_api_alloc (sizeof (*mp)); + clib_memset (mp, 0, sizeof (*mp)); + mp->_vl_msg_id = ntohs (VL_API_PROXY_ARP_INTFC_DETAILS + REPLY_MSG_ID_BASE); + mp->context = ctx->context; + mp->sw_if_index = htonl (sw_if_index); + + vl_api_send_msg (ctx->reg, (u8 *) mp); + + return (WALK_CONTINUE); +} + +static void +vl_api_proxy_arp_intfc_dump_t_handler (vl_api_proxy_arp_intfc_dump_t * mp) +{ + vl_api_registration_t *reg; + + reg = vl_api_client_index_to_registration (mp->client_index); + if (!reg) + return; + + proxy_arp_walk_ctx_t wctx = { + .reg = reg, + .context = mp->context, + }; + + proxy_arp_intfc_walk (send_proxy_arp_intfc_details, &wctx); +} + +static void + vl_api_proxy_arp_intfc_enable_disable_t_handler + (vl_api_proxy_arp_intfc_enable_disable_t * mp) +{ + vl_api_proxy_arp_intfc_enable_disable_reply_t *rmp; + int rv; + + VALIDATE_SW_IF_INDEX (mp); + + if (mp->enable) + rv = arp_proxy_enable (ntohl (mp->sw_if_index)); + else + rv = arp_proxy_disable (ntohl (mp->sw_if_index)); + + BAD_SW_IF_INDEX_LABEL; + + REPLY_MACRO (VL_API_PROXY_ARP_INTFC_ENABLE_DISABLE_REPLY); +} + +#include <vnet/arp/arp.api.c> + +static clib_error_t * +arp_api_init (vlib_main_t * vm) +{ + /* Ask for a correctly-sized block of API message decode slots */ + arp_base_msg_id = setup_message_id_table (); + + return 0; +} + +VLIB_INIT_FUNCTION (arp_api_init); + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/arp/arp_packet.h b/src/vnet/arp/arp_packet.h new file mode 100644 index 00000000000..a860c258f75 --- /dev/null +++ b/src/vnet/arp/arp_packet.h @@ -0,0 +1,91 @@ +/* + * ethernet/arp.c: IP v4 ARP node + * + * Copyright (c) 2010 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef __ARP_PACKET_H__ +#define __ARP_PACKET_H__ + +#include <vnet/arp/arp.h> + +/* Either we drop the packet or we send a reply to the sender. */ +typedef enum +{ + ARP_REPLY_NEXT_DROP, + ARP_REPLY_NEXT_REPLY_TX, + ARP_REPLY_N_NEXT, +} arp_reply_next_t; + +static_always_inline u32 +arp_mk_reply (vnet_main_t * vnm, + vlib_buffer_t * p0, + u32 sw_if_index0, + const ip4_address_t * if_addr0, + ethernet_arp_header_t * arp0, ethernet_header_t * eth_rx) +{ + vnet_hw_interface_t *hw_if0; + u8 *rewrite0, rewrite0_len; + ethernet_header_t *eth_tx; + u32 next0; + + /* Send a reply. + An adjacency to the sender is not always present, + so we use the interface to build us a rewrite string + which will contain all the necessary tags. */ + rewrite0 = ethernet_build_rewrite (vnm, sw_if_index0, + VNET_LINK_ARP, eth_rx->src_address); + rewrite0_len = vec_len (rewrite0); + + /* Figure out how much to rewind current data from adjacency. */ + vlib_buffer_advance (p0, -rewrite0_len); + eth_tx = vlib_buffer_get_current (p0); + + vnet_buffer (p0)->sw_if_index[VLIB_TX] = sw_if_index0; + hw_if0 = vnet_get_sup_hw_interface (vnm, sw_if_index0); + + /* Send reply back through input interface */ + vnet_buffer (p0)->sw_if_index[VLIB_TX] = sw_if_index0; + next0 = ARP_REPLY_NEXT_REPLY_TX; + + arp0->opcode = clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_reply); + + arp0->ip4_over_ethernet[1] = arp0->ip4_over_ethernet[0]; + + mac_address_from_bytes (&arp0->ip4_over_ethernet[0].mac, + hw_if0->hw_address); + clib_mem_unaligned (&arp0->ip4_over_ethernet[0].ip4.data_u32, u32) = + if_addr0->data_u32; + + /* Hardware must be ethernet-like. */ + ASSERT (vec_len (hw_if0->hw_address) == 6); + + /* the rx nd tx ethernet headers wil overlap in the case + * when we received a tagged VLAN=0 packet, but we are sending + * back untagged */ + clib_memcpy_fast (eth_tx, rewrite0, vec_len (rewrite0)); + vec_free (rewrite0); + + return (next0); +} + +#endif + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/arp/arp_proxy.c b/src/vnet/arp/arp_proxy.c new file mode 100644 index 00000000000..346a21775f8 --- /dev/null +++ b/src/vnet/arp/arp_proxy.c @@ -0,0 +1,407 @@ +/* + * ethernet/arp.c: IP v4 ARP node + * + * Copyright (c) 2010 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/arp/arp.h> +#include <vnet/arp/arp_packet.h> + +#include <vnet/fib/ip4_fib.h> + +typedef struct +{ + ip4_address_t lo_addr; + ip4_address_t hi_addr; + u32 fib_index; +} ethernet_proxy_arp_t; + +typedef struct arp_proxy_main_t_ +{ + /** Per interface state */ + bool *enabled_by_sw_if_index; + + /* Proxy arp vector */ + ethernet_proxy_arp_t *proxy_arps; +} arp_proxy_main_t; + +arp_proxy_main_t arp_proxy_main; + +void +proxy_arp_walk (proxy_arp_walk_t cb, void *data) +{ + arp_proxy_main_t *am = &arp_proxy_main; + ethernet_proxy_arp_t *pa; + + vec_foreach (pa, am->proxy_arps) + { + if (!cb (&pa->lo_addr, &pa->hi_addr, pa->fib_index, data)) + break; + } +} + +int +arp_proxy_disable (u32 sw_if_index) +{ + arp_proxy_main_t *am = &arp_proxy_main; + + vec_validate (am->enabled_by_sw_if_index, sw_if_index); + + if (am->enabled_by_sw_if_index[sw_if_index]) + { + vnet_feature_enable_disable ("arp", "arp-proxy", + sw_if_index, 0, NULL, 0); + } + am->enabled_by_sw_if_index[sw_if_index] = false; + + return (0); +} + +int +arp_proxy_enable (u32 sw_if_index) +{ + arp_proxy_main_t *am = &arp_proxy_main; + + vec_validate (am->enabled_by_sw_if_index, sw_if_index); + + if (!am->enabled_by_sw_if_index[sw_if_index]) + { + vnet_feature_enable_disable ("arp", "arp-proxy", + sw_if_index, 1, NULL, 0); + } + am->enabled_by_sw_if_index[sw_if_index] = true; + + return (0); +} + +static int +vnet_proxy_arp_add_del (const ip4_address_t * lo_addr, + const ip4_address_t * hi_addr, + u32 fib_index, int is_del) +{ + arp_proxy_main_t *am = &arp_proxy_main; + ethernet_proxy_arp_t *pa; + u32 found_at_index = ~0; + + vec_foreach (pa, am->proxy_arps) + { + if (pa->lo_addr.as_u32 == lo_addr->as_u32 && + pa->hi_addr.as_u32 == hi_addr->as_u32 && pa->fib_index == fib_index) + { + found_at_index = pa - am->proxy_arps; + break; + } + } + + if (found_at_index != ~0) + { + /* Delete, otherwise it's already in the table */ + if (is_del) + vec_delete (am->proxy_arps, 1, found_at_index); + return 0; + } + /* delete, no such entry */ + if (is_del) + return VNET_API_ERROR_NO_SUCH_ENTRY; + + /* add, not in table */ + vec_add2 (am->proxy_arps, pa, 1); + pa->lo_addr.as_u32 = lo_addr->as_u32; + pa->hi_addr.as_u32 = hi_addr->as_u32; + pa->fib_index = fib_index; + return 0; +} + +int +arp_proxy_add (u32 fib_index, + const ip4_address_t * lo, const ip4_address_t * hi) +{ + return (vnet_proxy_arp_add_del (lo, hi, fib_index, 0)); +} + +int +arp_proxy_del (u32 fib_index, + const ip4_address_t * lo, const ip4_address_t * hi) +{ + return (vnet_proxy_arp_add_del (lo, hi, fib_index, 1)); +} + +void +proxy_arp_intfc_walk (proxy_arp_intf_walk_t cb, void *data) +{ + arp_proxy_main_t *am = &arp_proxy_main; + bool *enabled; + + vec_foreach (enabled, am->enabled_by_sw_if_index) + { + if (*enabled) + cb (enabled - am->enabled_by_sw_if_index, data); + } +} + +static clib_error_t * +set_int_proxy_arp_command_fn (vlib_main_t * vm, + unformat_input_t * + input, vlib_cli_command_t * cmd) +{ + vnet_main_t *vnm = vnet_get_main (); + u32 sw_if_index; + int enable = 0; + + sw_if_index = ~0; + + while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (input, "%U", unformat_vnet_sw_interface, + vnm, &sw_if_index)) + ; + else if (unformat (input, "enable") || unformat (input, "on")) + enable = 1; + else if (unformat (input, "disable") || unformat (input, "off")) + enable = 0; + else + break; + } + + if (~0 == sw_if_index) + return clib_error_return (0, "unknown input '%U'", + format_unformat_error, input); + + if (enable) + arp_proxy_enable (sw_if_index); + else + arp_proxy_disable (sw_if_index); + + return 0; +} + +/* *INDENT-OFF* */ +/*? + * Enable proxy-arp on an interface. The vpp stack will answer ARP + * requests for the indicated address range. Multiple proxy-arp + * ranges may be provisioned. + * + * @note Proxy ARP as a technology is infamous for blackholing traffic. + * Also, the underlying implementation has not been performance-tuned. + * Avoid creating an unnecessarily large set of ranges. + * + * @cliexpar + * To enable proxy arp on a range of addresses, use: + * @cliexcmd{set ip arp proxy 6.0.0.1 - 6.0.0.11} + * Append 'del' to delete a range of proxy ARP addresses: + * @cliexcmd{set ip arp proxy 6.0.0.1 - 6.0.0.11 del} + * You must then specifically enable proxy arp on individual interfaces: + * @cliexcmd{set interface proxy-arp GigabitEthernet0/8/0 enable} + * To disable proxy arp on an individual interface: + * @cliexcmd{set interface proxy-arp GigabitEthernet0/8/0 disable} + ?*/ +VLIB_CLI_COMMAND (set_int_proxy_enable_command, static) = { + .path = "set interface proxy-arp", + .short_help = + "set interface proxy-arp <intfc> [enable|disable]", + .function = set_int_proxy_arp_command_fn, +}; +/* *INDENT-ON* */ + +typedef struct +{ + u8 packet_data[64]; +} ethernet_arp_input_trace_t; + +static u8 * +format_ethernet_arp_input_trace (u8 * s, va_list * va) +{ + CLIB_UNUSED (vlib_main_t * vm) = va_arg (*va, vlib_main_t *); + CLIB_UNUSED (vlib_node_t * node) = va_arg (*va, vlib_node_t *); + ethernet_arp_input_trace_t *t = va_arg (*va, ethernet_arp_input_trace_t *); + + s = format (s, "%U", + format_ethernet_arp_header, + t->packet_data, sizeof (t->packet_data)); + + return s; +} + +static uword +arp_proxy (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + arp_proxy_main_t *am = &arp_proxy_main; + vnet_main_t *vnm = vnet_get_main (); + u32 n_left_from, next_index, *from, *to_next; + u32 n_arp_replies_sent = 0; + + from = vlib_frame_vector_args (frame); + n_left_from = frame->n_vectors; + next_index = node->cached_next_index; + + if (node->flags & VLIB_NODE_FLAG_TRACE) + vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors, + /* stride */ 1, + sizeof (ethernet_arp_input_trace_t)); + + while (n_left_from > 0) + { + u32 n_left_to_next; + + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); + + while (n_left_from > 0 && n_left_to_next > 0) + { + vlib_buffer_t *p0; + ethernet_arp_header_t *arp0; + ethernet_header_t *eth_rx; + ip4_address_t proxy_src; + u32 pi0, error0, next0, sw_if_index0, fib_index0; + u8 is_request0; + ethernet_proxy_arp_t *pa; + + pi0 = from[0]; + to_next[0] = pi0; + from += 1; + to_next += 1; + n_left_from -= 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, pi0); + arp0 = vlib_buffer_get_current (p0); + /* Fill in ethernet header. */ + eth_rx = ethernet_buffer_get_header (p0); + + is_request0 = arp0->opcode + == clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_request); + + error0 = ETHERNET_ARP_ERROR_replies_sent; + sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX]; + next0 = ARP_REPLY_NEXT_DROP; + + fib_index0 = ip4_fib_table_get_index_for_sw_if_index (sw_if_index0); + if (~0 == fib_index0) + { + error0 = ETHERNET_ARP_ERROR_interface_no_table; + } + + if (0 == error0 && is_request0) + { + u32 this_addr = clib_net_to_host_u32 + (arp0->ip4_over_ethernet[1].ip4.as_u32); + + vec_foreach (pa, am->proxy_arps) + { + u32 lo_addr = clib_net_to_host_u32 (pa->lo_addr.as_u32); + u32 hi_addr = clib_net_to_host_u32 (pa->hi_addr.as_u32); + + /* an ARP request hit in the proxy-arp table? */ + if ((this_addr >= lo_addr && this_addr <= hi_addr) && + (fib_index0 == pa->fib_index)) + { + proxy_src.as_u32 = + arp0->ip4_over_ethernet[1].ip4.data_u32; + + /* + * change the interface address to the proxied + */ + n_arp_replies_sent++; + + next0 = + arp_mk_reply (vnm, p0, sw_if_index0, &proxy_src, arp0, + eth_rx); + } + } + } + else + { + p0->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); + } + + vlib_error_count (vm, node->node_index, + ETHERNET_ARP_ERROR_replies_sent, n_arp_replies_sent); + + return frame->n_vectors; +} + +static char *ethernet_arp_error_strings[] = { +#define _(sym,string) string, + foreach_ethernet_arp_error +#undef _ +}; + +VLIB_REGISTER_NODE (arp_proxy_node, static) = +{ + .function = arp_proxy,.name = "arp-proxy",.vector_size = + sizeof (u32),.n_errors = ETHERNET_ARP_N_ERROR,.error_strings = + ethernet_arp_error_strings,.n_next_nodes = ARP_REPLY_N_NEXT,.next_nodes = + { + [ARP_REPLY_NEXT_DROP] = "error-drop", + [ARP_REPLY_NEXT_REPLY_TX] = "interface-output",} +,.format_buffer = format_ethernet_arp_header,.format_trace = + format_ethernet_arp_input_trace,}; + +static clib_error_t * +show_ip4_arp (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) +{ + arp_proxy_main_t *am = &arp_proxy_main; + ethernet_proxy_arp_t *pa; + + if (vec_len (am->proxy_arps)) + { + vlib_cli_output (vm, "Proxy arps enabled for:"); + vec_foreach (pa, am->proxy_arps) + { + vlib_cli_output (vm, "Fib_index %d %U - %U ", + pa->fib_index, + format_ip4_address, &pa->lo_addr, + format_ip4_address, &pa->hi_addr); + } + } + + return (NULL); +} + +/*? + * Display all the IPv4 ARP proxy entries. + * + * @cliexpar + * Example of how to display the IPv4 ARP table: + * @cliexstart{show ip arp} + * Time FIB IP4 Flags Ethernet Interface + * 346.3028 0 6.1.1.3 de:ad:be:ef:ba:be GigabitEthernet2/0/0 + * 3077.4271 0 6.1.1.4 S de:ad:be:ef:ff:ff GigabitEthernet2/0/0 + * 2998.6409 1 6.2.2.3 de:ad:be:ef:00:01 GigabitEthernet2/0/0 + * Proxy arps enabled for: + * Fib_index 0 6.0.0.1 - 6.0.0.11 + * @cliexend + ?*/ +/* *INDENT-OFF* */ +VLIB_CLI_COMMAND (show_ip4_arp_command, static) = { + .path = "show arp proxy", + .function = show_ip4_arp, + .short_help = "show ip arp", +}; +/* *INDENT-ON* */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/arp/arp_test.c b/src/vnet/arp/arp_test.c new file mode 100644 index 00000000000..861d0739406 --- /dev/null +++ b/src/vnet/arp/arp_test.c @@ -0,0 +1,168 @@ +/* + *------------------------------------------------------------------ + * arp_test.c + * + * Copyright (c) 2019 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 <vat/vat.h> +#include <vlibapi/api.h> +#include <vlibmemory/api.h> +#include <vppinfra/error.h> + +#include <vnet/ip/ip_format_fns.h> + +#include <vpp/api/types.h> + +typedef struct +{ + /* API message ID base */ + u16 msg_id_base; + u32 ping_id; + vat_main_t *vat_main; +} arp_test_main_t; + +arp_test_main_t arp_test_main; + +#define __plugin_msg_base arp_test_main.msg_id_base +#include <vlibapi/vat_helper_macros.h> +uword unformat_sw_if_index (unformat_input_t * input, va_list * args); + +/* Declare message IDs */ +#include <arp/arp.api_enum.h> +#include <arp/arp.api_types.h> +#include <vpp/api/vpe.api_types.h> + +static int +api_proxy_arp_dump (vat_main_t * vam) +{ + return -1; +} + +static int +api_proxy_arp_intfc_dump (vat_main_t * vam) +{ + return -1; +} + +static void +vl_api_proxy_arp_details_t_handler (vl_api_proxy_arp_details_t * mp) +{ +} + +static void +vl_api_proxy_arp_intfc_details_t_handler (vl_api_proxy_arp_intfc_details_t * + mp) +{ +} + +static int +api_proxy_arp_add_del (vat_main_t * vam) +{ + unformat_input_t *i = vam->input; + vl_api_proxy_arp_add_del_t *mp; + u32 vrf_id = 0; + u8 is_add = 1; + vl_api_ip4_address_t lo, hi; + u8 range_set = 0; + int ret; + + while (unformat_check_input (i) != UNFORMAT_END_OF_INPUT) + { + if (unformat (i, "vrf %d", &vrf_id)) + ; + else if (unformat (i, "%U - %U", unformat_vl_api_ip4_address, &lo, + unformat_vl_api_ip4_address, &hi)) + range_set = 1; + else if (unformat (i, "del")) + is_add = 0; + else + { + clib_warning ("parse error '%U'", format_unformat_error, i); + return -99; + } + } + + if (range_set == 0) + { + errmsg ("address range not set"); + return -99; + } + + M (PROXY_ARP_ADD_DEL, mp); + + mp->proxy.table_id = ntohl (vrf_id); + mp->is_add = is_add; + clib_memcpy (mp->proxy.low, &lo, sizeof (lo)); + clib_memcpy (mp->proxy.hi, &hi, sizeof (hi)); + + S (mp); + W (ret); + return ret; +} + +static int +api_proxy_arp_intfc_enable_disable (vat_main_t * vam) +{ + unformat_input_t *i = vam->input; + vl_api_proxy_arp_intfc_enable_disable_t *mp; + u32 sw_if_index; + u8 enable = 1; + u8 sw_if_index_set = 0; + int ret; + + while (unformat_check_input (i) != UNFORMAT_END_OF_INPUT) + { + if (unformat (i, "%U", unformat_sw_if_index, vam, &sw_if_index)) + sw_if_index_set = 1; + else if (unformat (i, "sw_if_index %d", &sw_if_index)) + sw_if_index_set = 1; + else if (unformat (i, "enable")) + enable = 1; + else if (unformat (i, "disable")) + enable = 0; + else + { + clib_warning ("parse error '%U'", format_unformat_error, i); + return -99; + } + } + + if (sw_if_index_set == 0) + { + errmsg ("missing interface name or sw_if_index"); + return -99; + } + + M (PROXY_ARP_INTFC_ENABLE_DISABLE, mp); + + mp->sw_if_index = ntohl (sw_if_index); + mp->enable = enable; + + S (mp); + W (ret); + return ret; +} + +#include <arp/arp.api_test.c> + + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/bfd/bfd_api.h b/src/vnet/bfd/bfd_api.h index 9f0509d5f73..38d342a2e5b 100644 --- a/src/vnet/bfd/bfd_api.h +++ b/src/vnet/bfd/bfd_api.h @@ -21,7 +21,7 @@ #include <vnet/api_errno.h> #include <vnet/vnet.h> -#include <vnet/ip/ip6_packet.h> +#include <vnet/ip/ip46_address.h> #define foreach_bfd_transport(F) \ F (UDP4, "ip4-rewrite") \ diff --git a/src/vnet/bfd/bfd_udp.c b/src/vnet/bfd/bfd_udp.c index c545042c978..0a9106e78e5 100644 --- a/src/vnet/bfd/bfd_udp.c +++ b/src/vnet/bfd/bfd_udp.c @@ -29,7 +29,7 @@ #include <vnet/ip/ip4.h> #include <vnet/ip/ip6.h> #include <vnet/ip/ip6_packet.h> -#include <vnet/ip/ip6_neighbor.h> +#include <vnet/ip/ip6_link.h> #include <vnet/adj/adj.h> #include <vnet/adj/adj_nbr.h> #include <vnet/dpo/receive_dpo.h> @@ -612,9 +612,9 @@ bfd_udp_validate_api_input (u32 sw_if_index, if (ip6_address_is_link_local_unicast (&local_addr->ip6)) { - ip6_address_t ll_addr; - ll_addr = ip6_neighbor_get_link_local_address (sw_if_index); - if (ip6_address_is_equal (&ll_addr, &local_addr->ip6)) + const ip6_address_t *ll_addr; + ll_addr = ip6_get_link_local_address (sw_if_index); + if (ip6_address_is_equal (ll_addr, &local_addr->ip6)) { /* valid address for this interface */ local_ip_valid = 1; diff --git a/src/vnet/bonding/device.c b/src/vnet/bonding/device.c index 57f2806472f..550321b3763 100644 --- a/src/vnet/bonding/device.c +++ b/src/vnet/bonding/device.c @@ -24,7 +24,7 @@ #include <vnet/bonding/node.h> #include <vppinfra/lb_hash_hash.h> #include <vnet/ip/ip.h> -#include <vnet/ethernet/arp_packet.h> +#include <vnet/ip-neighbor/ip_neighbor.h> #define foreach_bond_tx_error \ _(NONE, "no error") \ @@ -799,8 +799,7 @@ bond_active_interface_switch_cb (vnet_main_t * vnm, u32 sw_if_index, { bond_main_t *bm = &bond_main; - send_ip4_garp (bm->vlib_main, sw_if_index); - send_ip6_na (bm->vlib_main, sw_if_index); + ip_neighbor_advertise (bm->vlib_main, IP46_TYPE_BOTH, NULL, sw_if_index); return (WALK_CONTINUE); } diff --git a/src/vnet/ethernet/arp.c b/src/vnet/ethernet/arp.c deleted file mode 100644 index 6b1069ecb6d..00000000000 --- a/src/vnet/ethernet/arp.c +++ /dev/null @@ -1,3062 +0,0 @@ -/* - * ethernet/arp.c: IP v4 ARP node - * - * Copyright (c) 2010 Cisco and/or its affiliates. - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at: - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -#include <vnet/ip/ip.h> -#include <vnet/ip/ip_neighbor.h> -#include <vnet/ip/ip6.h> -#include <vnet/ethernet/ethernet.h> -#include <vnet/ethernet/arp.h> -#include <vnet/l2/l2_input.h> -#include <vppinfra/mhash.h> -#include <vnet/fib/ip4_fib.h> -#include <vnet/fib/fib_entry_src.h> -#include <vnet/adj/adj_nbr.h> -#include <vnet/adj/adj_mcast.h> -#include <vnet/mpls/mpls.h> -#include <vnet/l2/feat_bitmap.h> - -#include <vlibmemory/api.h> - -/** - * @file - * @brief IPv4 ARP. - * - * This file contains code to manage the IPv4 ARP tables (IP Address - * to MAC Address lookup). - */ - - -/** - * @brief Per-interface ARP configuration and state - */ -typedef struct ethernet_arp_interface_t_ -{ - /** - * Hash table of ARP entries. - * Since this hash table is per-interface, the key is only the IPv4 address. - */ - uword *arp_entries; - /** - * Is ARP enabled on this interface - */ - u32 enabled; - /** - * Is Proxy ARP enabled on this interface - */ - u32 proxy_enabled; -} ethernet_arp_interface_t; - -typedef struct -{ - ip4_address_t lo_addr; - ip4_address_t hi_addr; - u32 fib_index; -} ethernet_proxy_arp_t; - -typedef struct -{ - u32 next_index; - uword node_index; - uword type_opaque; - uword data; - /* Used for arp event notification only */ - arp_change_event_cb_t data_callback; - u32 pid; -} pending_resolution_t; - -typedef struct -{ - /* Hash tables mapping name to opcode. */ - uword *opcode_by_name; - - /* lite beer "glean" adjacency handling */ - uword *pending_resolutions_by_address; - pending_resolution_t *pending_resolutions; - - /* Mac address change notification */ - uword *mac_changes_by_address; - pending_resolution_t *mac_changes; - - ethernet_arp_ip4_entry_t *ip4_entry_pool; - - /* ARP attack mitigation */ - u32 arp_delete_rotor; - u32 limit_arp_cache_size; - - /** Per interface state */ - ethernet_arp_interface_t *ethernet_arp_by_sw_if_index; - - /* Proxy arp vector */ - ethernet_proxy_arp_t *proxy_arps; - - uword wc_ip4_arp_publisher_node; - uword wc_ip4_arp_publisher_et; - - /* ARP feature arc index */ - u8 feature_arc_index; -} ethernet_arp_main_t; - -static ethernet_arp_main_t ethernet_arp_main; - -typedef struct -{ - u32 sw_if_index; - ip4_address_t ip4; - mac_address_t mac; - ip_neighbor_flags_t nbr_flags; - u32 flags; -#define ETHERNET_ARP_ARGS_REMOVE (1<<0) -#define ETHERNET_ARP_ARGS_FLUSH (1<<1) -#define ETHERNET_ARP_ARGS_POPULATE (1<<2) -#define ETHERNET_ARP_ARGS_WC_PUB (1<<3) -} vnet_arp_set_ip4_over_ethernet_rpc_args_t; - -static const u8 vrrp_prefix[] = { 0x00, 0x00, 0x5E, 0x00, 0x01 }; - -/* Node index for send_garp_na_process */ -u32 send_garp_na_process_node_index; - -static void -set_ip4_over_ethernet_rpc_callback (vnet_arp_set_ip4_over_ethernet_rpc_args_t - * a); - -static u8 * -format_ethernet_arp_hardware_type (u8 * s, va_list * va) -{ - ethernet_arp_hardware_type_t h = va_arg (*va, ethernet_arp_hardware_type_t); - char *t = 0; - switch (h) - { -#define _(n,f) case n: t = #f; break; - foreach_ethernet_arp_hardware_type; -#undef _ - - default: - return format (s, "unknown 0x%x", h); - } - - return format (s, "%s", t); -} - -static u8 * -format_ethernet_arp_opcode (u8 * s, va_list * va) -{ - ethernet_arp_opcode_t o = va_arg (*va, ethernet_arp_opcode_t); - char *t = 0; - switch (o) - { -#define _(f) case ETHERNET_ARP_OPCODE_##f: t = #f; break; - foreach_ethernet_arp_opcode; -#undef _ - - default: - return format (s, "unknown 0x%x", o); - } - - return format (s, "%s", t); -} - -static uword -unformat_ethernet_arp_opcode_host_byte_order (unformat_input_t * input, - va_list * args) -{ - int *result = va_arg (*args, int *); - ethernet_arp_main_t *am = ðernet_arp_main; - int x, i; - - /* Numeric opcode. */ - if (unformat (input, "0x%x", &x) || unformat (input, "%d", &x)) - { - if (x >= (1 << 16)) - return 0; - *result = x; - return 1; - } - - /* Named type. */ - if (unformat_user (input, unformat_vlib_number_by_name, - am->opcode_by_name, &i)) - { - *result = i; - return 1; - } - - return 0; -} - -static uword -unformat_ethernet_arp_opcode_net_byte_order (unformat_input_t * input, - va_list * args) -{ - int *result = va_arg (*args, int *); - if (!unformat_user - (input, unformat_ethernet_arp_opcode_host_byte_order, result)) - return 0; - - *result = clib_host_to_net_u16 ((u16) * result); - return 1; -} - -static u8 * -format_ethernet_arp_header (u8 * s, va_list * va) -{ - ethernet_arp_header_t *a = va_arg (*va, ethernet_arp_header_t *); - u32 max_header_bytes = va_arg (*va, u32); - u32 indent; - u16 l2_type, l3_type; - - if (max_header_bytes != 0 && sizeof (a[0]) > max_header_bytes) - return format (s, "ARP header truncated"); - - l2_type = clib_net_to_host_u16 (a->l2_type); - l3_type = clib_net_to_host_u16 (a->l3_type); - - indent = format_get_indent (s); - - s = format (s, "%U, type %U/%U, address size %d/%d", - format_ethernet_arp_opcode, clib_net_to_host_u16 (a->opcode), - format_ethernet_arp_hardware_type, l2_type, - format_ethernet_type, l3_type, - a->n_l2_address_bytes, a->n_l3_address_bytes); - - if (l2_type == ETHERNET_ARP_HARDWARE_TYPE_ethernet - && l3_type == ETHERNET_TYPE_IP4) - { - s = format (s, "\n%U%U/%U -> %U/%U", - format_white_space, indent, - format_mac_address_t, &a->ip4_over_ethernet[0].mac, - format_ip4_address, &a->ip4_over_ethernet[0].ip4, - format_mac_address_t, &a->ip4_over_ethernet[1].mac, - format_ip4_address, &a->ip4_over_ethernet[1].ip4); - } - else - { - uword n2 = a->n_l2_address_bytes; - uword n3 = a->n_l3_address_bytes; - s = format (s, "\n%U%U/%U -> %U/%U", - format_white_space, indent, - format_hex_bytes, a->data + 0 * n2 + 0 * n3, n2, - format_hex_bytes, a->data + 1 * n2 + 0 * n3, n3, - format_hex_bytes, a->data + 1 * n2 + 1 * n3, n2, - format_hex_bytes, a->data + 2 * n2 + 1 * n3, n3); - } - - return s; -} - -u8 * -format_ethernet_arp_ip4_entry (u8 * s, va_list * va) -{ - vnet_main_t *vnm = va_arg (*va, vnet_main_t *); - ethernet_arp_ip4_entry_t *e = va_arg (*va, ethernet_arp_ip4_entry_t *); - vnet_sw_interface_t *si; - - if (!e) - return format (s, "%=12s%=16s%=6s%=20s%=24s", "Time", "IP4", - "Flags", "Ethernet", "Interface"); - - si = vnet_get_sw_interface (vnm, e->sw_if_index); - - return format (s, "%=12U%=16U%=6U%=20U%U", - format_vlib_time, vnm->vlib_main, e->time_last_updated, - format_ip4_address, &e->ip4_address, - format_ip_neighbor_flags, e->flags, - format_mac_address_t, &e->mac, - format_vnet_sw_interface_name, vnm, si); -} - -typedef struct -{ - u8 packet_data[64]; -} ethernet_arp_input_trace_t; - -static u8 * -format_ethernet_arp_input_trace (u8 * s, va_list * va) -{ - CLIB_UNUSED (vlib_main_t * vm) = va_arg (*va, vlib_main_t *); - CLIB_UNUSED (vlib_node_t * node) = va_arg (*va, vlib_node_t *); - ethernet_arp_input_trace_t *t = va_arg (*va, ethernet_arp_input_trace_t *); - - s = format (s, "%U", - format_ethernet_arp_header, - t->packet_data, sizeof (t->packet_data)); - - return s; -} - -static u8 * -format_arp_term_input_trace (u8 * s, va_list * va) -{ - CLIB_UNUSED (vlib_main_t * vm) = va_arg (*va, vlib_main_t *); - CLIB_UNUSED (vlib_node_t * node) = va_arg (*va, vlib_node_t *); - ethernet_arp_input_trace_t *t = va_arg (*va, ethernet_arp_input_trace_t *); - - /* arp-term trace data saved is either arp or ip6/icmp6 packet: - - for arp, the 1st 16-bit field is hw type of value of 0x0001. - - for ip6, the first nibble has value of 6. */ - s = format (s, "%U", t->packet_data[0] == 0 ? - format_ethernet_arp_header : format_ip6_header, - t->packet_data, sizeof (t->packet_data)); - - return s; -} - -static void -arp_nbr_probe (ip_adjacency_t * adj) -{ - vnet_main_t *vnm = vnet_get_main (); - ip4_main_t *im = &ip4_main; - ip_interface_address_t *ia; - ethernet_arp_header_t *h; - vnet_hw_interface_t *hi; - vnet_sw_interface_t *si; - ip4_address_t *src; - vlib_buffer_t *b; - vlib_main_t *vm; - u32 bi = 0; - - vm = vlib_get_main (); - - si = vnet_get_sw_interface (vnm, adj->rewrite_header.sw_if_index); - - if (!(si->flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP)) - { - return; - } - - src = - ip4_interface_address_matching_destination (im, - &adj->sub_type.nbr.next_hop. - ip4, - adj->rewrite_header. - sw_if_index, &ia); - if (!src) - { - return; - } - - h = - vlib_packet_template_get_packet (vm, &im->ip4_arp_request_packet_template, - &bi); - if (!h) - return; - - hi = vnet_get_sup_hw_interface (vnm, adj->rewrite_header.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 = adj->sub_type.nbr.next_hop.ip4; - - b = vlib_get_buffer (vm, bi); - vnet_buffer (b)->sw_if_index[VLIB_RX] = - vnet_buffer (b)->sw_if_index[VLIB_TX] = adj->rewrite_header.sw_if_index; - - /* Add encapsulation string for software interface (e.g. ethernet header). */ - vnet_rewrite_one_header (adj[0], h, sizeof (ethernet_header_t)); - 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); - } -} - -static void -arp_mk_complete (adj_index_t ai, ethernet_arp_ip4_entry_t * e) -{ - adj_nbr_update_rewrite - (ai, ADJ_NBR_REWRITE_FLAG_COMPLETE, - ethernet_build_rewrite (vnet_get_main (), - e->sw_if_index, - adj_get_link_type (ai), &e->mac)); -} - -static void -arp_mk_incomplete (adj_index_t ai) -{ - ip_adjacency_t *adj = adj_get (ai); - - adj_nbr_update_rewrite - (ai, - ADJ_NBR_REWRITE_FLAG_INCOMPLETE, - ethernet_build_rewrite (vnet_get_main (), - adj->rewrite_header.sw_if_index, - VNET_LINK_ARP, - VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST)); -} - -static ethernet_arp_ip4_entry_t * -arp_entry_find (ethernet_arp_interface_t * eai, const ip4_address_t * addr) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_arp_ip4_entry_t *e = NULL; - uword *p; - - if (NULL != eai->arp_entries) - { - p = hash_get (eai->arp_entries, addr->as_u32); - if (!p) - return (NULL); - - e = pool_elt_at_index (am->ip4_entry_pool, p[0]); - } - - return (e); -} - -static adj_walk_rc_t -arp_mk_complete_walk (adj_index_t ai, void *ctx) -{ - ethernet_arp_ip4_entry_t *e = ctx; - - arp_mk_complete (ai, e); - - return (ADJ_WALK_RC_CONTINUE); -} - -static adj_walk_rc_t -arp_mk_incomplete_walk (adj_index_t ai, void *ctx) -{ - arp_mk_incomplete (ai); - - return (ADJ_WALK_RC_CONTINUE); -} - -static int -arp_is_enabled (ethernet_arp_main_t * am, u32 sw_if_index) -{ - if (vec_len (am->ethernet_arp_by_sw_if_index) <= sw_if_index) - return 0; - - return (am->ethernet_arp_by_sw_if_index[sw_if_index].enabled); -} - -static void -arp_enable (ethernet_arp_main_t * am, u32 sw_if_index) -{ - if (arp_is_enabled (am, sw_if_index)) - return; - - vec_validate (am->ethernet_arp_by_sw_if_index, sw_if_index); - - am->ethernet_arp_by_sw_if_index[sw_if_index].enabled = 1; - - vnet_feature_enable_disable ("arp", "arp-reply", sw_if_index, 1, NULL, 0); - vnet_feature_enable_disable ("arp", "arp-disabled", sw_if_index, 0, NULL, - 0); -} - -static int -vnet_arp_flush_ip4_over_ethernet_internal (vnet_main_t * vnm, - vnet_arp_set_ip4_over_ethernet_rpc_args_t - * args); - -static void -arp_disable (ethernet_arp_main_t * am, u32 sw_if_index) -{ - ethernet_arp_interface_t *eai; - ethernet_arp_ip4_entry_t *e; - u32 i, *to_delete = 0; - hash_pair_t *pair; - - if (!arp_is_enabled (am, sw_if_index)) - return; - - vnet_feature_enable_disable ("arp", "arp-disabled", sw_if_index, 1, NULL, - 0); - vnet_feature_enable_disable ("arp", "arp-reply", sw_if_index, 0, NULL, 0); - - eai = &am->ethernet_arp_by_sw_if_index[sw_if_index]; - - - /* *INDENT-OFF* */ - hash_foreach_pair (pair, eai->arp_entries, - ({ - e = pool_elt_at_index(am->ip4_entry_pool, - pair->value[0]); - vec_add1 (to_delete, e - am->ip4_entry_pool); - })); - /* *INDENT-ON* */ - - for (i = 0; i < vec_len (to_delete); i++) - { - e = pool_elt_at_index (am->ip4_entry_pool, to_delete[i]); - - vnet_arp_set_ip4_over_ethernet_rpc_args_t delme = { - .ip4.as_u32 = e->ip4_address.as_u32, - .sw_if_index = e->sw_if_index, - .flags = ETHERNET_ARP_ARGS_FLUSH, - }; - mac_address_copy (&delme.mac, &e->mac); - - vnet_arp_flush_ip4_over_ethernet_internal (vnet_get_main (), &delme); - } - - vec_free (to_delete); - - eai->enabled = 0; -} - -void -arp_update_adjacency (vnet_main_t * vnm, u32 sw_if_index, u32 ai) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_arp_interface_t *arp_int; - ethernet_arp_ip4_entry_t *e; - ip_adjacency_t *adj; - - adj = adj_get (ai); - - arp_enable (am, sw_if_index); - arp_int = &am->ethernet_arp_by_sw_if_index[sw_if_index]; - e = arp_entry_find (arp_int, &adj->sub_type.nbr.next_hop.ip4); - - switch (adj->lookup_next_index) - { - case IP_LOOKUP_NEXT_GLEAN: - adj_glean_update_rewrite (ai); - break; - case IP_LOOKUP_NEXT_ARP: - if (NULL != e) - { - adj_nbr_walk_nh4 (sw_if_index, - &e->ip4_address, arp_mk_complete_walk, e); - } - else - { - /* - * no matching ARP entry. - * construct the rewrite required to for an ARP packet, and stick - * that in the adj's pipe to smoke. - */ - adj_nbr_update_rewrite - (ai, - ADJ_NBR_REWRITE_FLAG_INCOMPLETE, - ethernet_build_rewrite - (vnm, - sw_if_index, - VNET_LINK_ARP, - VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST)); - - /* - * since the FIB has added this adj for a route, it makes sense it - * may want to forward traffic sometime soon. Let's send a - * speculative ARP. just one. If we were to do periodically that - * wouldn't be bad either, but that's more code than i'm prepared to - * write at this time for relatively little reward. - */ - arp_nbr_probe (adj); - } - break; - case IP_LOOKUP_NEXT_BCAST: - adj_nbr_update_rewrite (ai, - ADJ_NBR_REWRITE_FLAG_COMPLETE, - ethernet_build_rewrite - (vnm, - sw_if_index, - VNET_LINK_IP4, - VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST)); - break; - case IP_LOOKUP_NEXT_MCAST: - { - /* - * Construct a partial rewrite from the known ethernet mcast dest MAC - */ - u8 *rewrite; - u8 offset; - - rewrite = ethernet_build_rewrite (vnm, - sw_if_index, - adj->ia_link, - ethernet_ip4_mcast_dst_addr ()); - offset = vec_len (rewrite) - 2; - - /* - * Complete the remaining fields of the adj's rewrite to direct the - * complete of the rewrite at switch time by copying in the IP - * dst address's bytes. - * Offset is 2 bytes into the MAC destination address. - */ - adj_mcast_update_rewrite (ai, rewrite, offset); - - break; - } - case IP_LOOKUP_NEXT_DROP: - case IP_LOOKUP_NEXT_PUNT: - case IP_LOOKUP_NEXT_LOCAL: - case IP_LOOKUP_NEXT_REWRITE: - case IP_LOOKUP_NEXT_MCAST_MIDCHAIN: - case IP_LOOKUP_NEXT_MIDCHAIN: - case IP_LOOKUP_NEXT_ICMP_ERROR: - case IP_LOOKUP_N_NEXT: - ASSERT (0); - break; - } -} - -static void -arp_adj_fib_add (ethernet_arp_ip4_entry_t * e, u32 fib_index) -{ - fib_prefix_t pfx = { - .fp_len = 32, - .fp_proto = FIB_PROTOCOL_IP4, - .fp_addr.ip4 = e->ip4_address, - }; - - e->fib_entry_index = - fib_table_entry_path_add (fib_index, &pfx, FIB_SOURCE_ADJ, - FIB_ENTRY_FLAG_ATTACHED, - DPO_PROTO_IP4, &pfx.fp_addr, - e->sw_if_index, ~0, 1, NULL, - FIB_ROUTE_PATH_FLAG_NONE); - fib_table_lock (fib_index, FIB_PROTOCOL_IP4, FIB_SOURCE_ADJ); -} - -static void -arp_adj_fib_remove (ethernet_arp_ip4_entry_t * e, u32 fib_index) -{ - if (FIB_NODE_INDEX_INVALID != e->fib_entry_index) - { - fib_prefix_t pfx = { - .fp_len = 32, - .fp_proto = FIB_PROTOCOL_IP4, - .fp_addr.ip4 = e->ip4_address, - }; - u32 fib_index; - - fib_index = ip4_fib_table_get_index_for_sw_if_index (e->sw_if_index); - - fib_table_entry_path_remove (fib_index, &pfx, - FIB_SOURCE_ADJ, - DPO_PROTO_IP4, - &pfx.fp_addr, - e->sw_if_index, ~0, 1, - FIB_ROUTE_PATH_FLAG_NONE); - fib_table_unlock (fib_index, FIB_PROTOCOL_IP4, FIB_SOURCE_ADJ); - } -} - -static ethernet_arp_ip4_entry_t * -force_reuse_arp_entry (void) -{ - ethernet_arp_ip4_entry_t *e; - ethernet_arp_main_t *am = ðernet_arp_main; - u32 count = 0; - u32 index = pool_next_index (am->ip4_entry_pool, am->arp_delete_rotor); - if (index == ~0) /* Try again from elt 0 */ - index = pool_next_index (am->ip4_entry_pool, index); - - /* Find a non-static random entry to free up for reuse */ - do - { - if ((count++ == 100) || (index == ~0)) - return NULL; /* give up after 100 entries */ - e = pool_elt_at_index (am->ip4_entry_pool, index); - am->arp_delete_rotor = index; - index = pool_next_index (am->ip4_entry_pool, index); - } - while (e->flags & IP_NEIGHBOR_FLAG_STATIC); - - /* Remove ARP entry from its interface and update fib */ - hash_unset - (am->ethernet_arp_by_sw_if_index[e->sw_if_index].arp_entries, - e->ip4_address.as_u32); - arp_adj_fib_remove - (e, ip4_fib_table_get_index_for_sw_if_index (e->sw_if_index)); - adj_nbr_walk_nh4 (e->sw_if_index, - &e->ip4_address, arp_mk_incomplete_walk, e); - return e; -} - -static int -vnet_arp_set_ip4_over_ethernet_internal (vnet_main_t * vnm, - vnet_arp_set_ip4_over_ethernet_rpc_args_t - * args) -{ - ethernet_arp_ip4_entry_t *e = 0; - ethernet_arp_main_t *am = ðernet_arp_main; - vlib_main_t *vm = vlib_get_main (); - int make_new_arp_cache_entry = 1; - uword *p; - pending_resolution_t *pr, *mc; - ethernet_arp_interface_t *arp_int; - u32 sw_if_index = args->sw_if_index; - - arp_enable (am, sw_if_index); - - arp_int = &am->ethernet_arp_by_sw_if_index[sw_if_index]; - - if (NULL != arp_int->arp_entries) - { - p = hash_get (arp_int->arp_entries, args->ip4.as_u32); - if (p) - { - e = pool_elt_at_index (am->ip4_entry_pool, p[0]); - - /* Refuse to over-write static arp. */ - if (!(args->nbr_flags & IP_NEIGHBOR_FLAG_STATIC) && - (e->flags & IP_NEIGHBOR_FLAG_STATIC)) - { - /* if MAC address match, still check to send event */ - if (mac_address_equal (&e->mac, &args->mac)) - goto check_customers; - return -2; - } - make_new_arp_cache_entry = 0; - } - } - - if (make_new_arp_cache_entry) - { - if (am->limit_arp_cache_size && - pool_elts (am->ip4_entry_pool) >= am->limit_arp_cache_size) - { - e = force_reuse_arp_entry (); - if (NULL == e) - return -2; - } - else - pool_get (am->ip4_entry_pool, e); - - if (NULL == arp_int->arp_entries) - arp_int->arp_entries = hash_create (0, sizeof (u32)); - - hash_set (arp_int->arp_entries, args->ip4.as_u32, - e - am->ip4_entry_pool); - - e->sw_if_index = sw_if_index; - e->ip4_address = args->ip4; - e->fib_entry_index = FIB_NODE_INDEX_INVALID; - mac_address_copy (&e->mac, &args->mac); - - if (!(args->nbr_flags & IP_NEIGHBOR_FLAG_NO_FIB_ENTRY)) - { - arp_adj_fib_add (e, - ip4_fib_table_get_index_for_sw_if_index - (e->sw_if_index)); - } - else - { - e->flags |= IP_NEIGHBOR_FLAG_NO_FIB_ENTRY; - } - } - else - { - /* - * prevent a DoS attack from the data-plane that - * spams us with no-op updates to the MAC address - */ - if (mac_address_equal (&e->mac, &args->mac)) - { - e->time_last_updated = vlib_time_now (vm); - goto check_customers; - } - - /* Update ethernet address. */ - mac_address_copy (&e->mac, &args->mac); - } - - /* Update time stamp and flags. */ - e->time_last_updated = vlib_time_now (vm); - if (args->nbr_flags & IP_NEIGHBOR_FLAG_STATIC) - { - e->flags &= ~IP_NEIGHBOR_FLAG_DYNAMIC; - e->flags |= IP_NEIGHBOR_FLAG_STATIC; - } - else - { - e->flags &= ~IP_NEIGHBOR_FLAG_STATIC; - e->flags |= IP_NEIGHBOR_FLAG_DYNAMIC; - } - - adj_nbr_walk_nh4 (sw_if_index, &e->ip4_address, arp_mk_complete_walk, e); - -check_customers: - /* Customer(s) waiting for this address to be resolved? */ - p = hash_get (am->pending_resolutions_by_address, args->ip4.as_u32); - if (p) - { - u32 next_index; - next_index = p[0]; - - while (next_index != (u32) ~ 0) - { - pr = pool_elt_at_index (am->pending_resolutions, next_index); - vlib_process_signal_event (vm, pr->node_index, - pr->type_opaque, pr->data); - next_index = pr->next_index; - pool_put (am->pending_resolutions, pr); - } - - hash_unset (am->pending_resolutions_by_address, args->ip4.as_u32); - } - - /* Customer(s) requesting ARP event for this address? */ - p = hash_get (am->mac_changes_by_address, args->ip4.as_u32); - if (p) - { - u32 next_index; - next_index = p[0]; - - while (next_index != (u32) ~ 0) - { - int rv = 1; - mc = pool_elt_at_index (am->mac_changes, next_index); - - /* Call the user's data callback, return 1 to suppress dup events */ - if (mc->data_callback) - rv = (mc->data_callback) (mc->data, &args->mac, sw_if_index, 0); - - /* - * Signal the resolver process, as long as the user - * says they want to be notified - */ - if (rv == 0) - vlib_process_signal_event (vm, mc->node_index, - mc->type_opaque, mc->data); - next_index = mc->next_index; - } - } - - return 0; -} - -void -vnet_register_ip4_arp_resolution_event (vnet_main_t * vnm, - void *address_arg, - uword node_index, - uword type_opaque, uword data) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ip4_address_t *address = address_arg; - uword *p; - pending_resolution_t *pr; - - pool_get (am->pending_resolutions, pr); - - pr->next_index = ~0; - pr->node_index = node_index; - pr->type_opaque = type_opaque; - pr->data = data; - pr->data_callback = 0; - - p = hash_get (am->pending_resolutions_by_address, address->as_u32); - if (p) - { - /* Insert new resolution at the head of the list */ - pr->next_index = p[0]; - hash_unset (am->pending_resolutions_by_address, address->as_u32); - } - - hash_set (am->pending_resolutions_by_address, address->as_u32, - pr - am->pending_resolutions); -} - -int -vnet_add_del_ip4_arp_change_event (vnet_main_t * vnm, - arp_change_event_cb_t data_callback, - u32 pid, - void *address_arg, - uword node_index, - uword type_opaque, uword data, int is_add) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ip4_address_t *address = address_arg; - - /* Try to find an existing entry */ - u32 *first = (u32 *) hash_get (am->mac_changes_by_address, address->as_u32); - u32 *p = first; - pending_resolution_t *mc; - while (p && *p != ~0) - { - mc = pool_elt_at_index (am->mac_changes, *p); - if (mc->node_index == node_index && mc->type_opaque == type_opaque - && mc->pid == pid) - break; - p = &mc->next_index; - } - - int found = p && *p != ~0; - if (is_add) - { - if (found) - return VNET_API_ERROR_ENTRY_ALREADY_EXISTS; - - pool_get (am->mac_changes, mc); - /* *INDENT-OFF* */ - *mc = (pending_resolution_t) - { - .next_index = ~0, - .node_index = node_index, - .type_opaque = type_opaque, - .data = data, - .data_callback = data_callback, - .pid = pid, - }; - /* *INDENT-ON* */ - - /* Insert new resolution at the end of the list */ - u32 new_idx = mc - am->mac_changes; - if (p) - p[0] = new_idx; - else - hash_set (am->mac_changes_by_address, address->as_u32, new_idx); - } - else - { - if (!found) - return VNET_API_ERROR_NO_SUCH_ENTRY; - - /* Clients may need to clean up pool entries, too */ - if (data_callback) - /* no new mac addrs */ - (data_callback) (mc->data, NULL, ~0, NULL); - - /* Remove the entry from the list and delete the entry */ - *p = mc->next_index; - pool_put (am->mac_changes, mc); - - /* Remove from hash if we deleted the last entry */ - if (*p == ~0 && p == first) - hash_unset (am->mac_changes_by_address, address->as_u32); - } - return 0; -} - -/* Either we drop the packet or we send a reply to the sender. */ -typedef enum -{ - ARP_REPLY_NEXT_DROP, - ARP_REPLY_NEXT_REPLY_TX, - ARP_REPLY_N_NEXT, -} arp_reply_next_t; - -#define foreach_ethernet_arp_error \ - _ (replies_sent, "ARP replies sent") \ - _ (l2_type_not_ethernet, "L2 type not ethernet") \ - _ (l3_type_not_ip4, "L3 type not IP4") \ - _ (l3_src_address_not_local, "IP4 source address not local to subnet") \ - _ (l3_dst_address_not_local, "IP4 destination address not local to subnet") \ - _ (l3_dst_address_unset, "IP4 destination address is unset") \ - _ (l3_src_address_is_local, "IP4 source address matches local interface") \ - _ (l3_src_address_learned, "ARP request IP4 source address learned") \ - _ (replies_received, "ARP replies received") \ - _ (opcode_not_request, "ARP opcode not request") \ - _ (proxy_arp_replies_sent, "Proxy ARP replies sent") \ - _ (l2_address_mismatch, "ARP hw addr does not match L2 frame src addr") \ - _ (gratuitous_arp, "ARP probe or announcement dropped") \ - _ (interface_no_table, "Interface is not mapped to an IP table") \ - _ (interface_not_ip_enabled, "Interface is not IP enabled") \ - _ (unnumbered_mismatch, "RX interface is unnumbered to different subnet") \ - -typedef enum -{ -#define _(sym,string) ETHERNET_ARP_ERROR_##sym, - foreach_ethernet_arp_error -#undef _ - ETHERNET_ARP_N_ERROR, -} ethernet_arp_reply_error_t; - -static int -arp_unnumbered (vlib_buffer_t * p0, - u32 input_sw_if_index, u32 conn_sw_if_index) -{ - vnet_main_t *vnm = vnet_get_main (); - vnet_interface_main_t *vim = &vnm->interface_main; - vnet_sw_interface_t *si; - - /* verify that the input interface is unnumbered to the connected. - * the connected interface is the interface on which the subnet is - * configured */ - si = &vim->sw_interfaces[input_sw_if_index]; - - if (!(si->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED && - (si->unnumbered_sw_if_index == conn_sw_if_index))) - { - /* the input interface is not unnumbered to the interface on which - * the sub-net is configured that covers the ARP request. - * So this is not the case for unnumbered.. */ - return 0; - } - - return !0; -} - -static u32 -arp_learn (vnet_main_t * vnm, - ethernet_arp_main_t * am, u32 sw_if_index, - const ethernet_arp_ip4_over_ethernet_address_t * addr) -{ - vnet_arp_set_ip4_over_ethernet (vnm, sw_if_index, addr, 0); - return (ETHERNET_ARP_ERROR_l3_src_address_learned); -} - -typedef enum arp_input_next_t_ -{ - ARP_INPUT_NEXT_DROP, - ARP_INPUT_NEXT_DISABLED, - ARP_INPUT_N_NEXT, -} arp_input_next_t; - -static uword -arp_input (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) -{ - u32 n_left_from, next_index, *from, *to_next, n_left_to_next; - ethernet_arp_main_t *am = ðernet_arp_main; - - from = vlib_frame_vector_args (frame); - n_left_from = frame->n_vectors; - next_index = node->cached_next_index; - - if (node->flags & VLIB_NODE_FLAG_TRACE) - vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors, - /* stride */ 1, - sizeof (ethernet_arp_input_trace_t)); - - 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) - { - const ethernet_arp_header_t *arp0; - arp_input_next_t next0; - vlib_buffer_t *p0; - u32 pi0, error0; - - pi0 = to_next[0] = from[0]; - from += 1; - to_next += 1; - n_left_from -= 1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer (vm, pi0); - arp0 = vlib_buffer_get_current (p0); - - error0 = ETHERNET_ARP_ERROR_replies_sent; - next0 = ARP_INPUT_NEXT_DROP; - - error0 = - (arp0->l2_type != - clib_net_to_host_u16 (ETHERNET_ARP_HARDWARE_TYPE_ethernet) ? - ETHERNET_ARP_ERROR_l2_type_not_ethernet : error0); - error0 = - (arp0->l3_type != - clib_net_to_host_u16 (ETHERNET_TYPE_IP4) ? - ETHERNET_ARP_ERROR_l3_type_not_ip4 : error0); - error0 = - (0 == arp0->ip4_over_ethernet[0].ip4.as_u32 ? - ETHERNET_ARP_ERROR_l3_dst_address_unset : error0); - - if (ETHERNET_ARP_ERROR_replies_sent == error0) - { - next0 = ARP_INPUT_NEXT_DISABLED; - vnet_feature_arc_start (am->feature_arc_index, - vnet_buffer (p0)->sw_if_index[VLIB_RX], - &next0, p0); - } - else - p0->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; -} - -typedef enum arp_disabled_next_t_ -{ - ARP_DISABLED_NEXT_DROP, - ARP_DISABLED_N_NEXT, -} arp_disabled_next_t; - -#define foreach_arp_disabled_error \ - _ (DISABLED, "ARP Disabled on this interface") \ - -typedef enum -{ -#define _(sym,string) ARP_DISABLED_ERROR_##sym, - foreach_arp_disabled_error -#undef _ - ARP_DISABLED_N_ERROR, -} arp_disabled_error_t; - -static char *arp_disabled_error_strings[] = { -#define _(sym,string) string, - foreach_arp_disabled_error -#undef _ -}; - -static uword -arp_disabled (vlib_main_t * vm, - vlib_node_runtime_t * node, vlib_frame_t * frame) -{ - u32 n_left_from, next_index, *from, *to_next, n_left_to_next; - - from = vlib_frame_vector_args (frame); - n_left_from = frame->n_vectors; - next_index = node->cached_next_index; - - if (node->flags & VLIB_NODE_FLAG_TRACE) - vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors, - /* stride */ 1, - sizeof (ethernet_arp_input_trace_t)); - - 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) - { - arp_disabled_next_t next0 = ARP_DISABLED_NEXT_DROP; - vlib_buffer_t *p0; - u32 pi0, error0; - - next0 = ARP_DISABLED_NEXT_DROP; - error0 = ARP_DISABLED_ERROR_DISABLED; - - pi0 = to_next[0] = from[0]; - from += 1; - to_next += 1; - n_left_from -= 1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer (vm, pi0); - p0->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_always_inline u32 -arp_mk_reply (vnet_main_t * vnm, - vlib_buffer_t * p0, - u32 sw_if_index0, - const ip4_address_t * if_addr0, - ethernet_arp_header_t * arp0, ethernet_header_t * eth_rx) -{ - vnet_hw_interface_t *hw_if0; - u8 *rewrite0, rewrite0_len; - ethernet_header_t *eth_tx; - u32 next0; - - /* Send a reply. - An adjacency to the sender is not always present, - so we use the interface to build us a rewrite string - which will contain all the necessary tags. */ - rewrite0 = ethernet_build_rewrite (vnm, sw_if_index0, - VNET_LINK_ARP, eth_rx->src_address); - rewrite0_len = vec_len (rewrite0); - - /* Figure out how much to rewind current data from adjacency. */ - vlib_buffer_advance (p0, -rewrite0_len); - eth_tx = vlib_buffer_get_current (p0); - - vnet_buffer (p0)->sw_if_index[VLIB_TX] = sw_if_index0; - hw_if0 = vnet_get_sup_hw_interface (vnm, sw_if_index0); - - /* Send reply back through input interface */ - vnet_buffer (p0)->sw_if_index[VLIB_TX] = sw_if_index0; - next0 = ARP_REPLY_NEXT_REPLY_TX; - - arp0->opcode = clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_reply); - - arp0->ip4_over_ethernet[1] = arp0->ip4_over_ethernet[0]; - - mac_address_from_bytes (&arp0->ip4_over_ethernet[0].mac, - hw_if0->hw_address); - clib_mem_unaligned (&arp0->ip4_over_ethernet[0].ip4.data_u32, u32) = - if_addr0->data_u32; - - /* Hardware must be ethernet-like. */ - ASSERT (vec_len (hw_if0->hw_address) == 6); - - /* the rx nd tx ethernet headers wil overlap in the case - * when we received a tagged VLAN=0 packet, but we are sending - * back untagged */ - clib_memcpy_fast (eth_tx, rewrite0, vec_len (rewrite0)); - vec_free (rewrite0); - - return (next0); -} - -enum arp_dst_fib_type -{ - ARP_DST_FIB_NONE, - ARP_DST_FIB_ADJ, - ARP_DST_FIB_CONN -}; - -/* - * we're looking for FIB sources that indicate the destination - * is attached. There may be interposed DPO prior to the one - * we are looking for - */ -static enum arp_dst_fib_type -arp_dst_fib_check (const fib_node_index_t fei, fib_entry_flag_t * flags) -{ - const fib_entry_t *entry = fib_entry_get (fei); - const fib_entry_src_t *entry_src; - fib_source_t src; - /* *INDENT-OFF* */ - FOR_EACH_SRC_ADDED(entry, entry_src, src, - ({ - *flags = fib_entry_get_flags_for_source (fei, src); - if (fib_entry_is_sourced (fei, FIB_SOURCE_ADJ)) - return ARP_DST_FIB_ADJ; - else if (FIB_ENTRY_FLAG_CONNECTED & *flags) - return ARP_DST_FIB_CONN; - })) - /* *INDENT-ON* */ - - return ARP_DST_FIB_NONE; -} - -static uword -arp_reply (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - vnet_main_t *vnm = vnet_get_main (); - u32 n_left_from, next_index, *from, *to_next; - u32 n_replies_sent = 0; - - from = vlib_frame_vector_args (frame); - n_left_from = frame->n_vectors; - next_index = node->cached_next_index; - - if (node->flags & VLIB_NODE_FLAG_TRACE) - vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors, - /* stride */ 1, - sizeof (ethernet_arp_input_trace_t)); - - while (n_left_from > 0) - { - u32 n_left_to_next; - - vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); - - while (n_left_from > 0 && n_left_to_next > 0) - { - vlib_buffer_t *p0; - ethernet_arp_header_t *arp0; - ethernet_header_t *eth_rx; - const ip4_address_t *if_addr0; - u32 pi0, error0, next0, sw_if_index0, conn_sw_if_index0, fib_index0; - u8 dst_is_local0, is_vrrp_reply0; - fib_node_index_t dst_fei, src_fei; - const fib_prefix_t *pfx0; - fib_entry_flag_t src_flags, dst_flags; - - pi0 = from[0]; - to_next[0] = pi0; - from += 1; - to_next += 1; - n_left_from -= 1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer (vm, pi0); - arp0 = vlib_buffer_get_current (p0); - /* Fill in ethernet header. */ - eth_rx = ethernet_buffer_get_header (p0); - - next0 = ARP_REPLY_NEXT_DROP; - error0 = ETHERNET_ARP_ERROR_replies_sent; - sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX]; - - /* Check that IP address is local and matches incoming interface. */ - fib_index0 = ip4_fib_table_get_index_for_sw_if_index (sw_if_index0); - if (~0 == fib_index0) - { - error0 = ETHERNET_ARP_ERROR_interface_no_table; - goto drop; - - } - - { - /* - * we're looking for FIB entries that indicate the source - * is attached. There may be more specific non-attached - * routes that match the source, but these do not influence - * whether we respond to an ARP request, i.e. they do not - * influence whether we are the correct way for the sender - * to reach us, they only affect how we reach the sender. - */ - fib_entry_t *src_fib_entry; - const fib_prefix_t *pfx; - fib_entry_src_t *src; - fib_source_t source; - int attached; - int mask; - - mask = 32; - attached = 0; - - do - { - src_fei = ip4_fib_table_lookup (ip4_fib_get (fib_index0), - &arp0-> - ip4_over_ethernet[0].ip4, - mask); - src_fib_entry = fib_entry_get (src_fei); - - /* - * It's possible that the source that provides the - * flags we need, or the flags we must not have, - * is not the best source, so check then all. - */ - /* *INDENT-OFF* */ - FOR_EACH_SRC_ADDED(src_fib_entry, src, source, - ({ - src_flags = fib_entry_get_flags_for_source (src_fei, source); - - /* Reject requests/replies with our local interface - address. */ - if (FIB_ENTRY_FLAG_LOCAL & src_flags) - { - error0 = ETHERNET_ARP_ERROR_l3_src_address_is_local; - /* - * When VPP has an interface whose address is also - * applied to a TAP interface on the host, then VPP's - * TAP interface will be unnumbered to the 'real' - * interface and do proxy ARP from the host. - * The curious aspect of this setup is that ARP requests - * from the host will come from the VPP's own address. - * So don't drop immediately here, instead go see if this - * is a proxy ARP case. - */ - goto next_feature; - } - /* A Source must also be local to subnet of matching - * interface address. */ - if ((FIB_ENTRY_FLAG_ATTACHED & src_flags) || - (FIB_ENTRY_FLAG_CONNECTED & src_flags)) - { - attached = 1; - break; - } - /* - * else - * The packet was sent from an address that is not - * connected nor attached i.e. it is not from an - * address that is covered by a link's sub-net, - * nor is it a already learned host resp. - */ - })); - /* *INDENT-ON* */ - - /* - * shorter mask lookup for the next iteration. - */ - pfx = fib_entry_get_prefix (src_fei); - mask = pfx->fp_len - 1; - - /* - * continue until we hit the default route or we find - * the attached we are looking for. The most likely - * outcome is we find the attached with the first source - * on the first lookup. - */ - } - while (!attached && - !fib_entry_is_sourced (src_fei, FIB_SOURCE_DEFAULT_ROUTE)); - - if (!attached) - { - /* - * the matching route is a not attached, i.e. it was - * added as a result of routing, rather than interface/ARP - * configuration. If the matching route is not a host route - * (i.e. a /32) - */ - error0 = ETHERNET_ARP_ERROR_l3_src_address_not_local; - goto drop; - } - } - - dst_fei = ip4_fib_table_lookup (ip4_fib_get (fib_index0), - &arp0->ip4_over_ethernet[1].ip4, - 32); - switch (arp_dst_fib_check (dst_fei, &dst_flags)) - { - case ARP_DST_FIB_ADJ: - /* - * We matched an adj-fib on ths source subnet (a /32 previously - * added as a result of ARP). If this request is a gratuitous - * ARP, then learn from it. - * The check for matching an adj-fib, is to prevent hosts - * from spamming us with gratuitous ARPS that might otherwise - * blow our ARP cache - */ - if (arp0->ip4_over_ethernet[0].ip4.as_u32 == - arp0->ip4_over_ethernet[1].ip4.as_u32) - error0 = arp_learn (vnm, am, sw_if_index0, - &arp0->ip4_over_ethernet[0]); - goto drop; - case ARP_DST_FIB_CONN: - /* destination is connected, continue to process */ - break; - case ARP_DST_FIB_NONE: - /* destination is not connected, stop here */ - error0 = ETHERNET_ARP_ERROR_l3_dst_address_not_local; - goto next_feature; - } - - dst_is_local0 = (FIB_ENTRY_FLAG_LOCAL & dst_flags); - pfx0 = fib_entry_get_prefix (dst_fei); - if_addr0 = &pfx0->fp_addr.ip4; - - is_vrrp_reply0 = - ((arp0->opcode == - clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_reply)) - && - (!memcmp - (arp0->ip4_over_ethernet[0].mac.bytes, vrrp_prefix, - sizeof (vrrp_prefix)))); - - /* Trash ARP packets whose ARP-level source addresses do not - match their L2-frame-level source addresses, unless it's - a reply from a VRRP virtual router */ - if (!ethernet_mac_address_equal - (eth_rx->src_address, - arp0->ip4_over_ethernet[0].mac.bytes) && !is_vrrp_reply0) - { - error0 = ETHERNET_ARP_ERROR_l2_address_mismatch; - goto drop; - } - - /* Learn or update sender's mapping only for replies to addresses - * that are local to the subnet */ - if (arp0->opcode == - clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_reply)) - { - if (dst_is_local0) - error0 = arp_learn (vnm, am, sw_if_index0, - &arp0->ip4_over_ethernet[0]); - else - /* a reply for a non-local destination could be a GARP. - * GARPs for hosts we know were handled above, so this one - * we drop */ - error0 = ETHERNET_ARP_ERROR_l3_dst_address_not_local; - - goto next_feature; - } - else if (arp0->opcode == - clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_request) && - (dst_is_local0 == 0)) - { - goto next_feature; - } - - /* Honor unnumbered interface, if any */ - conn_sw_if_index0 = fib_entry_get_resolving_interface (dst_fei); - if (sw_if_index0 != conn_sw_if_index0 || - sw_if_index0 != fib_entry_get_resolving_interface (src_fei)) - { - /* - * The interface the ARP is sent to or was received on is not the - * interface on which the covering prefix is configured. - * Maybe this is a case for unnumbered. - */ - if (!arp_unnumbered (p0, sw_if_index0, conn_sw_if_index0)) - { - error0 = ETHERNET_ARP_ERROR_unnumbered_mismatch; - goto drop; - } - } - if (arp0->ip4_over_ethernet[0].ip4.as_u32 == - arp0->ip4_over_ethernet[1].ip4.as_u32) - { - error0 = ETHERNET_ARP_ERROR_gratuitous_arp; - goto drop; - } - - next0 = arp_mk_reply (vnm, p0, sw_if_index0, - if_addr0, arp0, eth_rx); - - /* We are going to reply to this request, so, in the absence of - errors, learn the sender */ - if (!error0) - error0 = arp_learn (vnm, am, sw_if_index0, - &arp0->ip4_over_ethernet[1]); - - n_replies_sent += 1; - goto enqueue; - - next_feature: - vnet_feature_next (&next0, p0); - goto enqueue; - - drop: - p0->error = node->errors[error0]; - - enqueue: - 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); - } - - vlib_error_count (vm, node->node_index, - ETHERNET_ARP_ERROR_replies_sent, n_replies_sent); - - return frame->n_vectors; -} - -static uword -arp_proxy (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - vnet_main_t *vnm = vnet_get_main (); - u32 n_left_from, next_index, *from, *to_next; - u32 n_arp_replies_sent = 0; - - from = vlib_frame_vector_args (frame); - n_left_from = frame->n_vectors; - next_index = node->cached_next_index; - - if (node->flags & VLIB_NODE_FLAG_TRACE) - vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors, - /* stride */ 1, - sizeof (ethernet_arp_input_trace_t)); - - while (n_left_from > 0) - { - u32 n_left_to_next; - - vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); - - while (n_left_from > 0 && n_left_to_next > 0) - { - vlib_buffer_t *p0; - ethernet_arp_header_t *arp0; - ethernet_header_t *eth_rx; - ip4_address_t proxy_src; - u32 pi0, error0, next0, sw_if_index0, fib_index0; - u8 is_request0; - ethernet_proxy_arp_t *pa; - - pi0 = from[0]; - to_next[0] = pi0; - from += 1; - to_next += 1; - n_left_from -= 1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer (vm, pi0); - arp0 = vlib_buffer_get_current (p0); - /* Fill in ethernet header. */ - eth_rx = ethernet_buffer_get_header (p0); - - is_request0 = arp0->opcode - == clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_request); - - error0 = ETHERNET_ARP_ERROR_replies_sent; - sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX]; - next0 = ARP_REPLY_NEXT_DROP; - - fib_index0 = ip4_fib_table_get_index_for_sw_if_index (sw_if_index0); - if (~0 == fib_index0) - { - error0 = ETHERNET_ARP_ERROR_interface_no_table; - } - - if (0 == error0 && is_request0) - { - u32 this_addr = clib_net_to_host_u32 - (arp0->ip4_over_ethernet[1].ip4.as_u32); - - vec_foreach (pa, am->proxy_arps) - { - u32 lo_addr = clib_net_to_host_u32 (pa->lo_addr.as_u32); - u32 hi_addr = clib_net_to_host_u32 (pa->hi_addr.as_u32); - - /* an ARP request hit in the proxy-arp table? */ - if ((this_addr >= lo_addr && this_addr <= hi_addr) && - (fib_index0 == pa->fib_index)) - { - proxy_src.as_u32 = - arp0->ip4_over_ethernet[1].ip4.data_u32; - - /* - * change the interface address to the proxied - */ - n_arp_replies_sent++; - - next0 = - arp_mk_reply (vnm, p0, sw_if_index0, &proxy_src, arp0, - eth_rx); - } - } - } - else - { - p0->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); - } - - vlib_error_count (vm, node->node_index, - ETHERNET_ARP_ERROR_replies_sent, n_arp_replies_sent); - - return frame->n_vectors; -} - -static char *ethernet_arp_error_strings[] = { -#define _(sym,string) string, - foreach_ethernet_arp_error -#undef _ -}; - -/* *INDENT-OFF* */ - -VLIB_REGISTER_NODE (arp_input_node, static) = -{ - .function = arp_input, - .name = "arp-input", - .vector_size = sizeof (u32), - .n_errors = ETHERNET_ARP_N_ERROR, - .error_strings = ethernet_arp_error_strings, - .n_next_nodes = ARP_INPUT_N_NEXT, - .next_nodes = { - [ARP_INPUT_NEXT_DROP] = "error-drop", - [ARP_INPUT_NEXT_DISABLED] = "arp-disabled", - }, - .format_buffer = format_ethernet_arp_header, - .format_trace = format_ethernet_arp_input_trace, -}; - -VLIB_REGISTER_NODE (arp_disabled_node, static) = -{ - .function = arp_disabled, - .name = "arp-disabled", - .vector_size = sizeof (u32), - .n_errors = ARP_DISABLED_N_ERROR, - .error_strings = arp_disabled_error_strings, - .n_next_nodes = ARP_DISABLED_N_NEXT, - .next_nodes = { - [ARP_INPUT_NEXT_DROP] = "error-drop", - }, - .format_buffer = format_ethernet_arp_header, - .format_trace = format_ethernet_arp_input_trace, -}; - -VLIB_REGISTER_NODE (arp_reply_node, static) = -{ - .function = arp_reply, - .name = "arp-reply", - .vector_size = sizeof (u32), - .n_errors = ETHERNET_ARP_N_ERROR, - .error_strings = ethernet_arp_error_strings, - .n_next_nodes = ARP_REPLY_N_NEXT, - .next_nodes = { - [ARP_REPLY_NEXT_DROP] = "error-drop", - [ARP_REPLY_NEXT_REPLY_TX] = "interface-output", - }, - .format_buffer = format_ethernet_arp_header, - .format_trace = format_ethernet_arp_input_trace, -}; - -VLIB_REGISTER_NODE (arp_proxy_node, static) = -{ - .function = arp_proxy, - .name = "arp-proxy", - .vector_size = sizeof (u32), - .n_errors = ETHERNET_ARP_N_ERROR, - .error_strings = ethernet_arp_error_strings, - .n_next_nodes = ARP_REPLY_N_NEXT, - .next_nodes = { - [ARP_REPLY_NEXT_DROP] = "error-drop", - [ARP_REPLY_NEXT_REPLY_TX] = "interface-output", - }, - .format_buffer = format_ethernet_arp_header, - .format_trace = format_ethernet_arp_input_trace, -}; - -/* Built-in ARP rx feature path definition */ -VNET_FEATURE_ARC_INIT (arp_feat, static) = -{ - .arc_name = "arp", - .start_nodes = VNET_FEATURES ("arp-input"), - .last_in_arc = "error-drop", - .arc_index_ptr = ðernet_arp_main.feature_arc_index, -}; - -VNET_FEATURE_INIT (arp_reply_feat_node, static) = -{ - .arc_name = "arp", - .node_name = "arp-reply", - .runs_before = VNET_FEATURES ("arp-disabled"), -}; - -VNET_FEATURE_INIT (arp_proxy_feat_node, static) = -{ - .arc_name = "arp", - .node_name = "arp-proxy", - .runs_after = VNET_FEATURES ("arp-reply"), - .runs_before = VNET_FEATURES ("arp-disabled"), -}; - -VNET_FEATURE_INIT (arp_disabled_feat_node, static) = -{ - .arc_name = "arp", - .node_name = "arp-disabled", - .runs_before = VNET_FEATURES ("error-drop"), -}; - -VNET_FEATURE_INIT (arp_drop_feat_node, static) = -{ - .arc_name = "arp", - .node_name = "error-drop", - .runs_before = 0, /* last feature */ -}; - -/* *INDENT-ON* */ - -static int -ip4_arp_entry_sort (void *a1, void *a2) -{ - ethernet_arp_ip4_entry_t *e1 = a1; - ethernet_arp_ip4_entry_t *e2 = a2; - - int cmp; - vnet_main_t *vnm = vnet_get_main (); - - cmp = vnet_sw_interface_compare (vnm, e1->sw_if_index, e2->sw_if_index); - if (!cmp) - cmp = ip4_address_compare (&e1->ip4_address, &e2->ip4_address); - return cmp; -} - -ethernet_arp_ip4_entry_t * -ip4_neighbors_pool (void) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - return am->ip4_entry_pool; -} - -ethernet_arp_ip4_entry_t * -ip4_neighbor_entries (u32 sw_if_index) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_arp_ip4_entry_t *n, *ns = 0; - - /* *INDENT-OFF* */ - pool_foreach (n, am->ip4_entry_pool, ({ - if (sw_if_index != ~0 && n->sw_if_index != sw_if_index) - continue; - vec_add1 (ns, n[0]); - })); - /* *INDENT-ON* */ - - if (ns) - vec_sort_with_function (ns, ip4_arp_entry_sort); - return ns; -} - -static clib_error_t * -show_ip4_arp (vlib_main_t * vm, - unformat_input_t * input, vlib_cli_command_t * cmd) -{ - vnet_main_t *vnm = vnet_get_main (); - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_arp_ip4_entry_t *e, *es; - ethernet_proxy_arp_t *pa; - clib_error_t *error = 0; - u32 sw_if_index; - - /* Filter entries by interface if given. */ - sw_if_index = ~0; - (void) unformat_user (input, unformat_vnet_sw_interface, vnm, &sw_if_index); - - es = ip4_neighbor_entries (sw_if_index); - if (es) - { - vlib_cli_output (vm, "%U", format_ethernet_arp_ip4_entry, vnm, 0); - vec_foreach (e, es) - { - vlib_cli_output (vm, "%U", format_ethernet_arp_ip4_entry, vnm, e); - } - vec_free (es); - } - - if (vec_len (am->proxy_arps)) - { - vlib_cli_output (vm, "Proxy arps enabled for:"); - vec_foreach (pa, am->proxy_arps) - { - vlib_cli_output (vm, "Fib_index %d %U - %U ", - pa->fib_index, - format_ip4_address, &pa->lo_addr, - format_ip4_address, &pa->hi_addr); - } - } - - return error; -} - -/*? - * Display all the IPv4 ARP entries. - * - * @cliexpar - * Example of how to display the IPv4 ARP table: - * @cliexstart{show ip arp} - * Time FIB IP4 Flags Ethernet Interface - * 346.3028 0 6.1.1.3 de:ad:be:ef:ba:be GigabitEthernet2/0/0 - * 3077.4271 0 6.1.1.4 S de:ad:be:ef:ff:ff GigabitEthernet2/0/0 - * 2998.6409 1 6.2.2.3 de:ad:be:ef:00:01 GigabitEthernet2/0/0 - * Proxy arps enabled for: - * Fib_index 0 6.0.0.1 - 6.0.0.11 - * @cliexend - ?*/ -/* *INDENT-OFF* */ -VLIB_CLI_COMMAND (show_ip4_arp_command, static) = { - .path = "show ip arp", - .function = show_ip4_arp, - .short_help = "show ip arp", -}; -/* *INDENT-ON* */ - -typedef struct -{ - pg_edit_t l2_type, l3_type; - pg_edit_t n_l2_address_bytes, n_l3_address_bytes; - pg_edit_t opcode; - struct - { - pg_edit_t mac; - pg_edit_t ip4; - } ip4_over_ethernet[2]; -} pg_ethernet_arp_header_t; - -static inline void -pg_ethernet_arp_header_init (pg_ethernet_arp_header_t * p) -{ - /* Initialize fields that are not bit fields in the IP header. */ -#define _(f) pg_edit_init (&p->f, ethernet_arp_header_t, f); - _(l2_type); - _(l3_type); - _(n_l2_address_bytes); - _(n_l3_address_bytes); - _(opcode); - _(ip4_over_ethernet[0].mac); - _(ip4_over_ethernet[0].ip4); - _(ip4_over_ethernet[1].mac); - _(ip4_over_ethernet[1].ip4); -#undef _ -} - -uword -unformat_pg_arp_header (unformat_input_t * input, va_list * args) -{ - pg_stream_t *s = va_arg (*args, pg_stream_t *); - pg_ethernet_arp_header_t *p; - u32 group_index; - - p = pg_create_edit_group (s, sizeof (p[0]), sizeof (ethernet_arp_header_t), - &group_index); - pg_ethernet_arp_header_init (p); - - /* Defaults. */ - pg_edit_set_fixed (&p->l2_type, ETHERNET_ARP_HARDWARE_TYPE_ethernet); - pg_edit_set_fixed (&p->l3_type, ETHERNET_TYPE_IP4); - pg_edit_set_fixed (&p->n_l2_address_bytes, 6); - pg_edit_set_fixed (&p->n_l3_address_bytes, 4); - - if (!unformat (input, "%U: %U/%U -> %U/%U", - unformat_pg_edit, - unformat_ethernet_arp_opcode_net_byte_order, &p->opcode, - unformat_pg_edit, - unformat_mac_address_t, &p->ip4_over_ethernet[0].mac, - unformat_pg_edit, - unformat_ip4_address, &p->ip4_over_ethernet[0].ip4, - unformat_pg_edit, - unformat_mac_address_t, &p->ip4_over_ethernet[1].mac, - unformat_pg_edit, - unformat_ip4_address, &p->ip4_over_ethernet[1].ip4)) - { - /* Free up any edits we may have added. */ - pg_free_edit_group (s); - return 0; - } - return 1; -} - -clib_error_t * -ip4_set_arp_limit (u32 arp_limit) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - - am->limit_arp_cache_size = arp_limit; - return 0; -} - -/** - * @brief Control Plane hook to remove an ARP entry - */ -int -vnet_arp_unset_ip4_over_ethernet (vnet_main_t * vnm, - u32 sw_if_index, - const - ethernet_arp_ip4_over_ethernet_address_t * - a) -{ - vnet_arp_set_ip4_over_ethernet_rpc_args_t args = { - .sw_if_index = sw_if_index, - .flags = ETHERNET_ARP_ARGS_REMOVE, - .ip4 = a->ip4, - .mac = a->mac, - }; - - vl_api_rpc_call_main_thread (set_ip4_over_ethernet_rpc_callback, - (u8 *) & args, sizeof (args)); - return 0; -} - -/** - * @brief publish wildcard arp event - * @param sw_if_index The interface on which the ARP entries are acted - */ -static int -vnet_arp_wc_publish (u32 sw_if_index, - const ethernet_arp_ip4_over_ethernet_address_t * a) -{ - vnet_arp_set_ip4_over_ethernet_rpc_args_t args = { - .flags = ETHERNET_ARP_ARGS_WC_PUB, - .sw_if_index = sw_if_index, - .ip4 = a->ip4, - .mac = a->mac, - }; - - vl_api_rpc_call_main_thread (set_ip4_over_ethernet_rpc_callback, - (u8 *) & args, sizeof (args)); - return 0; -} - -static void -vnet_arp_wc_publish_internal (vnet_main_t * vnm, - vnet_arp_set_ip4_over_ethernet_rpc_args_t * - args) -{ - vlib_main_t *vm = vlib_get_main (); - ethernet_arp_main_t *am = ðernet_arp_main; - uword ni = am->wc_ip4_arp_publisher_node; - uword et = am->wc_ip4_arp_publisher_et; - - if (ni == (uword) ~ 0) - return; - wc_arp_report_t *r = - vlib_process_signal_event_data (vm, ni, et, 1, sizeof *r); - r->ip.as_u32 = args->ip4.as_u32; - r->sw_if_index = args->sw_if_index; - mac_address_copy (&r->mac, &args->mac); -} - -void -wc_arp_set_publisher_node (uword node_index, uword event_type) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - am->wc_ip4_arp_publisher_node = node_index; - am->wc_ip4_arp_publisher_et = event_type; -} - -static void -arp_entry_free (ethernet_arp_interface_t * eai, ethernet_arp_ip4_entry_t * e); - -static int -vnet_arp_flush_ip4_over_ethernet_internal (vnet_main_t * vnm, - vnet_arp_set_ip4_over_ethernet_rpc_args_t - * args) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_arp_ip4_entry_t *e; - ethernet_arp_interface_t *eai; - - if (vec_len (am->ethernet_arp_by_sw_if_index) <= args->sw_if_index) - return 0; - - eai = &am->ethernet_arp_by_sw_if_index[args->sw_if_index]; - - e = arp_entry_find (eai, &args->ip4); - - if (NULL != e) - { - adj_nbr_walk_nh4 (e->sw_if_index, - &e->ip4_address, arp_mk_incomplete_walk, e); - - /* - * The difference between flush and unset, is that an unset - * means delete for static and dynamic entries. A flush - * means delete only for dynamic. Flushing is what the DP - * does in response to interface events. unset is only done - * by the control plane. - */ - if (e->flags & IP_NEIGHBOR_FLAG_STATIC) - { - e->flags &= ~IP_NEIGHBOR_FLAG_DYNAMIC; - } - else if (e->flags & IP_NEIGHBOR_FLAG_DYNAMIC) - { - arp_entry_free (eai, e); - } - } - return (0); -} - -/* - * arp_add_del_interface_address - * - * callback when an interface address is added or deleted - */ -static void -arp_enable_disable_interface (ip4_main_t * im, - uword opaque, u32 sw_if_index, u32 is_enable) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - - if (is_enable) - arp_enable (am, sw_if_index); - else - arp_disable (am, sw_if_index); -} - -/* - * arp_add_del_interface_address - * - * callback when an interface address is added or deleted - */ -static void -arp_add_del_interface_address (ip4_main_t * im, - uword opaque, - u32 sw_if_index, - ip4_address_t * address, - u32 address_length, - u32 if_address_index, u32 is_del) -{ - /* - * Flush the ARP cache of all entries covered by the address - * that is being removed. - */ - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_arp_ip4_entry_t *e; - - if (vec_len (am->ethernet_arp_by_sw_if_index) <= sw_if_index) - return; - - if (is_del) - { - ethernet_arp_interface_t *eai; - u32 i, *to_delete = 0; - hash_pair_t *pair; - - eai = &am->ethernet_arp_by_sw_if_index[sw_if_index]; - - /* *INDENT-OFF* */ - hash_foreach_pair (pair, eai->arp_entries, - ({ - e = pool_elt_at_index(am->ip4_entry_pool, - pair->value[0]); - if (ip4_destination_matches_route (im, &e->ip4_address, - address, address_length)) - { - vec_add1 (to_delete, e - am->ip4_entry_pool); - } - })); - /* *INDENT-ON* */ - - for (i = 0; i < vec_len (to_delete); i++) - { - e = pool_elt_at_index (am->ip4_entry_pool, to_delete[i]); - - vnet_arp_set_ip4_over_ethernet_rpc_args_t delme = { - .ip4.as_u32 = e->ip4_address.as_u32, - .sw_if_index = e->sw_if_index, - .flags = ETHERNET_ARP_ARGS_FLUSH, - }; - mac_address_copy (&delme.mac, &e->mac); - - vnet_arp_flush_ip4_over_ethernet_internal (vnet_get_main (), - &delme); - } - - vec_free (to_delete); - } -} - -static void -arp_table_bind (ip4_main_t * im, - uword opaque, - u32 sw_if_index, u32 new_fib_index, u32 old_fib_index) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_arp_interface_t *eai; - ethernet_arp_ip4_entry_t *e; - hash_pair_t *pair; - - /* - * the IP table that the interface is bound to has changed. - * reinstall all the adj fibs. - */ - - if (vec_len (am->ethernet_arp_by_sw_if_index) <= sw_if_index) - return; - - eai = &am->ethernet_arp_by_sw_if_index[sw_if_index]; - - /* *INDENT-OFF* */ - hash_foreach_pair (pair, eai->arp_entries, - ({ - e = pool_elt_at_index(am->ip4_entry_pool, - pair->value[0]); - /* - * remove the adj-fib from the old table and add to the new - */ - arp_adj_fib_remove(e, old_fib_index); - arp_adj_fib_add(e, new_fib_index); - })); - /* *INDENT-ON* */ - -} - -static clib_error_t * -ethernet_arp_init (vlib_main_t * vm) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ip4_main_t *im = &ip4_main; - pg_node_t *pn; - - ethernet_register_input_type (vm, ETHERNET_TYPE_ARP, arp_input_node.index); - - pn = pg_get_node (arp_input_node.index); - pn->unformat_edit = unformat_pg_arp_header; - - am->opcode_by_name = hash_create_string (0, sizeof (uword)); -#define _(o) hash_set_mem (am->opcode_by_name, #o, ETHERNET_ARP_OPCODE_##o); - foreach_ethernet_arp_opcode; -#undef _ - - /* $$$ configurable */ - am->limit_arp_cache_size = 50000; - - am->pending_resolutions_by_address = hash_create (0, sizeof (uword)); - am->mac_changes_by_address = hash_create (0, sizeof (uword)); - am->wc_ip4_arp_publisher_node = (uword) ~ 0; - - /* don't trace ARP error packets */ - { - vlib_node_runtime_t *rt = - vlib_node_get_runtime (vm, arp_input_node.index); - -#define _(a,b) \ - vnet_pcap_drop_trace_filter_add_del \ - (rt->errors[ETHERNET_ARP_ERROR_##a], \ - 1 /* is_add */); - foreach_ethernet_arp_error -#undef _ - } - - ip4_add_del_interface_address_callback_t cb; - cb.function = arp_add_del_interface_address; - cb.function_opaque = 0; - vec_add1 (im->add_del_interface_address_callbacks, cb); - - ip4_enable_disable_interface_callback_t cbe; - cbe.function = arp_enable_disable_interface; - cbe.function_opaque = 0; - vec_add1 (im->enable_disable_interface_callbacks, cbe); - - ip4_table_bind_callback_t cbt; - cbt.function = arp_table_bind; - cbt.function_opaque = 0; - vec_add1 (im->table_bind_callbacks, cbt); - - return 0; -} -/* *INDENT-OFF* */ -VLIB_INIT_FUNCTION (ethernet_arp_init) = -{ - .runs_after = VLIB_INITS("ethernet_init"), -}; -/* *INDENT-ON* */ - -static void -arp_entry_free (ethernet_arp_interface_t * eai, ethernet_arp_ip4_entry_t * e) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - - arp_adj_fib_remove - (e, ip4_fib_table_get_index_for_sw_if_index (e->sw_if_index)); - hash_unset (eai->arp_entries, e->ip4_address.as_u32); - pool_put (am->ip4_entry_pool, e); -} - -static inline int -vnet_arp_unset_ip4_over_ethernet_internal (vnet_main_t * vnm, - vnet_arp_set_ip4_over_ethernet_rpc_args_t - * args) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_arp_ip4_entry_t *e; - ethernet_arp_interface_t *eai; - - if (vec_len (am->ethernet_arp_by_sw_if_index) <= args->sw_if_index) - return 0; - - eai = &am->ethernet_arp_by_sw_if_index[args->sw_if_index]; - - e = arp_entry_find (eai, &args->ip4); - - if (NULL != e) - { - adj_nbr_walk_nh4 (e->sw_if_index, - &e->ip4_address, arp_mk_incomplete_walk, e); - arp_entry_free (eai, e); - } - - return 0; -} - - -static int -vnet_arp_populate_ip4_over_ethernet_internal (vnet_main_t * vnm, - vnet_arp_set_ip4_over_ethernet_rpc_args_t - * args) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_arp_ip4_entry_t *e; - ethernet_arp_interface_t *eai; - - arp_enable (am, args->sw_if_index); - eai = &am->ethernet_arp_by_sw_if_index[args->sw_if_index]; - - e = arp_entry_find (eai, &args->ip4); - - if (NULL != e) - { - adj_nbr_walk_nh4 (e->sw_if_index, - &e->ip4_address, arp_mk_complete_walk, e); - } - return (0); -} - -static void -set_ip4_over_ethernet_rpc_callback (vnet_arp_set_ip4_over_ethernet_rpc_args_t - * a) -{ - vnet_main_t *vm = vnet_get_main (); - ASSERT (vlib_get_thread_index () == 0); - - if (a->flags & ETHERNET_ARP_ARGS_REMOVE) - vnet_arp_unset_ip4_over_ethernet_internal (vm, a); - else if (a->flags & ETHERNET_ARP_ARGS_FLUSH) - vnet_arp_flush_ip4_over_ethernet_internal (vm, a); - else if (a->flags & ETHERNET_ARP_ARGS_POPULATE) - vnet_arp_populate_ip4_over_ethernet_internal (vm, a); - else if (a->flags & ETHERNET_ARP_ARGS_WC_PUB) - vnet_arp_wc_publish_internal (vm, a); - else - vnet_arp_set_ip4_over_ethernet_internal (vm, a); -} - -/** - * @brief Invoked when the interface's admin state changes - */ -static clib_error_t * -ethernet_arp_sw_interface_up_down (vnet_main_t * vnm, - u32 sw_if_index, u32 flags) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_arp_ip4_entry_t *e; - u32 i, *to_update = 0; - - /* *INDENT-OFF* */ - pool_foreach (e, am->ip4_entry_pool, - ({ - if (e->sw_if_index == sw_if_index) - vec_add1 (to_update, - e - am->ip4_entry_pool); - })); - /* *INDENT-ON* */ - - for (i = 0; i < vec_len (to_update); i++) - { - e = pool_elt_at_index (am->ip4_entry_pool, to_update[i]); - - vnet_arp_set_ip4_over_ethernet_rpc_args_t update_me = { - .ip4.as_u32 = e->ip4_address.as_u32, - .sw_if_index = e->sw_if_index, - }; - mac_address_copy (&update_me.mac, &e->mac); - - if (flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP) - { - update_me.flags = ETHERNET_ARP_ARGS_POPULATE; - vnet_arp_populate_ip4_over_ethernet_internal (vnm, &update_me); - } - else - { - update_me.flags = ETHERNET_ARP_ARGS_FLUSH; - vnet_arp_flush_ip4_over_ethernet_internal (vnm, &update_me); - } - } - vec_free (to_update); - - return 0; -} - -VNET_SW_INTERFACE_ADMIN_UP_DOWN_FUNCTION (ethernet_arp_sw_interface_up_down); - -static void -increment_ip4_and_mac_address (ethernet_arp_ip4_over_ethernet_address_t * a) -{ - u8 old; - int i; - - for (i = 3; i >= 0; i--) - { - old = a->ip4.as_u8[i]; - a->ip4.as_u8[i] += 1; - if (old < a->ip4.as_u8[i]) - break; - } - - for (i = 5; i >= 0; i--) - { - old = a->mac.bytes[i]; - a->mac.bytes[i] += 1; - if (old < a->mac.bytes[i]) - break; - } -} - -int -vnet_arp_set_ip4_over_ethernet (vnet_main_t * vnm, - u32 sw_if_index, - const ethernet_arp_ip4_over_ethernet_address_t - * a, ip_neighbor_flags_t flags) -{ - vnet_arp_set_ip4_over_ethernet_rpc_args_t args = { - .sw_if_index = sw_if_index, - .nbr_flags = flags, - .flags = 0, - .ip4.as_u32 = a->ip4.as_u32, - .mac = a->mac, - }; - - vl_api_rpc_call_main_thread (set_ip4_over_ethernet_rpc_callback, - (u8 *) & args, sizeof (args)); - return 0; -} - -void -proxy_arp_walk (proxy_arp_walk_t cb, void *data) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_proxy_arp_t *pa; - - vec_foreach (pa, am->proxy_arps) - { - if (!cb (&pa->lo_addr, &pa->hi_addr, pa->fib_index, data)) - break; - } -} - -int -vnet_proxy_arp_enable_disable (vnet_main_t * vnm, u32 sw_if_index, u8 enable) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_arp_interface_t *eai; - - vec_validate (am->ethernet_arp_by_sw_if_index, sw_if_index); - - eai = &am->ethernet_arp_by_sw_if_index[sw_if_index]; - - if (enable) - { - if (!eai->proxy_enabled) - { - vnet_feature_enable_disable ("arp", "arp-proxy", - sw_if_index, 1, NULL, 0); - } - eai->proxy_enabled = 1; - } - else - { - if (eai->proxy_enabled) - { - vnet_feature_enable_disable ("arp", "arp-proxy", - sw_if_index, 0, NULL, 0); - } - eai->proxy_enabled = 0; - } - - return (0); -} - -int -vnet_proxy_arp_add_del (ip4_address_t * lo_addr, - ip4_address_t * hi_addr, u32 fib_index, int is_del) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_proxy_arp_t *pa; - u32 found_at_index = ~0; - - vec_foreach (pa, am->proxy_arps) - { - if (pa->lo_addr.as_u32 == lo_addr->as_u32 && - pa->hi_addr.as_u32 == hi_addr->as_u32 && pa->fib_index == fib_index) - { - found_at_index = pa - am->proxy_arps; - break; - } - } - - if (found_at_index != ~0) - { - /* Delete, otherwise it's already in the table */ - if (is_del) - vec_delete (am->proxy_arps, 1, found_at_index); - return 0; - } - /* delete, no such entry */ - if (is_del) - return VNET_API_ERROR_NO_SUCH_ENTRY; - - /* add, not in table */ - vec_add2 (am->proxy_arps, pa, 1); - pa->lo_addr.as_u32 = lo_addr->as_u32; - pa->hi_addr.as_u32 = hi_addr->as_u32; - pa->fib_index = fib_index; - return 0; -} - -void -proxy_arp_intfc_walk (proxy_arp_intf_walk_t cb, void *data) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_arp_interface_t *eai; - - vec_foreach (eai, am->ethernet_arp_by_sw_if_index) - { - if (eai->proxy_enabled) - cb (eai - am->ethernet_arp_by_sw_if_index, data); - } -} - -static clib_error_t * -ip_arp_add_del_command_fn (vlib_main_t * vm, - unformat_input_t * input, vlib_cli_command_t * cmd) -{ - vnet_main_t *vnm = vnet_get_main (); - u32 sw_if_index; - ethernet_arp_ip4_over_ethernet_address_t lo_addr, hi_addr, addr; - int addr_valid = 0; - int is_del = 0; - int count = 1; - u32 fib_index = 0; - u32 fib_id; - int is_proxy = 0; - ip_neighbor_flags_t flags; - - flags = IP_NEIGHBOR_FLAG_NONE; - - while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT) - { - /* set ip arp TenGigE1/1/0/1 1.2.3.4 aa:bb:... or aabb.ccdd... */ - if (unformat (input, "%U %U %U", - unformat_vnet_sw_interface, vnm, &sw_if_index, - unformat_ip4_address, &addr.ip4, - unformat_mac_address_t, &addr.mac)) - addr_valid = 1; - - else if (unformat (input, "delete") || unformat (input, "del")) - is_del = 1; - - else if (unformat (input, "static")) - flags |= IP_NEIGHBOR_FLAG_STATIC; - - else if (unformat (input, "no-fib-entry")) - flags |= IP_NEIGHBOR_FLAG_NO_FIB_ENTRY; - - else if (unformat (input, "count %d", &count)) - ; - - else if (unformat (input, "fib-id %d", &fib_id)) - { - fib_index = fib_table_find (FIB_PROTOCOL_IP4, fib_id); - - if (~0 == fib_index) - return clib_error_return (0, "fib ID %d doesn't exist\n", fib_id); - } - - else if (unformat (input, "proxy %U - %U", - unformat_ip4_address, &lo_addr.ip4, - unformat_ip4_address, &hi_addr.ip4)) - is_proxy = 1; - else - break; - } - - if (is_proxy) - { - (void) vnet_proxy_arp_add_del (&lo_addr.ip4, &hi_addr.ip4, - fib_index, is_del); - return 0; - } - - if (addr_valid) - { - int i; - - for (i = 0; i < count; i++) - { - if (is_del == 0) - { - uword event_type, *event_data = 0; - - /* Park the debug CLI until the arp entry is installed */ - vnet_register_ip4_arp_resolution_event - (vnm, &addr.ip4, vlib_current_process (vm), - 1 /* type */ , 0 /* data */ ); - - vnet_arp_set_ip4_over_ethernet (vnm, sw_if_index, &addr, flags); - - vlib_process_wait_for_event (vm); - event_type = vlib_process_get_events (vm, &event_data); - vec_reset_length (event_data); - if (event_type != 1) - clib_warning ("event type %d unexpected", event_type); - } - else - vnet_arp_unset_ip4_over_ethernet (vnm, sw_if_index, &addr); - - increment_ip4_and_mac_address (&addr); - } - } - else - { - return clib_error_return (0, "unknown input `%U'", - format_unformat_error, input); - } - - return 0; -} - -/* *INDENT-OFF* */ -/*? - * Add or delete IPv4 ARP cache entries. - * - * @note 'set ip arp' options (e.g. delete, static, 'fib-id <id>', - * 'count <number>', 'interface ip4_addr mac_addr') can be added in - * any order and combination. - * - * @cliexpar - * @parblock - * Add or delete IPv4 ARP cache entries as follows. MAC Address can be in - * either aa:bb:cc:dd:ee:ff format or aabb.ccdd.eeff format. - * @cliexcmd{set ip arp GigabitEthernet2/0/0 6.0.0.3 dead.beef.babe} - * @cliexcmd{set ip arp delete GigabitEthernet2/0/0 6.0.0.3 de:ad:be:ef:ba:be} - * - * To add or delete an IPv4 ARP cache entry to or from a specific fib - * table: - * @cliexcmd{set ip arp fib-id 1 GigabitEthernet2/0/0 6.0.0.3 dead.beef.babe} - * @cliexcmd{set ip arp fib-id 1 delete GigabitEthernet2/0/0 6.0.0.3 dead.beef.babe} - * - * Add or delete IPv4 static ARP cache entries as follows: - * @cliexcmd{set ip arp static GigabitEthernet2/0/0 6.0.0.3 dead.beef.babe} - * @cliexcmd{set ip arp static delete GigabitEthernet2/0/0 6.0.0.3 dead.beef.babe} - * - * For testing / debugging purposes, the 'set ip arp' command can add or - * delete multiple entries. Supply the 'count N' parameter: - * @cliexcmd{set ip arp count 10 GigabitEthernet2/0/0 6.0.0.3 dead.beef.babe} - * @endparblock - ?*/ -VLIB_CLI_COMMAND (ip_arp_add_del_command, static) = { - .path = "set ip arp", - .short_help = - "set ip arp [del] <intfc> <ip-address> <mac-address> [static] [no-fib-entry] [count <count>] [fib-id <fib-id>] [proxy <lo-addr> - <hi-addr>]", - .function = ip_arp_add_del_command_fn, -}; -/* *INDENT-ON* */ - -static clib_error_t * -set_int_proxy_arp_command_fn (vlib_main_t * vm, - unformat_input_t * - input, vlib_cli_command_t * cmd) -{ - vnet_main_t *vnm = vnet_get_main (); - u32 sw_if_index; - int enable = 0; - - sw_if_index = ~0; - - while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT) - { - if (unformat (input, "%U", unformat_vnet_sw_interface, - vnm, &sw_if_index)) - ; - else if (unformat (input, "enable") || unformat (input, "on")) - enable = 1; - else if (unformat (input, "disable") || unformat (input, "off")) - enable = 0; - else - break; - } - - if (~0 == sw_if_index) - return clib_error_return (0, "unknown input '%U'", - format_unformat_error, input); - - vnet_proxy_arp_enable_disable (vnm, sw_if_index, enable); - - return 0; -} - -/* *INDENT-OFF* */ -/*? - * Enable proxy-arp on an interface. The vpp stack will answer ARP - * requests for the indicated address range. Multiple proxy-arp - * ranges may be provisioned. - * - * @note Proxy ARP as a technology is infamous for blackholing traffic. - * Also, the underlying implementation has not been performance-tuned. - * Avoid creating an unnecessarily large set of ranges. - * - * @cliexpar - * To enable proxy arp on a range of addresses, use: - * @cliexcmd{set ip arp proxy 6.0.0.1 - 6.0.0.11} - * Append 'del' to delete a range of proxy ARP addresses: - * @cliexcmd{set ip arp proxy 6.0.0.1 - 6.0.0.11 del} - * You must then specifically enable proxy arp on individual interfaces: - * @cliexcmd{set interface proxy-arp GigabitEthernet0/8/0 enable} - * To disable proxy arp on an individual interface: - * @cliexcmd{set interface proxy-arp GigabitEthernet0/8/0 disable} - ?*/ -VLIB_CLI_COMMAND (set_int_proxy_enable_command, static) = { - .path = "set interface proxy-arp", - .short_help = - "set interface proxy-arp <intfc> [enable|disable]", - .function = set_int_proxy_arp_command_fn, -}; -/* *INDENT-ON* */ - - -/* - * ARP/ND Termination in a L2 Bridge Domain based on IP4/IP6 to MAC - * hash tables mac_by_ip4 and mac_by_ip6 for each BD. - */ -typedef enum -{ - ARP_TERM_NEXT_L2_OUTPUT, - ARP_TERM_NEXT_DROP, - ARP_TERM_N_NEXT, -} arp_term_next_t; - -u32 arp_term_next_node_index[32]; - -static uword -arp_term_l2bd (vlib_main_t * vm, - vlib_node_runtime_t * node, vlib_frame_t * frame) -{ - l2input_main_t *l2im = &l2input_main; - u32 n_left_from, next_index, *from, *to_next; - u32 n_replies_sent = 0; - u16 last_bd_index = ~0; - l2_bridge_domain_t *last_bd_config = 0; - l2_input_config_t *cfg0; - - from = vlib_frame_vector_args (frame); - n_left_from = frame->n_vectors; - next_index = node->cached_next_index; - - while (n_left_from > 0) - { - u32 n_left_to_next; - - vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); - - while (n_left_from > 0 && n_left_to_next > 0) - { - vlib_buffer_t *p0; - ethernet_header_t *eth0; - ethernet_arp_header_t *arp0; - ip6_header_t *iph0; - u8 *l3h0; - u32 pi0, error0, next0, sw_if_index0; - u16 ethertype0; - u16 bd_index0; - u32 ip0; - u8 *macp0; - - pi0 = from[0]; - to_next[0] = pi0; - from += 1; - to_next += 1; - n_left_from -= 1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer (vm, pi0); - // Terminate only local (SHG == 0) ARP - if (vnet_buffer (p0)->l2.shg != 0) - goto next_l2_feature; - - eth0 = vlib_buffer_get_current (p0); - l3h0 = (u8 *) eth0 + vnet_buffer (p0)->l2.l2_len; - ethertype0 = clib_net_to_host_u16 (*(u16 *) (l3h0 - 2)); - arp0 = (ethernet_arp_header_t *) l3h0; - - if (ethertype0 != ETHERNET_TYPE_ARP) - goto check_ip6_nd; - - if ((arp0->opcode != - clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_request)) && - (arp0->opcode != - clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_reply))) - goto check_ip6_nd; - - /* Must be ARP request/reply packet here */ - if (PREDICT_FALSE ((node->flags & VLIB_NODE_FLAG_TRACE) && - (p0->flags & VLIB_BUFFER_IS_TRACED))) - { - u8 *t0 = vlib_add_trace (vm, node, p0, - sizeof (ethernet_arp_input_trace_t)); - clib_memcpy_fast (t0, l3h0, - sizeof (ethernet_arp_input_trace_t)); - } - - error0 = 0; - error0 = - (arp0->l2_type != - clib_net_to_host_u16 (ETHERNET_ARP_HARDWARE_TYPE_ethernet) - ? ETHERNET_ARP_ERROR_l2_type_not_ethernet : error0); - error0 = - (arp0->l3_type != - clib_net_to_host_u16 (ETHERNET_TYPE_IP4) ? - ETHERNET_ARP_ERROR_l3_type_not_ip4 : error0); - - sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX]; - - if (error0) - goto drop; - - /* Trash ARP packets whose ARP-level source addresses do not - match, or if requester address is mcast */ - if (PREDICT_FALSE - (!ethernet_mac_address_equal (eth0->src_address, - arp0->ip4_over_ethernet[0]. - mac.bytes)) - || ethernet_address_cast (arp0->ip4_over_ethernet[0].mac.bytes)) - { - /* VRRP virtual MAC may be different to SMAC in ARP reply */ - if (!ethernet_mac_address_equal - (arp0->ip4_over_ethernet[0].mac.bytes, vrrp_prefix)) - { - error0 = ETHERNET_ARP_ERROR_l2_address_mismatch; - goto drop; - } - } - if (PREDICT_FALSE - (ip4_address_is_multicast (&arp0->ip4_over_ethernet[0].ip4))) - { - error0 = ETHERNET_ARP_ERROR_l3_src_address_not_local; - goto drop; - } - - /* Check if anyone want ARP request events for L2 BDs */ - { - ethernet_arp_main_t *am = ðernet_arp_main; - if (am->wc_ip4_arp_publisher_node != (uword) ~ 0) - vnet_arp_wc_publish (sw_if_index0, &arp0->ip4_over_ethernet[0]); - } - - /* lookup BD mac_by_ip4 hash table for MAC entry */ - ip0 = arp0->ip4_over_ethernet[1].ip4.as_u32; - bd_index0 = vnet_buffer (p0)->l2.bd_index; - if (PREDICT_FALSE ((bd_index0 != last_bd_index) - || (last_bd_index == (u16) ~ 0))) - { - last_bd_index = bd_index0; - last_bd_config = vec_elt_at_index (l2im->bd_configs, bd_index0); - } - macp0 = (u8 *) hash_get (last_bd_config->mac_by_ip4, ip0); - - if (PREDICT_FALSE (!macp0)) - goto next_l2_feature; /* MAC not found */ - if (PREDICT_FALSE (arp0->ip4_over_ethernet[0].ip4.as_u32 == - arp0->ip4_over_ethernet[1].ip4.as_u32)) - goto next_l2_feature; /* GARP */ - - /* MAC found, send ARP reply - - Convert ARP request packet to ARP reply */ - arp0->opcode = clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_reply); - arp0->ip4_over_ethernet[1] = arp0->ip4_over_ethernet[0]; - arp0->ip4_over_ethernet[0].ip4.as_u32 = ip0; - mac_address_from_bytes (&arp0->ip4_over_ethernet[0].mac, macp0); - clib_memcpy_fast (eth0->dst_address, eth0->src_address, 6); - clib_memcpy_fast (eth0->src_address, macp0, 6); - n_replies_sent += 1; - - output_response: - /* For BVI, need to use l2-fwd node to send ARP reply as - l2-output node cannot output packet to BVI properly */ - cfg0 = vec_elt_at_index (l2im->configs, sw_if_index0); - if (PREDICT_FALSE (cfg0->bvi)) - { - vnet_buffer (p0)->l2.feature_bitmap |= L2INPUT_FEAT_FWD; - vnet_buffer (p0)->sw_if_index[VLIB_RX] = 0; - goto next_l2_feature; - } - - /* Send ARP/ND reply back out input interface through l2-output */ - vnet_buffer (p0)->sw_if_index[VLIB_TX] = sw_if_index0; - next0 = ARP_TERM_NEXT_L2_OUTPUT; - vlib_validate_buffer_enqueue_x1 (vm, node, next_index, - to_next, n_left_to_next, pi0, - next0); - continue; - - check_ip6_nd: - /* IP6 ND event notification or solicitation handling to generate - local response instead of flooding */ - iph0 = (ip6_header_t *) l3h0; - if (PREDICT_FALSE (ethertype0 == ETHERNET_TYPE_IP6 && - iph0->protocol == IP_PROTOCOL_ICMP6 && - !ip6_address_is_unspecified - (&iph0->src_address))) - { - sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX]; - if (vnet_ip6_nd_term - (vm, node, p0, eth0, iph0, sw_if_index0, - vnet_buffer (p0)->l2.bd_index)) - goto output_response; - } - - next_l2_feature: - { - next0 = vnet_l2_feature_next (p0, arp_term_next_node_index, - L2INPUT_FEAT_ARP_TERM); - vlib_validate_buffer_enqueue_x1 (vm, node, next_index, - to_next, n_left_to_next, - pi0, next0); - continue; - } - - drop: - if (0 == arp0->ip4_over_ethernet[0].ip4.as_u32 || - (arp0->ip4_over_ethernet[0].ip4.as_u32 == - arp0->ip4_over_ethernet[1].ip4.as_u32)) - { - error0 = ETHERNET_ARP_ERROR_gratuitous_arp; - } - next0 = ARP_TERM_NEXT_DROP; - p0->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); - } - - vlib_error_count (vm, node->node_index, - ETHERNET_ARP_ERROR_replies_sent, n_replies_sent); - return frame->n_vectors; -} - -/* *INDENT-OFF* */ -VLIB_REGISTER_NODE (arp_term_l2bd_node, static) = { - .function = arp_term_l2bd, - .name = "arp-term-l2bd", - .vector_size = sizeof (u32), - .n_errors = ETHERNET_ARP_N_ERROR, - .error_strings = ethernet_arp_error_strings, - .n_next_nodes = ARP_TERM_N_NEXT, - .next_nodes = { - [ARP_TERM_NEXT_L2_OUTPUT] = "l2-output", - [ARP_TERM_NEXT_DROP] = "error-drop", - }, - .format_buffer = format_ethernet_arp_header, - .format_trace = format_arp_term_input_trace, -}; -/* *INDENT-ON* */ - -clib_error_t * -arp_term_init (vlib_main_t * vm) -{ - // Initialize the feature next-node indexes - feat_bitmap_init_next_nodes (vm, - arp_term_l2bd_node.index, - L2INPUT_N_FEAT, - l2input_get_feat_names (), - arp_term_next_node_index); - return 0; -} - -VLIB_INIT_FUNCTION (arp_term_init); - -void -change_arp_mac (u32 sw_if_index, ethernet_arp_ip4_entry_t * e) -{ - if (e->sw_if_index == sw_if_index) - { - adj_nbr_walk_nh4 (e->sw_if_index, - &e->ip4_address, arp_mk_complete_walk, e); - } -} - -void -ethernet_arp_change_mac (u32 sw_if_index) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - ethernet_arp_ip4_entry_t *e; - adj_index_t ai; - - /* *INDENT-OFF* */ - pool_foreach (e, am->ip4_entry_pool, - ({ - change_arp_mac (sw_if_index, e); - })); - /* *INDENT-ON* */ - - ai = adj_glean_get (FIB_PROTOCOL_IP4, sw_if_index); - - if (ADJ_INDEX_INVALID != ai) - adj_glean_update_rewrite (ai); -} - -void -send_ip4_garp (vlib_main_t * vm, u32 sw_if_index) -{ - ip4_main_t *i4m = &ip4_main; - ip4_address_t *ip4_addr = ip4_interface_first_address (i4m, sw_if_index, 0); - - send_ip4_garp_w_addr (vm, ip4_addr, sw_if_index); -} - -void -send_ip4_garp_w_addr (vlib_main_t * vm, - const ip4_address_t * ip4_addr, u32 sw_if_index) -{ - ip4_main_t *i4m = &ip4_main; - vnet_main_t *vnm = vnet_get_main (); - u8 *rewrite, rewrite_len; - vnet_hw_interface_t *hi = vnet_get_sup_hw_interface (vnm, sw_if_index); - - if (ip4_addr) - { - clib_warning ("Sending GARP for IP4 address %U on sw_if_idex %d", - format_ip4_address, ip4_addr, sw_if_index); - - /* Form GARP packet for output - Gratuitous ARP is an ARP request packet - where the interface IP/MAC pair is used for both source and request - MAC/IP pairs in the request */ - u32 bi = 0; - ethernet_arp_header_t *h = vlib_packet_template_get_packet - (vm, &i4m->ip4_arp_request_packet_template, &bi); - - if (!h) - return; - - mac_address_from_bytes (&h->ip4_over_ethernet[0].mac, hi->hw_address); - mac_address_from_bytes (&h->ip4_over_ethernet[1].mac, hi->hw_address); - h->ip4_over_ethernet[0].ip4 = ip4_addr[0]; - h->ip4_over_ethernet[1].ip4 = ip4_addr[0]; - - /* Setup MAC header with ARP Etype and broadcast DMAC */ - vlib_buffer_t *b = vlib_get_buffer (vm, bi); - rewrite = - ethernet_build_rewrite (vnm, sw_if_index, VNET_LINK_ARP, - VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST); - rewrite_len = vec_len (rewrite); - vlib_buffer_advance (b, -rewrite_len); - ethernet_header_t *e = vlib_buffer_get_current (b); - clib_memcpy_fast (e->dst_address, rewrite, rewrite_len); - vec_free (rewrite); - - /* Send GARP packet out the specified interface */ - vnet_buffer (b)->sw_if_index[VLIB_RX] = - vnet_buffer (b)->sw_if_index[VLIB_TX] = sw_if_index; - 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); - } -} - -/* - * Remove any arp entries associated with the specified interface - */ -static clib_error_t * -vnet_arp_delete_sw_interface (vnet_main_t * vnm, u32 sw_if_index, u32 is_add) -{ - ethernet_arp_main_t *am = ðernet_arp_main; - - if (!is_add && sw_if_index != ~0) - { - ethernet_arp_ip4_entry_t *e; - /* *INDENT-OFF* */ - pool_foreach (e, am->ip4_entry_pool, ({ - if (e->sw_if_index != sw_if_index) - continue; - vnet_arp_set_ip4_over_ethernet_rpc_args_t args = { - .sw_if_index = sw_if_index, - .ip4 = e->ip4_address, - }; - vnet_arp_unset_ip4_over_ethernet_internal (vnm, &args); - })); - /* *INDENT-ON* */ - arp_disable (am, sw_if_index); - } - else if (is_add) - { - vnet_feature_enable_disable ("arp", "arp-disabled", - sw_if_index, 1, NULL, 0); - } - - return (NULL); -} - -VNET_SW_INTERFACE_ADD_DEL_FUNCTION (vnet_arp_delete_sw_interface); - -/* - * fd.io coding-style-patch-verification: ON - * - * Local Variables: - * eval: (c-set-style "gnu") - * End: - */ diff --git a/src/vnet/ethernet/arp.h b/src/vnet/ethernet/arp.h deleted file mode 100644 index de92c4f419c..00000000000 --- a/src/vnet/ethernet/arp.h +++ /dev/null @@ -1,123 +0,0 @@ -/* - * 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. - */ - -#ifndef __ARP_H__ -#define __ARP_H__ - -#include <vnet/ethernet/ethernet.h> -#include <vnet/ethernet/arp_packet.h> -#include <vnet/ip/ip.h> -#include <vnet/ip/ip_neighbor.h> - -typedef struct -{ - u32 sw_if_index; - ip4_address_t ip4_address; - - mac_address_t mac; - - ip_neighbor_flags_t flags; - - f64 time_last_updated; - - /** - * The index of the adj-fib entry created - */ - fib_node_index_t fib_entry_index; -} ethernet_arp_ip4_entry_t; - -extern u8 *format_ethernet_arp_ip4_entry (u8 * s, va_list * va); - -ethernet_arp_ip4_entry_t *ip4_neighbors_pool (void); -ethernet_arp_ip4_entry_t *ip4_neighbor_entries (u32 sw_if_index); - -extern int vnet_proxy_arp_add_del (ip4_address_t * lo_addr, - ip4_address_t * hi_addr, - u32 fib_index, int is_del); - -extern int vnet_proxy_arp_enable_disable (vnet_main_t * vnm, - u32 sw_if_index, u8 enable); - -extern int vnet_arp_set_ip4_over_ethernet (vnet_main_t * vnm, - u32 sw_if_index, - const - ethernet_arp_ip4_over_ethernet_address_t - * a, ip_neighbor_flags_t flags); - -extern int vnet_arp_unset_ip4_over_ethernet (vnet_main_t * vnm, - u32 sw_if_index, - const - ethernet_arp_ip4_over_ethernet_address_t - * a); - -void vnet_register_ip4_arp_resolution_event (vnet_main_t * vnm, - void *address_arg, - uword node_index, - uword type_opaque, uword data); - -typedef int (*arp_change_event_cb_t) (u32 pool_index, - const mac_address_t * mac, - u32 sw_if_index, - const ip4_address_t * address); - -int vnet_add_del_ip4_arp_change_event (vnet_main_t * vnm, - arp_change_event_cb_t data_callback, - u32 pid, - void *address_arg, - uword node_index, - uword type_opaque, - uword data, int is_add); - -void wc_arp_set_publisher_node (uword inode_index, uword event_type); - -void ethernet_arp_change_mac (u32 sw_if_index); -void ethernet_ndp_change_mac (u32 sw_if_index); - -void arp_update_adjacency (vnet_main_t * vnm, u32 sw_if_index, u32 ai); - -typedef struct -{ - u32 sw_if_index; - ip4_address_t ip; - mac_address_t mac; -} wc_arp_report_t; - -/** - * call back function when walking the DB of proxy ARPs - * @return 0 to stop the walk !0 to continue - */ -typedef walk_rc_t (proxy_arp_walk_t) (const ip4_address_t * lo_addr, - const ip4_address_t * hi_addr, - u32 fib_index, void *dat); - -extern void proxy_arp_walk (proxy_arp_walk_t cb, void *data); - -/** - * call back function when walking the DB of proxy ARP interface - * @return 0 to stop the walk !0 to continue - */ -typedef walk_rc_t (proxy_arp_intf_walk_t) (u32 sw_if_index, void *data); - -extern void proxy_arp_intfc_walk (proxy_arp_intf_walk_t cb, void *data); - -#endif - -/* - * fd.io coding-style-patch-verification: ON - * - * Local Variables: - * eval: (c-set-style "gnu") - * End: - */ diff --git a/src/vnet/ethernet/arp_packet.c b/src/vnet/ethernet/arp_packet.c new file mode 100644 index 00000000000..c72dab99ad5 --- /dev/null +++ b/src/vnet/ethernet/arp_packet.c @@ -0,0 +1,111 @@ +/* + * ethernet/arp.c: IP v4 ARP node + * + * Copyright (c) 2010 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/ethernet/arp_packet.h> +#include <vnet/ethernet/ethernet.h> +#include <vnet/ip/format.h> + +u8 * +format_ethernet_arp_opcode (u8 * s, va_list * va) +{ + ethernet_arp_opcode_t o = va_arg (*va, ethernet_arp_opcode_t); + char *t = 0; + switch (o) + { +#define _(f) case ETHERNET_ARP_OPCODE_##f: t = #f; break; + foreach_ethernet_arp_opcode; +#undef _ + + default: + return format (s, "unknown 0x%x", o); + } + + return format (s, "%s", t); +} + +u8 * +format_ethernet_arp_hardware_type (u8 * s, va_list * va) +{ + ethernet_arp_hardware_type_t h = va_arg (*va, ethernet_arp_hardware_type_t); + char *t = 0; + switch (h) + { +#define _(n,f) case n: t = #f; break; + foreach_ethernet_arp_hardware_type; +#undef _ + + default: + return format (s, "unknown 0x%x", h); + } + + return format (s, "%s", t); +} + +u8 * +format_ethernet_arp_header (u8 * s, va_list * va) +{ + ethernet_arp_header_t *a = va_arg (*va, ethernet_arp_header_t *); + u32 max_header_bytes = va_arg (*va, u32); + u32 indent; + u16 l2_type, l3_type; + + if (max_header_bytes != 0 && sizeof (a[0]) > max_header_bytes) + return format (s, "ARP header truncated"); + + l2_type = clib_net_to_host_u16 (a->l2_type); + l3_type = clib_net_to_host_u16 (a->l3_type); + + indent = format_get_indent (s); + + s = format (s, "%U, type %U/%U, address size %d/%d", + format_ethernet_arp_opcode, clib_net_to_host_u16 (a->opcode), + format_ethernet_arp_hardware_type, l2_type, + format_ethernet_type, l3_type, + a->n_l2_address_bytes, a->n_l3_address_bytes); + + if (l2_type == ETHERNET_ARP_HARDWARE_TYPE_ethernet + && l3_type == ETHERNET_TYPE_IP4) + { + s = format (s, "\n%U%U/%U -> %U/%U", + format_white_space, indent, + format_mac_address_t, &a->ip4_over_ethernet[0].mac, + format_ip4_address, &a->ip4_over_ethernet[0].ip4, + format_mac_address_t, &a->ip4_over_ethernet[1].mac, + format_ip4_address, &a->ip4_over_ethernet[1].ip4); + } + else + { + uword n2 = a->n_l2_address_bytes; + uword n3 = a->n_l3_address_bytes; + s = format (s, "\n%U%U/%U -> %U/%U", + format_white_space, indent, + format_hex_bytes, a->data + 0 * n2 + 0 * n3, n2, + format_hex_bytes, a->data + 1 * n2 + 0 * n3, n3, + format_hex_bytes, a->data + 1 * n2 + 1 * n3, n2, + format_hex_bytes, a->data + 2 * n2 + 1 * n3, n3); + } + + return s; +} + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ethernet/arp_packet.h b/src/vnet/ethernet/arp_packet.h index b4e021cce19..a35adf17c85 100644 --- a/src/vnet/ethernet/arp_packet.h +++ b/src/vnet/ethernet/arp_packet.h @@ -18,6 +18,7 @@ #ifndef included_ethernet_arp_packet_h #define included_ethernet_arp_packet_h +#include <vnet/ip/ip4_packet.h> #include <vnet/ethernet/mac_address.h> #define foreach_ethernet_arp_hardware_type \ @@ -145,9 +146,9 @@ typedef struct }; } ethernet_arp_header_t; -void send_ip4_garp (vlib_main_t * vm, u32 sw_if_index); -void send_ip4_garp_w_addr (vlib_main_t * vm, - const ip4_address_t * ip4_addr, u32 sw_if_index); +extern u8 *format_ethernet_arp_header (u8 * s, va_list * va); +extern u8 *format_ethernet_arp_opcode (u8 * s, va_list * va); +extern u8 *format_ethernet_arp_hardware_type (u8 * s, va_list * va); #endif /* included_ethernet_arp_packet_h */ diff --git a/src/vnet/ethernet/ethernet.h b/src/vnet/ethernet/ethernet.h index cde6fb90038..2e5209c4423 100644 --- a/src/vnet/ethernet/ethernet.h +++ b/src/vnet/ethernet/ethernet.h @@ -257,8 +257,19 @@ typedef struct u32 input_next_mpls; } next_by_ethertype_t; +struct ethernet_main_t_; + +typedef void (ethernet_address_change_function_t) + (struct ethernet_main_t_ * im, u32 sw_if_index, uword opaque); + typedef struct { + ethernet_address_change_function_t *function; + uword function_opaque; +} ethernet_address_change_ctx_t; + +typedef struct ethernet_main_t_ +{ vlib_main_t *vlib_main; /* next node index for the L3 input node of each ethertype */ @@ -300,6 +311,10 @@ typedef struct /* Allocated loopback instances */ uword *bm_loopback_instances; + + /** Functions to call when interface hw address changes. */ + ethernet_address_change_ctx_t *address_change_callbacks; + } ethernet_main_t; extern ethernet_main_t ethernet_main; @@ -552,8 +567,6 @@ void ethernet_update_adjacency (vnet_main_t * vnm, u32 sw_if_index, u32 ai); u8 *ethernet_build_rewrite (vnet_main_t * vnm, u32 sw_if_index, vnet_link_t link_type, const void *dst_address); -const u8 *ethernet_ip4_mcast_dst_addr (void); -const u8 *ethernet_ip6_mcast_dst_addr (void); void ethernet_input_init (vlib_main_t * vm, ethernet_main_t * em); extern vlib_node_registration_t ethernet_input_node; diff --git a/src/vnet/ethernet/interface.c b/src/vnet/ethernet/interface.c index d79669206a9..39e5cfb3cf8 100644 --- a/src/vnet/ethernet/interface.c +++ b/src/vnet/ethernet/interface.c @@ -41,10 +41,12 @@ #include <vnet/ip/ip.h> #include <vnet/pg/pg.h> #include <vnet/ethernet/ethernet.h> -#include <vnet/ethernet/arp.h> +//#include <vnet/ethernet/arp.h> #include <vnet/l2/l2_input.h> #include <vnet/l2/l2_bd.h> #include <vnet/adj/adj.h> +#include <vnet/adj/adj_mcast.h> +#include <vnet/ip-neighbor/ip_neighbor.h> /** * @file @@ -53,7 +55,7 @@ * This file contains code to manage loopback interfaces. */ -const u8 * +static const u8 * ethernet_ip4_mcast_dst_addr (void) { const static u8 ethernet_mcast_dst_mac[] = { @@ -63,7 +65,7 @@ ethernet_ip4_mcast_dst_addr (void) return (ethernet_mcast_dst_mac); } -const u8 * +static const u8 * ethernet_ip6_mcast_dst_addr (void) { const static u8 ethernet_mcast_dst_mac[] = { @@ -195,27 +197,74 @@ ethernet_build_rewrite (vnet_main_t * vnm, void ethernet_update_adjacency (vnet_main_t * vnm, u32 sw_if_index, u32 ai) { - ip_adjacency_t *adj; - - adj = adj_get (ai); - vnet_sw_interface_t *si = vnet_get_sw_interface (vnm, sw_if_index); + if ((si->type == VNET_SW_INTERFACE_TYPE_P2P) || (si->type == VNET_SW_INTERFACE_TYPE_PIPE)) { default_update_adjacency (vnm, sw_if_index, ai); } - else if (FIB_PROTOCOL_IP4 == adj->ia_nh_proto) - { - arp_update_adjacency (vnm, sw_if_index, ai); - } - else if (FIB_PROTOCOL_IP6 == adj->ia_nh_proto) - { - ip6_ethernet_update_adjacency (vnm, sw_if_index, ai); - } else { - ASSERT (0); + ip_adjacency_t *adj; + + adj = adj_get (ai); + + switch (adj->lookup_next_index) + { + case IP_LOOKUP_NEXT_GLEAN: + adj_glean_update_rewrite (ai); + break; + case IP_LOOKUP_NEXT_ARP: + ip_neighbor_update (vnm, ai); + break; + case IP_LOOKUP_NEXT_BCAST: + adj_nbr_update_rewrite (ai, + ADJ_NBR_REWRITE_FLAG_COMPLETE, + ethernet_build_rewrite + (vnm, + adj->rewrite_header.sw_if_index, + adj->ia_link, + VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST)); + break; + case IP_LOOKUP_NEXT_MCAST: + { + /* + * Construct a partial rewrite from the known ethernet mcast dest MAC + */ + u8 *rewrite; + u8 offset; + + rewrite = ethernet_build_rewrite + (vnm, + sw_if_index, + adj->ia_link, + (adj->ia_nh_proto == FIB_PROTOCOL_IP6 ? + ethernet_ip6_mcast_dst_addr () : + ethernet_ip4_mcast_dst_addr ())); + + /* + * Complete the remaining fields of the adj's rewrite to direct the + * complete of the rewrite at switch time by copying in the IP + * dst address's bytes. + * Ofset is 2 bytes into the destintation address. + */ + offset = vec_len (rewrite) - 2; + adj_mcast_update_rewrite (ai, rewrite, offset); + + break; + } + case IP_LOOKUP_NEXT_DROP: + case IP_LOOKUP_NEXT_PUNT: + case IP_LOOKUP_NEXT_LOCAL: + case IP_LOOKUP_NEXT_REWRITE: + case IP_LOOKUP_NEXT_MCAST_MIDCHAIN: + case IP_LOOKUP_NEXT_MIDCHAIN: + case IP_LOOKUP_NEXT_ICMP_ERROR: + case IP_LOOKUP_N_NEXT: + ASSERT (0); + break; + } } } @@ -234,8 +283,12 @@ ethernet_mac_change (vnet_hw_interface_t * hi, clib_memcpy (hi->hw_address, mac_address, vec_len (hi->hw_address)); clib_memcpy (ei->address, (u8 *) mac_address, sizeof (ei->address)); - ethernet_arp_change_mac (hi->sw_if_index); - ethernet_ndp_change_mac (hi->sw_if_index); + + { + ethernet_address_change_ctx_t *cb; + vec_foreach (cb, em->address_change_callbacks) + cb->function (em, hi->sw_if_index, cb->function_opaque); + } return (NULL); } diff --git a/src/vnet/ethernet/mac_address.c b/src/vnet/ethernet/mac_address.c index 6f40e50efa2..b7981299700 100644 --- a/src/vnet/ethernet/mac_address.c +++ b/src/vnet/ethernet/mac_address.c @@ -61,6 +61,16 @@ unformat_mac_address_t (unformat_input_t * input, va_list * args) return (0); } +void +mac_address_increment (mac_address_t * mac) +{ + u64 a; + + a = mac_address_as_u64 (mac); + a++; + mac_address_from_u64 (mac, a); +} + /* * fd.io coding-style-patch-verification: ON * diff --git a/src/vnet/ethernet/mac_address.h b/src/vnet/ethernet/mac_address.h index 01fb76e002b..aa773c5c449 100644 --- a/src/vnet/ethernet/mac_address.h +++ b/src/vnet/ethernet/mac_address.h @@ -149,6 +149,7 @@ mac_address_set_zero (mac_address_t * mac) mac->u.last_2 = 0; } +extern void mac_address_increment (mac_address_t * mac); extern uword unformat_mac_address_t (unformat_input_t * input, va_list * args); extern u8 *format_mac_address_t (u8 * s, va_list * args); diff --git a/src/vnet/fib/fib_entry.c b/src/vnet/fib/fib_entry.c index 0e5482840bf..d8c57fd1032 100644 --- a/src/vnet/fib/fib_entry.c +++ b/src/vnet/fib/fib_entry.c @@ -1713,6 +1713,29 @@ fib_entry_pool_size (void) return (pool_elts(fib_entry_pool)); } +#ifdef CLIB_DEBUG +void +fib_table_assert_empty (const fib_table_t *fib_table) +{ + fib_node_index_t *fei, *feis = NULL; + fib_entry_t *fib_entry; + + pool_foreach (fib_entry, fib_entry_pool, + ({ + if (fib_entry->fe_fib_index == fib_table->ft_index) + vec_add1 (feis, fib_entry_get_index(fib_entry)); + })); + + if (vec_len(feis)) + { + vec_foreach (fei, feis) + clib_error ("%U", format_fib_entry, *fei, FIB_ENTRY_FORMAT_DETAIL); + } + + ASSERT(0); +} +#endif + static clib_error_t * show_fib_entry_command (vlib_main_t * vm, unformat_input_t * input, diff --git a/src/vnet/fib/fib_path.c b/src/vnet/fib/fib_path.c index 1eb195d3a52..ef5d58c9e2b 100644 --- a/src/vnet/fib/fib_path.c +++ b/src/vnet/fib/fib_path.c @@ -1590,8 +1590,8 @@ fib_path_cmp_i (const fib_path_t *path1, path2->attached.fp_interface); break; case FIB_PATH_TYPE_RECURSIVE: - res = ip46_address_cmp(&path1->recursive.fp_nh, - &path2->recursive.fp_nh); + res = ip46_address_cmp(&path1->recursive.fp_nh.fp_ip, + &path2->recursive.fp_nh.fp_ip); if (0 == res) { diff --git a/src/vnet/fib/fib_table.h b/src/vnet/fib/fib_table.h index 59ebb0b0161..a11f0560338 100644 --- a/src/vnet/fib/fib_table.h +++ b/src/vnet/fib/fib_table.h @@ -957,4 +957,12 @@ extern void fib_table_sub_tree_walk(u32 fib_index, */ extern u8 *format_fib_table_memory(u8 *s, va_list *args); +/** + * Debug function + */ +#ifdef CLIB_DEBUG +extern void fib_table_assert_empty(const fib_table_t *fib_table); +#endif + + #endif diff --git a/src/vnet/fib/fib_types.c b/src/vnet/fib/fib_types.c index 3ac5c1d1183..c859913eeff 100644 --- a/src/vnet/fib/fib_types.c +++ b/src/vnet/fib/fib_types.c @@ -166,6 +166,21 @@ fib_prefix_is_cover (const fib_prefix_t *p1, return (0); } +u8 +fib_prefix_get_host_length (fib_protocol_t proto) +{ + switch (proto) + { + case FIB_PROTOCOL_IP4: + return (32); + case FIB_PROTOCOL_IP6: + return (128); + case FIB_PROTOCOL_MPLS: + return (21); + } + return (0); +} + int fib_prefix_is_host (const fib_prefix_t *prefix) { diff --git a/src/vnet/fib/fib_types.h b/src/vnet/fib/fib_types.h index 11e5bf4ebae..333b2abd240 100644 --- a/src/vnet/fib/fib_types.h +++ b/src/vnet/fib/fib_types.h @@ -18,7 +18,7 @@ #include <stdbool.h> #include <vlib/vlib.h> -#include <vnet/ip/ip6_packet.h> +#include <vnet/ip/ip46_address.h> #include <vnet/mpls/packet.h> #include <vnet/dpo/dpo.h> #include <vnet/bier/bier_types.h> @@ -265,7 +265,7 @@ extern int fib_prefix_is_cover(const fib_prefix_t *p1, * \brief Return true is the prefix is a host prefix */ extern int fib_prefix_is_host(const fib_prefix_t *p); - +extern u8 fib_prefix_get_host_length (fib_protocol_t proto); /** * \brief Host prefix from ip diff --git a/src/vnet/fib/ip4_fib.c b/src/vnet/fib/ip4_fib.c index e4ff1bf77cc..d4ef698bc9a 100644 --- a/src/vnet/fib/ip4_fib.c +++ b/src/vnet/fib/ip4_fib.c @@ -182,7 +182,10 @@ ip4_fib_table_destroy (u32 fib_index) /* * validate no more routes. */ - ASSERT(0 == fib_table->ft_total_route_counts); +#ifdef CLIB_DEBUG + if (0 != fib_table->ft_total_route_counts) + fib_table_assert_empty(fib_table); +#endif vec_foreach(n_locks, fib_table->ft_src_route_counts) { diff --git a/src/vnet/fib/ip6_fib.c b/src/vnet/fib/ip6_fib.c index 06160c5a15a..784f52c0460 100644 --- a/src/vnet/fib/ip6_fib.c +++ b/src/vnet/fib/ip6_fib.c @@ -154,7 +154,11 @@ ip6_fib_table_destroy (u32 fib_index) /* * validate no more routes. */ - ASSERT(0 == fib_table->ft_total_route_counts); +#ifdef CLIB_DEBUG + if (0 != fib_table->ft_total_route_counts) + fib_table_assert_empty(fib_table); +#endif + vec_foreach_index(source, fib_table->ft_src_route_counts) { ASSERT(0 == fib_table->ft_src_route_counts[source]); diff --git a/src/vnet/ip-neighbor/ip4_neighbor.c b/src/vnet/ip-neighbor/ip4_neighbor.c new file mode 100644 index 00000000000..2a9e2675a78 --- /dev/null +++ b/src/vnet/ip-neighbor/ip4_neighbor.c @@ -0,0 +1,319 @@ +/* + * 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 <vnet/ip-neighbor/ip4_neighbor.h> +#include <vnet/ethernet/ethernet.h> + +void +ip4_neighbor_probe_dst (const ip_adjacency_t * adj, const ip4_address_t * dst) +{ + ip_interface_address_t *ia; + ip4_address_t *src; + + src = ip4_interface_address_matching_destination + (&ip4_main, + &adj->sub_type.nbr.next_hop.ip4, adj->rewrite_header.sw_if_index, &ia); + if (!src) + return; + + ip4_neighbor_probe (vlib_get_main (), vnet_get_main (), adj, src, dst); +} + +void +ip4_neighbor_advertise (vlib_main_t * vm, + vnet_main_t * vnm, + u32 sw_if_index, const ip4_address_t * addr) +{ + vnet_hw_interface_t *hi = vnet_get_sup_hw_interface (vnm, sw_if_index); + ip4_main_t *i4m = &ip4_main; + u8 *rewrite, rewrite_len; + + if (NULL == addr) + { + ip4_main_t *i4m = &ip4_main; + addr = ip4_interface_first_address (i4m, sw_if_index, 0); + } + + if (addr) + { + clib_warning ("Sending GARP for IP4 address %U on sw_if_idex %d", + format_ip4_address, addr, sw_if_index); + + /* Form GARP packet for output - Gratuitous ARP is an ARP request packet + where the interface IP/MAC pair is used for both source and request + MAC/IP pairs in the request */ + u32 bi = 0; + ethernet_arp_header_t *h = vlib_packet_template_get_packet + (vm, &i4m->ip4_arp_request_packet_template, &bi); + + if (!h) + return; + + mac_address_from_bytes (&h->ip4_over_ethernet[0].mac, hi->hw_address); + mac_address_from_bytes (&h->ip4_over_ethernet[1].mac, hi->hw_address); + h->ip4_over_ethernet[0].ip4 = addr[0]; + h->ip4_over_ethernet[1].ip4 = addr[0]; + + /* Setup MAC header with ARP Etype and broadcast DMAC */ + vlib_buffer_t *b = vlib_get_buffer (vm, bi); + rewrite = + ethernet_build_rewrite (vnm, sw_if_index, VNET_LINK_ARP, + VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST); + rewrite_len = vec_len (rewrite); + vlib_buffer_advance (b, -rewrite_len); + ethernet_header_t *e = vlib_buffer_get_current (b); + clib_memcpy_fast (e->dst_address, rewrite, rewrite_len); + vec_free (rewrite); + + /* Send GARP packet out the specified interface */ + vnet_buffer (b)->sw_if_index[VLIB_RX] = + vnet_buffer (b)->sw_if_index[VLIB_TX] = sw_if_index; + 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); + } +} + +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, adj_index0, sw_if_index0; + ip4_address_t resolve0, src0; + vlib_buffer_t *p0, *b0; + ip_adjacency_t *adj0; + 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); + sw_if_index0 = adj0->rewrite_header.sw_if_index; + + if (is_glean) + { + /* resolve the packet's destination */ + ip4_header_t *ip0 = vlib_buffer_get_current (p0); + resolve0 = ip0->dst_address; + src0 = adj0->sub_type.glean.receive_addr.ip4; + } + else + { + /* resolve the incomplete adj */ + resolve0 = adj0->sub_type.nbr.next_hop.ip4; + /* Src IP address in ARP header. */ + if (ip4_src_address_for_packet (lm, sw_if_index0, &src0)) + { + /* No source address available */ + p0->error = node->errors[IP4_ARP_ERROR_NO_SOURCE_ADDRESS]; + continue; + } + } + + /* combine the address and interface for the hash key */ + 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. */ + b0 = ip4_neighbor_probe (vm, vnm, adj0, &src0, &resolve0); + + if (PREDICT_TRUE (NULL != b0)) + { + /* copy the persistent fields from the original */ + clib_memcpy_fast (b0->opaque2, p0->opaque2, + sizeof (p0->opaque2)); + p0->error = node->errors[IP4_ARP_ERROR_REQUEST_SENT]; + } + else + { + p0->error = node->errors[IP4_ARP_ERROR_NO_BUFFERS]; + continue; + } + } + + 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] = "ip4-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] = "ip4-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); + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip-neighbor/ip4_neighbor.h b/src/vnet/ip-neighbor/ip4_neighbor.h new file mode 100644 index 00000000000..c52e2d446af --- /dev/null +++ b/src/vnet/ip-neighbor/ip4_neighbor.h @@ -0,0 +1,85 @@ +/* + * Copyright (c) 2019 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef __IP4_NEIGHBOR_H__ +#define __IP4_NEIGHBOR_H__ + +#include <vnet/ip/ip.h> +#include <vnet/ethernet/arp_packet.h> + +extern void ip4_neighbor_probe_dst (const ip_adjacency_t * adj, + const ip4_address_t * dst); +extern void ip4_neighbor_advertise (vlib_main_t * vm, + vnet_main_t * vnm, + u32 sw_if_index, + const ip4_address_t * addr); + +always_inline vlib_buffer_t * +ip4_neighbor_probe (vlib_main_t * vm, + vnet_main_t * vnm, + const ip_adjacency_t * adj0, + const ip4_address_t * src, const ip4_address_t * dst) +{ + vnet_hw_interface_t *hw_if0; + ethernet_arp_header_t *h0; + vlib_buffer_t *b0; + u32 bi0; + + /* Send ARP request. */ + h0 = vlib_packet_template_get_packet (vm, + &ip4_main.ip4_arp_request_packet_template, + &bi0); + /* Seems we're out of buffers */ + if (PREDICT_FALSE (!h0)) + return (NULL); + + b0 = vlib_get_buffer (vm, bi0); + + /* 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, adj0->rewrite_header.sw_if_index); + + /* Src ethernet address in ARP header. */ + mac_address_from_bytes (&h0->ip4_over_ethernet[0].mac, hw_if0->hw_address); + + h0->ip4_over_ethernet[0].ip4 = *src; + h0->ip4_over_ethernet[1].ip4 = *dst; + + VLIB_BUFFER_TRACE_TRAJECTORY_INIT (b0); + vnet_buffer (b0)->sw_if_index[VLIB_TX] = adj0->rewrite_header.sw_if_index; + + vlib_buffer_advance (b0, -adj0->rewrite_header.data_bytes); + + { + vlib_frame_t *f = vlib_get_frame_to_node (vm, hw_if0->output_node_index); + u32 *to_next = vlib_frame_vector_args (f); + to_next[0] = bi0; + f->n_vectors = 1; + vlib_put_frame_to_node (vm, hw_if0->output_node_index, f); + } + + return b0; +} + +#endif + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip-neighbor/ip6_neighbor.c b/src/vnet/ip-neighbor/ip6_neighbor.c new file mode 100644 index 00000000000..325db8c6277 --- /dev/null +++ b/src/vnet/ip-neighbor/ip6_neighbor.c @@ -0,0 +1,338 @@ +/* + * ip/ip6_neighbor.c: IP6 neighbor handling + * + * Copyright (c) 2010 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/ip-neighbor/ip6_neighbor.h> + +void +ip6_neighbor_probe_dst (const ip_adjacency_t * adj, const ip6_address_t * dst) +{ + ip_interface_address_t *ia; + ip6_address_t *src; + + src = ip6_interface_address_matching_destination + (&ip6_main, dst, adj->rewrite_header.sw_if_index, &ia); + + if (!src) + return; + + ip6_neighbor_probe (vlib_get_main (), vnet_get_main (), adj, src, dst); +} + +void +ip6_neighbor_advertise (vlib_main_t * vm, + vnet_main_t * vnm, + u32 sw_if_index, const ip6_address_t * addr) +{ + vnet_hw_interface_t *hi = vnet_get_sup_hw_interface (vnm, sw_if_index); + ip6_main_t *i6m = &ip6_main; + u8 *rewrite, rewrite_len; + u8 dst_address[6]; + + if (NULL == addr) + addr = ip6_interface_first_address (i6m, sw_if_index); + + if (addr) + { + clib_warning + ("Sending unsolicitated NA IP6 address %U on sw_if_idex %d", + format_ip6_address, addr, sw_if_index); + + /* Form unsolicited neighbor advertisement packet from NS pkt template */ + int bogus_length; + u32 bi = 0; + icmp6_neighbor_solicitation_header_t *h = + vlib_packet_template_get_packet (vm, + &ip6_neighbor_packet_template, + &bi); + if (!h) + return; + + ip6_set_reserved_multicast_address (&h->ip.dst_address, + IP6_MULTICAST_SCOPE_link_local, + IP6_MULTICAST_GROUP_ID_all_hosts); + h->ip.src_address = addr[0]; + h->neighbor.icmp.type = ICMP6_neighbor_advertisement; + h->neighbor.target_address = addr[0]; + h->neighbor.advertisement_flags = clib_host_to_net_u32 + (ICMP6_NEIGHBOR_ADVERTISEMENT_FLAG_OVERRIDE); + h->link_layer_option.header.type = + ICMP6_NEIGHBOR_DISCOVERY_OPTION_target_link_layer_address; + clib_memcpy (h->link_layer_option.ethernet_address, + hi->hw_address, vec_len (hi->hw_address)); + h->neighbor.icmp.checksum = + ip6_tcp_udp_icmp_compute_checksum (vm, 0, &h->ip, &bogus_length); + ASSERT (bogus_length == 0); + + /* Setup MAC header with IP6 Etype and mcast DMAC */ + vlib_buffer_t *b = vlib_get_buffer (vm, bi); + ip6_multicast_ethernet_address (dst_address, + IP6_MULTICAST_GROUP_ID_all_hosts); + rewrite = + ethernet_build_rewrite (vnm, sw_if_index, VNET_LINK_IP6, dst_address); + rewrite_len = vec_len (rewrite); + vlib_buffer_advance (b, -rewrite_len); + ethernet_header_t *e = vlib_buffer_get_current (b); + clib_memcpy (e->dst_address, rewrite, rewrite_len); + vec_free (rewrite); + + /* Send unsolicited ND advertisement packet out the specified interface */ + vnet_buffer (b)->sw_if_index[VLIB_RX] = + vnet_buffer (b)->sw_if_index[VLIB_TX] = sw_if_index; + 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); + } +} + +typedef enum +{ + IP6_NBR_NEXT_DROP, + IP6_NBR_NEXT_REPLY_TX, + IP6_NBR_N_NEXT, +} ip6_discover_neighbor_next_t; + +typedef enum +{ + IP6_NBR_ERROR_DROP, + IP6_NBR_ERROR_REQUEST_SENT, + IP6_NBR_ERROR_NO_SOURCE_ADDRESS, + IP6_NBR_ERROR_NO_BUFFERS, +} ip6_discover_neighbor_error_t; + +static uword +ip6_discover_neighbor_inline (vlib_main_t * vm, + vlib_node_runtime_t * node, + vlib_frame_t * frame, int is_glean) +{ + vnet_main_t *vnm = vnet_get_main (); + ip6_main_t *im = &ip6_main; + u32 *from, *to_next_drop; + uword n_left_from, n_left_to_next_drop; + u64 seed; + u32 thread_index = vm->thread_index; + + if (node->flags & VLIB_NODE_FLAG_TRACE) + ip6_forward_next_trace (vm, node, frame, VLIB_TX); + + seed = throttle_seed (&im->nd_throttle, thread_index, vlib_time_now (vm)); + + from = vlib_frame_vector_args (frame); + n_left_from = frame->n_vectors; + + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, IP6_NBR_NEXT_DROP, + to_next_drop, n_left_to_next_drop); + + while (n_left_from > 0 && n_left_to_next_drop > 0) + { + u32 pi0, adj_index0, sw_if_index0, drop0, r0; + vnet_hw_interface_t *hw_if0; + vlib_buffer_t *p0, *b0; + ip_adjacency_t *adj0; + ip6_address_t src; + ip6_header_t *ip0; + + pi0 = from[0]; + + p0 = vlib_get_buffer (vm, pi0); + + adj_index0 = vnet_buffer (p0)->ip.adj_index[VLIB_TX]; + + ip0 = vlib_buffer_get_current (p0); + + adj0 = adj_get (adj_index0); + + if (!is_glean) + { + ip0->dst_address.as_u64[0] = + adj0->sub_type.nbr.next_hop.ip6.as_u64[0]; + ip0->dst_address.as_u64[1] = + adj0->sub_type.nbr.next_hop.ip6.as_u64[1]; + } + + sw_if_index0 = adj0->rewrite_header.sw_if_index; + vnet_buffer (p0)->sw_if_index[VLIB_TX] = sw_if_index0; + + /* combine the address and interface for a hash */ + r0 = ip6_address_hash_to_u64 (&ip0->dst_address) ^ sw_if_index0; + + drop0 = throttle_check (&im->nd_throttle, thread_index, r0, seed); + + from += 1; + n_left_from -= 1; + to_next_drop[0] = pi0; + to_next_drop += 1; + n_left_to_next_drop -= 1; + + hw_if0 = vnet_get_sup_hw_interface (vnm, sw_if_index0); + + /* If the interface is link-down, drop the pkt */ + if (!(hw_if0->flags & VNET_HW_INTERFACE_FLAG_LINK_UP)) + drop0 = 1; + + if (!ip6_link_is_enabled (sw_if_index0)) + drop0 = 1; + + /* + * 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) + drop0 = 1; + + if (drop0) + { + p0->error = node->errors[IP6_NBR_ERROR_DROP]; + continue; + } + + /* + * Choose source address based on destination lookup + * adjacency. + */ + if (!ip6_src_address_for_packet (sw_if_index0, + &ip0->dst_address, &src)) + { + /* There is no address on the interface */ + p0->error = node->errors[IP6_NBR_ERROR_NO_SOURCE_ADDRESS]; + continue; + } + + b0 = ip6_neighbor_probe (vm, vnm, adj0, &src, &ip0->dst_address); + + if (PREDICT_TRUE (NULL != b0)) + { + clib_memcpy_fast (b0->opaque2, p0->opaque2, + sizeof (p0->opaque2)); + b0->flags |= p0->flags & VLIB_BUFFER_IS_TRACED; + b0->trace_handle = p0->trace_handle; + p0->error = node->errors[IP6_NBR_ERROR_REQUEST_SENT]; + } + else + { + /* There is no address on the interface */ + p0->error = node->errors[IP6_NBR_ERROR_NO_BUFFERS]; + continue; + } + } + + vlib_put_next_frame (vm, node, IP6_NBR_NEXT_DROP, n_left_to_next_drop); + } + + return frame->n_vectors; +} + +static uword +ip6_discover_neighbor (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + return (ip6_discover_neighbor_inline (vm, node, frame, 0)); +} + +static uword +ip6_glean (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + return (ip6_discover_neighbor_inline (vm, node, frame, 1)); +} + +static char *ip6_discover_neighbor_error_strings[] = { + [IP6_NBR_ERROR_DROP] = "address overflow drops", + [IP6_NBR_ERROR_REQUEST_SENT] = "neighbor solicitations sent", + [IP6_NBR_ERROR_NO_SOURCE_ADDRESS] = "no source address for ND solicitation", + [IP6_NBR_ERROR_NO_BUFFERS] = "no buffers", +}; + +/* *INDENT-OFF* */ +VLIB_REGISTER_NODE (ip6_glean_node) = +{ + .function = ip6_glean, + .name = "ip6-glean", + .vector_size = sizeof (u32), + .format_trace = format_ip6_forward_next_trace, + .n_errors = ARRAY_LEN (ip6_discover_neighbor_error_strings), + .error_strings = ip6_discover_neighbor_error_strings, + .n_next_nodes = IP6_NBR_N_NEXT, + .next_nodes = + { + [IP6_NBR_NEXT_DROP] = "ip6-drop", + [IP6_NBR_NEXT_REPLY_TX] = "ip6-rewrite-mcast", + }, +}; +VLIB_REGISTER_NODE (ip6_discover_neighbor_node) = +{ + .function = ip6_discover_neighbor, + .name = "ip6-discover-neighbor", + .vector_size = sizeof (u32), + .format_trace = format_ip6_forward_next_trace, + .n_errors = ARRAY_LEN (ip6_discover_neighbor_error_strings), + .error_strings = ip6_discover_neighbor_error_strings, + .n_next_nodes = IP6_NBR_N_NEXT, + .next_nodes = + { + [IP6_NBR_NEXT_DROP] = "ip6-drop", + [IP6_NBR_NEXT_REPLY_TX] = "ip6-rewrite-mcast", + }, +}; +/* *INDENT-ON* */ + +/* Template used to generate IP6 neighbor solicitation packets. */ +vlib_packet_template_t ip6_neighbor_packet_template; + +static clib_error_t * +ip6_neighbor_init (vlib_main_t * vm) +{ + icmp6_neighbor_solicitation_header_t p; + + clib_memset (&p, 0, sizeof (p)); + + p.ip.ip_version_traffic_class_and_flow_label = + clib_host_to_net_u32 (0x6 << 28); + p.ip.payload_length = + clib_host_to_net_u16 (sizeof (p) - + STRUCT_OFFSET_OF + (icmp6_neighbor_solicitation_header_t, neighbor)); + p.ip.protocol = IP_PROTOCOL_ICMP6; + p.ip.hop_limit = 255; + ip6_set_solicited_node_multicast_address (&p.ip.dst_address, 0); + + p.neighbor.icmp.type = ICMP6_neighbor_solicitation; + + p.link_layer_option.header.type = + ICMP6_NEIGHBOR_DISCOVERY_OPTION_source_link_layer_address; + p.link_layer_option.header.n_data_u64s = + sizeof (p.link_layer_option) / sizeof (u64); + + vlib_packet_template_init (vm, + &ip6_neighbor_packet_template, &p, sizeof (p), + /* alloc chunk size */ 8, + "ip6 neighbor discovery"); + + return NULL; +} + +VLIB_INIT_FUNCTION (ip6_neighbor_init); + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip-neighbor/ip6_neighbor.h b/src/vnet/ip-neighbor/ip6_neighbor.h new file mode 100644 index 00000000000..934a2885ee1 --- /dev/null +++ b/src/vnet/ip-neighbor/ip6_neighbor.h @@ -0,0 +1,112 @@ +/* + * ip/ip6_neighbor.h: IP6 NS transmit + * + * Copyright (c) 2019 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef __IP6_NEIGHBOR_H__ +#define __IP6_NEIGHBOR_H__ + +#include <vlib/vlib.h> +#include <vnet/ip/format.h> +#include <vnet/ip/ip6.h> +#include <vnet/ip/ip6_link.h> +#include <vnet/ip/icmp46_packet.h> +#include <vnet/ethernet/ethernet.h> +#include <vnet/adj/adj_internal.h> + +/* Template used to generate IP6 neighbor solicitation packets. */ +extern vlib_packet_template_t ip6_neighbor_packet_template; + +extern void ip6_neighbor_advertise (vlib_main_t * vm, + vnet_main_t * vnm, + u32 sw_if_index, + const ip6_address_t * addr); + +extern void ip6_neighbor_probe_dst (const ip_adjacency_t * adj, + const ip6_address_t * dst); + +always_inline vlib_buffer_t * +ip6_neighbor_probe (vlib_main_t * vm, + vnet_main_t * vnm, + const ip_adjacency_t * adj, + const ip6_address_t * src, const ip6_address_t * dst) +{ + icmp6_neighbor_solicitation_header_t *h0; + vnet_hw_interface_t *hw_if0; + vlib_buffer_t *b0; + int bogus_length; + u32 bi0 = 0; + + h0 = vlib_packet_template_get_packet + (vm, &ip6_neighbor_packet_template, &bi0); + if (!h0) + return NULL;; + + b0 = vlib_get_buffer (vm, bi0); + + hw_if0 = vnet_get_sup_hw_interface (vnm, adj->rewrite_header.sw_if_index); + + /* + * Destination address is a solicited node multicast address. + * We need to fill in + * the low 24 bits with low 24 bits of target's address. + */ + h0->ip.src_address = *src; + h0->ip.dst_address.as_u8[13] = dst->as_u8[13]; + h0->ip.dst_address.as_u8[14] = dst->as_u8[14]; + h0->ip.dst_address.as_u8[15] = dst->as_u8[15]; + + h0->neighbor.target_address = *dst; + + clib_memcpy (h0->link_layer_option.ethernet_address, + hw_if0->hw_address, vec_len (hw_if0->hw_address)); + + /* $$$$ appears we need this; why is the checksum non-zero? */ + h0->neighbor.icmp.checksum = 0; + h0->neighbor.icmp.checksum = + ip6_tcp_udp_icmp_compute_checksum (vm, 0, &h0->ip, &bogus_length); + + ASSERT (bogus_length == 0); + VLIB_BUFFER_TRACE_TRAJECTORY_INIT (b0); + + vnet_buffer (b0)->sw_if_index[VLIB_TX] = adj->rewrite_header.sw_if_index; + + /* Use the link's mcast adj to ship the packet */ + vnet_buffer (b0)->ip.adj_index[VLIB_TX] = + ip6_link_get_mcast_adj (adj->rewrite_header.sw_if_index); + adj = adj_get (vnet_buffer (b0)->ip.adj_index[VLIB_TX]); + + b0->flags |= VNET_BUFFER_F_LOCALLY_ORIGINATED; + + { + vlib_frame_t *f = vlib_get_frame_to_node (vm, adj->ia_node_index); + u32 *to_next = vlib_frame_vector_args (f); + to_next[0] = bi0; + f->n_vectors = 1; + vlib_put_frame_to_node (vm, adj->ia_node_index, f); + } + + return b0; +} + +#endif + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip-neighbor/ip_neighbor.api b/src/vnet/ip-neighbor/ip_neighbor.api new file mode 100644 index 00000000000..a8fb24a712c --- /dev/null +++ b/src/vnet/ip-neighbor/ip_neighbor.api @@ -0,0 +1,168 @@ +/* Hey Emacs use -*- mode: C -*- */ +/* + * Copyright (c) 2018 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. + */ + +/** \file + + This file defines vpp IP neighbor control-plane API messages which are generally + called through a shared memory interface. +*/ + +option version = "1.0.0"; + +import "vnet/ip/ip_types.api"; +import "vnet/ethernet/ethernet_types.api"; +import "vnet/interface_types.api"; + +/** \brief IP neighbor flags + @param is_static - A static neighbor Entry - there are not flushed + If the interface goes down. + @param is_no_fib_entry - Do not create a corresponding entry in the FIB + table for the neighbor. +*/ +enum ip_neighbor_flags: u8 +{ + IP_API_NEIGHBOR_FLAG_NONE = 0, + IP_API_NEIGHBOR_FLAG_STATIC = 0x1, + IP_API_NEIGHBOR_FLAG_NO_FIB_ENTRY = 0x2, +}; + +/** \brief IP neighbor + @param sw_if_index - interface used to reach neighbor + @param flags - flags for the nieghbor + @param mac_address - l2 address of the neighbor + @param ip_address - ip4 or ip6 address of the neighbor +*/ +typedef ip_neighbor { + vl_api_interface_index_t sw_if_index; + vl_api_ip_neighbor_flags_t flags; + vl_api_mac_address_t mac_address; + vl_api_address_t ip_address; +}; + +/** \brief IP neighbor add / del request + @param client_index - opaque cookie to identify the sender + @param context - sender context, to match reply w/ request + @param is_add - 1 to add neighbor, 0 to delete + @param neighbor - the neighor to add/remove +*/ +define ip_neighbor_add_del +{ + u32 client_index; + u32 context; + /* 1 = add, 0 = delete */ + bool is_add; + vl_api_ip_neighbor_t neighbor; +}; +/** \brief IP neighbor add / del reply + @param client_index - opaque cookie to identify the sender + @param context - sender context, to match reply w/ request + @param retval - return value + @param stats_index - the index to use for this neighbor in the stats segement +*/ +define ip_neighbor_add_del_reply +{ + u32 context; + i32 retval; + u32 stats_index; +}; + +/** \brief Dump IP neighboors + @param client_index - opaque cookie to identify the sender + @param context - sender context, to match reply w/ request + @param sw_if_index - the interface to dump neighboors, ~0 == all + @param af - address family is ipv[6|4] +*/ +define ip_neighbor_dump +{ + u32 client_index; + u32 context; + vl_api_interface_index_t sw_if_index [default=0xffffffff]; + vl_api_address_family_t af; +}; + +/** \brief IP neighboors dump response + @param context - sender context which was passed in the request + @param neighbour - the neighbor +*/ +define ip_neighbor_details { + u32 context; + vl_api_ip_neighbor_t neighbor; +}; + +/** \brief Enable/disable periodic IP neighbor scan + @param client_index - opaque cookie to identify the sender + @param context - sender context, to match reply w/ request + @param af - Address family v4/v6 + @param max_number - The maximum number of neighbours that will be created. + default 50k + @param max_age - The maximum age (in seconds) before an inactive neighbour + is flushed + default 0 => never + @param recycle - If max_number of neighbours is reached and new ones need + to be created should the oldest neighbour be 'recycled'. +*/ +autoreply define ip_neighbor_config +{ + u32 client_index; + u32 context; + vl_api_address_family_t af; + u32 max_number; + u32 max_age; + bool recycle; +}; + +/** \brief Register for IP4 ARP resolution event on receing ARP reply or + MAC/IP info from ARP requests in L2 BDs + @param client_index - opaque cookie to identify the sender + @param context - sender context, to match reply w/ request + @param enable - 1 => register for events, 0 => cancel registration + @param pid - sender's pid + @param ip - exact IP address of interested neighbor resolution event + @param sw_if_index - interface on which the IP address is present. +*/ +autoreply define want_ip_neighbor_events +{ + u32 client_index; + u32 context; + bool enable; + u32 pid; + vl_api_address_t ip; + vl_api_interface_index_t sw_if_index [default=0xffffffff]; +}; + +/** \brief Tell client about an IP4 ARP resolution event or + MAC/IP info from ARP requests in L2 BDs + @param client_index - opaque cookie to identify the sender + @param pid - client pid registered to receive notification + @param neighbor - new neighbor created +*/ +define ip_neighbor_event +{ + u32 client_index; + u32 pid; + vl_api_ip_neighbor_t neighbor; +}; + +service { + rpc want_ip_neighbor_events returns want_ip_neighbor_events_reply + events ip_neighbor_event; +}; + +/* + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip-neighbor/ip_neighbor.c b/src/vnet/ip-neighbor/ip_neighbor.c new file mode 100644 index 00000000000..11684eb9de5 --- /dev/null +++ b/src/vnet/ip-neighbor/ip_neighbor.c @@ -0,0 +1,1664 @@ +/* + * src/vnet/ip/ip_neighboor.c: ip neighbor generic handling + * + * Copyright (c) 2018 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vppinfra/llist.h> + +#include <vnet/ip-neighbor/ip_neighbor.h> +#include <vnet/ip-neighbor/ip4_neighbor.h> +#include <vnet/ip-neighbor/ip6_neighbor.h> +#include <vnet/ip-neighbor/ip_neighbor_watch.h> + +#include <vnet/ip/ip6_ll_table.h> +#include <vnet/fib/fib_table.h> +#include <vnet/adj/adj_mcast.h> + +/** Pool for All IP neighbors */ +static ip_neighbor_t *ip_neighbor_pool; + +/** protocol specific lists of time sorted neighbors */ +index_t ip_neighbor_list_head[IP46_N_TYPES]; + +typedef struct ip_neighbor_elt_t_ +{ + clib_llist_anchor_t ipne_anchor; + index_t ipne_index; +} ip_neighbor_elt_t; + +/** Pool of linked list elemeents */ +ip_neighbor_elt_t *ip_neighbor_elt_pool; + +typedef struct ip_neighbor_db_t_ +{ + /** per interface hash */ + uword **ipndb_hash; + /** per-protocol limit - max number of neighbors*/ + u32 ipndb_limit; + /** max age of a neighbor before it's forcibly evicted */ + u32 ipndb_age; + /** when the limit is reached and new neighbors are created, should + * we recycle an old one */ + bool ipndb_recycle; + /** per-protocol number of elements */ + u32 ipndb_n_elts; + /** per-protocol number of elements per-fib-index*/ + u32 *ipndb_n_elts_per_fib; +} ip_neighbor_db_t; + +static vlib_log_class_t ipn_logger; + +/* DBs of neighbours one per AF */ +/* *INDENT-OFF* */ +static ip_neighbor_db_t ip_neighbor_db[IP46_N_TYPES] = { + [IP46_TYPE_IP4] = { + .ipndb_limit = 50000, + /* Default to not aging and not recycling */ + .ipndb_age = 0, + .ipndb_recycle = false, + }, + [IP46_TYPE_IP6] = { + .ipndb_limit = 50000, + /* Default to not aging and not recycling */ + .ipndb_age = 0, + .ipndb_recycle = false, + } +}; +/* *INDENT-ON* */ + +#define IP_NEIGHBOR_DBG(...) \ + vlib_log_debug (ipn_logger, __VA_ARGS__); + +#define IP_NEIGHBOR_INFO(...) \ + vlib_log_notice (ipn_logger, __VA_ARGS__); + +ip_neighbor_t * +ip_neighbor_get (index_t ipni) +{ + if (pool_is_free_index (ip_neighbor_pool, ipni)) + return (NULL); + + return (pool_elt_at_index (ip_neighbor_pool, ipni)); +} + +static index_t +ip_neighbor_get_index (const ip_neighbor_t * ipn) +{ + return (ipn - ip_neighbor_pool); +} + +static bool +ip_neighbor_is_dynamic (const ip_neighbor_t * ipn) +{ + return (ipn->ipn_flags & IP_NEIGHBOR_FLAG_DYNAMIC); +} + +const ip46_address_t * +ip_neighbor_get_ip (const ip_neighbor_t * ipn) +{ + return (&ipn->ipn_key->ipnk_ip); +} + +const mac_address_t * +ip_neighbor_get_mac (const ip_neighbor_t * ipn) +{ + return (&ipn->ipn_mac); +} + +const u32 +ip_neighbor_get_sw_if_index (const ip_neighbor_t * ipn) +{ + return (ipn->ipn_key->ipnk_sw_if_index); +} + +static void +ip_neighbor_list_remove (ip_neighbor_t * ipn) +{ + /* new neighbours, are added to the head of the list, since the + * list is time sorted, newest first */ + ip_neighbor_elt_t *elt; + + if (~0 != ipn->ipn_elt) + { + elt = pool_elt_at_index (ip_neighbor_elt_pool, ipn->ipn_elt); + + clib_llist_remove (ip_neighbor_elt_pool, ipne_anchor, elt); + } +} + +static void +ip_neighbor_refresh (ip_neighbor_t * ipn) +{ + /* new neighbours, are added to the head of the list, since the + * list is time sorted, newest first */ + ip_neighbor_elt_t *elt, *head; + + ipn->ipn_time_last_updated = vlib_time_now (vlib_get_main ()); + ipn->ipn_n_probes = 0; + + if (ip_neighbor_is_dynamic (ipn)) + { + if (~0 == ipn->ipn_elt) + /* first time insertion */ + pool_get_zero (ip_neighbor_elt_pool, elt); + else + { + /* already inserted - extract first */ + elt = pool_elt_at_index (ip_neighbor_elt_pool, ipn->ipn_elt); + + clib_llist_remove (ip_neighbor_elt_pool, ipne_anchor, elt); + } + head = pool_elt_at_index (ip_neighbor_elt_pool, + ip_neighbor_list_head[ipn-> + ipn_key->ipnk_type]); + + elt->ipne_index = ip_neighbor_get_index (ipn); + clib_llist_add (ip_neighbor_elt_pool, ipne_anchor, elt, head); + ipn->ipn_elt = elt - ip_neighbor_elt_pool; + } +} + +static void +ip_neighbor_db_add (const ip_neighbor_t * ipn) +{ + vec_validate (ip_neighbor_db[ipn->ipn_key->ipnk_type].ipndb_hash, + ipn->ipn_key->ipnk_sw_if_index); + + if (!ip_neighbor_db[ipn->ipn_key->ipnk_type].ipndb_hash + [ipn->ipn_key->ipnk_sw_if_index]) + ip_neighbor_db[ipn->ipn_key->ipnk_type].ipndb_hash[ipn-> + ipn_key->ipnk_sw_if_index] + = hash_create_mem (0, sizeof (ip_neighbor_key_t), sizeof (index_t)); + + hash_set_mem (ip_neighbor_db[ipn->ipn_key->ipnk_type].ipndb_hash + [ipn->ipn_key->ipnk_sw_if_index], ipn->ipn_key, + ip_neighbor_get_index (ipn)); + + ip_neighbor_db[ipn->ipn_key->ipnk_type].ipndb_n_elts++; +} + +static void +ip_neighbor_db_remove (const ip_neighbor_key_t * key) +{ + vec_validate (ip_neighbor_db[key->ipnk_type].ipndb_hash, + key->ipnk_sw_if_index); + + hash_unset_mem (ip_neighbor_db[key->ipnk_type].ipndb_hash + [key->ipnk_sw_if_index], key); + + ip_neighbor_db[key->ipnk_type].ipndb_n_elts--; +} + +static ip_neighbor_t * +ip_neighbor_db_find (const ip_neighbor_key_t * key) +{ + uword *p; + + if (key->ipnk_sw_if_index >= + vec_len (ip_neighbor_db[key->ipnk_type].ipndb_hash)) + return NULL; + + p = + hash_get_mem (ip_neighbor_db[key->ipnk_type].ipndb_hash + [key->ipnk_sw_if_index], key); + + if (p) + return ip_neighbor_get (p[0]); + + return (NULL); +} + +static u8 +ip46_type_pfx_len (ip46_type_t type) +{ + return (type == IP46_TYPE_IP4 ? 32 : 128); +} + +static void +ip_neighbor_adj_fib_add (ip_neighbor_t * ipn, u32 fib_index) +{ + if (ipn->ipn_key->ipnk_type == IP46_TYPE_IP6 && + ip6_address_is_link_local_unicast (&ipn->ipn_key->ipnk_ip.ip6)) + { + ip6_ll_prefix_t pfx = { + .ilp_addr = ipn->ipn_key->ipnk_ip.ip6, + .ilp_sw_if_index = ipn->ipn_key->ipnk_sw_if_index, + }; + ipn->ipn_fib_entry_index = + ip6_ll_table_entry_update (&pfx, FIB_ROUTE_PATH_FLAG_NONE); + } + else + { + fib_protocol_t fproto; + + fproto = fib_proto_from_ip46 (ipn->ipn_key->ipnk_type); + + fib_prefix_t pfx = { + .fp_len = ip46_type_pfx_len (ipn->ipn_key->ipnk_type), + .fp_proto = fproto, + .fp_addr = ipn->ipn_key->ipnk_ip, + }; + + ipn->ipn_fib_entry_index = + fib_table_entry_path_add (fib_index, &pfx, FIB_SOURCE_ADJ, + FIB_ENTRY_FLAG_ATTACHED, + fib_proto_to_dpo (fproto), + &pfx.fp_addr, + ipn->ipn_key->ipnk_sw_if_index, + ~0, 1, NULL, FIB_ROUTE_PATH_FLAG_NONE); + + vec_validate (ip_neighbor_db + [ipn->ipn_key->ipnk_type].ipndb_n_elts_per_fib, + fib_index); + + ip_neighbor_db[ipn->ipn_key-> + ipnk_type].ipndb_n_elts_per_fib[fib_index]++; + + if (1 == + ip_neighbor_db[ipn->ipn_key-> + ipnk_type].ipndb_n_elts_per_fib[fib_index]) + fib_table_lock (fib_index, fproto, FIB_SOURCE_ADJ); + } +} + +static void +ip_neighbor_adj_fib_remove (ip_neighbor_t * ipn, u32 fib_index) +{ + if (FIB_NODE_INDEX_INVALID != ipn->ipn_fib_entry_index) + { + if (ipn->ipn_key->ipnk_type == IP46_TYPE_IP6 && + ip6_address_is_link_local_unicast (&ipn->ipn_key->ipnk_ip.ip6)) + { + ip6_ll_prefix_t pfx = { + .ilp_addr = ipn->ipn_key->ipnk_ip.ip6, + .ilp_sw_if_index = ipn->ipn_key->ipnk_sw_if_index, + }; + ip6_ll_table_entry_delete (&pfx); + } + else + { + fib_protocol_t fproto; + + fproto = fib_proto_from_ip46 (ipn->ipn_key->ipnk_type); + + fib_prefix_t pfx = { + .fp_len = ip46_type_pfx_len (ipn->ipn_key->ipnk_type), + .fp_proto = fproto, + .fp_addr = ipn->ipn_key->ipnk_ip, + }; + + fib_table_entry_path_remove (fib_index, + &pfx, + FIB_SOURCE_ADJ, + fib_proto_to_dpo (fproto), + &pfx.fp_addr, + ipn->ipn_key->ipnk_sw_if_index, + ~0, 1, FIB_ROUTE_PATH_FLAG_NONE); + + ip_neighbor_db[ipn->ipn_key-> + ipnk_type].ipndb_n_elts_per_fib[fib_index]--; + + if (0 == + ip_neighbor_db[ipn->ipn_key-> + ipnk_type].ipndb_n_elts_per_fib[fib_index]) + fib_table_unlock (fib_index, fproto, FIB_SOURCE_ADJ); + } + } +} + +static void +ip_neighbor_mk_complete (adj_index_t ai, ip_neighbor_t * ipn) +{ + adj_nbr_update_rewrite (ai, ADJ_NBR_REWRITE_FLAG_COMPLETE, + ethernet_build_rewrite (vnet_get_main (), + ipn-> + ipn_key->ipnk_sw_if_index, + adj_get_link_type (ai), + ipn->ipn_mac.bytes)); +} + +static void +ip_neighbor_mk_incomplete (adj_index_t ai) +{ + ip_adjacency_t *adj = adj_get (ai); + + adj_nbr_update_rewrite (ai, + ADJ_NBR_REWRITE_FLAG_INCOMPLETE, + ethernet_build_rewrite (vnet_get_main (), + adj-> + rewrite_header.sw_if_index, + adj_get_link_type (ai), + VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST)); +} + +static adj_walk_rc_t +ip_neighbor_mk_complete_walk (adj_index_t ai, void *ctx) +{ + ip_neighbor_t *ipn = ctx; + + ip_neighbor_mk_complete (ai, ipn); + + return (ADJ_WALK_RC_CONTINUE); +} + +static adj_walk_rc_t +ip_neighbor_mk_incomplete_walk (adj_index_t ai, void *ctx) +{ + ip_neighbor_mk_incomplete (ai); + + return (ADJ_WALK_RC_CONTINUE); +} + +static void +ip_neighbor_free (ip_neighbor_t * ipn) +{ + IP_NEIGHBOR_DBG ("free: %U", format_ip_neighbor, + ip_neighbor_get_index (ipn)); + + adj_nbr_walk_nh (ipn->ipn_key->ipnk_sw_if_index, + fib_proto_from_ip46 (ipn->ipn_key->ipnk_type), + &ipn->ipn_key->ipnk_ip, + ip_neighbor_mk_incomplete_walk, ipn); + ip_neighbor_adj_fib_remove + (ipn, + fib_table_get_index_for_sw_if_index + (fib_proto_from_ip46 (ipn->ipn_key->ipnk_type), + ipn->ipn_key->ipnk_sw_if_index)); + + ip_neighbor_list_remove (ipn); + ip_neighbor_db_remove (ipn->ipn_key); + clib_mem_free (ipn->ipn_key); + + pool_put (ip_neighbor_pool, ipn); +} + +static bool +ip_neighbor_force_reuse (ip46_type_t type) +{ + if (!ip_neighbor_db[type].ipndb_recycle) + return false; + + /* pluck the oldest entry, which is the one from the end of the list */ + ip_neighbor_elt_t *elt, *head; + + head = + pool_elt_at_index (ip_neighbor_elt_pool, ip_neighbor_list_head[type]); + + if (clib_llist_is_empty (ip_neighbor_elt_pool, ipne_anchor, head)) + return (false); + + elt = clib_llist_prev (ip_neighbor_elt_pool, ipne_anchor, head); + ip_neighbor_free (ip_neighbor_get (elt->ipne_index)); + + return (true); +} + +static ip_neighbor_t * +ip_neighbor_alloc (const ip_neighbor_key_t * key, + const mac_address_t * mac, ip_neighbor_flags_t flags) +{ + ip_neighbor_t *ipn; + + if (ip_neighbor_db[key->ipnk_type].ipndb_limit && + (ip_neighbor_db[key->ipnk_type].ipndb_n_elts >= + ip_neighbor_db[key->ipnk_type].ipndb_limit)) + { + if (!ip_neighbor_force_reuse (key->ipnk_type)) + return (NULL); + } + + pool_get_zero (ip_neighbor_pool, ipn); + + ipn->ipn_key = clib_mem_alloc (sizeof (*ipn->ipn_key)); + clib_memcpy (ipn->ipn_key, key, sizeof (*ipn->ipn_key)); + + ipn->ipn_fib_entry_index = FIB_NODE_INDEX_INVALID; + ipn->ipn_flags = flags; + ipn->ipn_elt = ~0; + + mac_address_copy (&ipn->ipn_mac, mac); + + ip_neighbor_db_add (ipn); + + /* create the adj-fib. the entry in the FIB table for the peer's interface */ + if (!(ipn->ipn_flags & IP_NEIGHBOR_FLAG_NO_FIB_ENTRY)) + ip_neighbor_adj_fib_add + (ipn, fib_table_get_index_for_sw_if_index + (fib_proto_from_ip46 (ipn->ipn_key->ipnk_type), + ipn->ipn_key->ipnk_sw_if_index)); + + return (ipn); +} + +int +ip_neighbor_add (const ip46_address_t * ip, + ip46_type_t type, + const mac_address_t * mac, + u32 sw_if_index, + ip_neighbor_flags_t flags, u32 * stats_index) +{ + fib_protocol_t fproto; + ip_neighbor_t *ipn; + + /* main thread only */ + ASSERT (0 == vlib_get_thread_index ()); + + fproto = fib_proto_from_ip46 (type); + + const ip_neighbor_key_t key = { + .ipnk_ip = *ip, + .ipnk_sw_if_index = sw_if_index, + .ipnk_type = type, + }; + + ipn = ip_neighbor_db_find (&key); + + if (ipn) + { + IP_NEIGHBOR_DBG ("update: %U, %U", + format_vnet_sw_if_index_name, vnet_get_main (), + sw_if_index, format_ip46_address, ip, type, + format_ip_neighbor_flags, flags, format_mac_address_t, + mac); + + /* Refuse to over-write static neighbor entry. */ + if (!(flags & IP_NEIGHBOR_FLAG_STATIC) && + (ipn->ipn_flags & IP_NEIGHBOR_FLAG_STATIC)) + { + /* if MAC address match, still check to send event */ + if (0 == mac_address_cmp (&ipn->ipn_mac, mac)) + goto check_customers; + return -2; + } + + /* + * prevent a DoS attack from the data-plane that + * spams us with no-op updates to the MAC address + */ + if (0 == mac_address_cmp (&ipn->ipn_mac, mac)) + { + ip_neighbor_refresh (ipn); + goto check_customers; + } + + mac_address_copy (&ipn->ipn_mac, mac); + + /* A dynamic entry can become static, but not vice-versa. + * i.e. since if it was programmed by the CP then it must + * be removed by the CP */ + if ((flags & IP_NEIGHBOR_FLAG_STATIC) && + !(ipn->ipn_flags & IP_NEIGHBOR_FLAG_STATIC)) + { + ip_neighbor_list_remove (ipn); + ipn->ipn_flags |= IP_NEIGHBOR_FLAG_STATIC; + ipn->ipn_flags &= ~IP_NEIGHBOR_FLAG_DYNAMIC; + } + } + else + { + IP_NEIGHBOR_INFO ("add: %U, %U", + format_vnet_sw_if_index_name, vnet_get_main (), + sw_if_index, format_ip46_address, ip, type, + format_ip_neighbor_flags, flags, format_mac_address_t, + mac); + + ipn = ip_neighbor_alloc (&key, mac, flags); + + if (NULL == ipn) + return VNET_API_ERROR_LIMIT_EXCEEDED; + } + + /* Update time stamp and flags. */ + ip_neighbor_refresh (ipn); + + adj_nbr_walk_nh (ipn->ipn_key->ipnk_sw_if_index, + fproto, &ipn->ipn_key->ipnk_ip, + ip_neighbor_mk_complete_walk, ipn); + +check_customers: + /* Customer(s) requesting event for this address? */ + ip_neighbor_publish (ip_neighbor_get_index (ipn)); + + if (stats_index) + *stats_index = adj_nbr_find (fproto, + fib_proto_to_link (fproto), + &ipn->ipn_key->ipnk_ip, + ipn->ipn_key->ipnk_sw_if_index); + return 0; +} + +int +ip_neighbor_del (const ip46_address_t * ip, ip46_type_t type, u32 sw_if_index) +{ + ip_neighbor_t *ipn; + + /* main thread only */ + ASSERT (0 == vlib_get_thread_index ()); + + IP_NEIGHBOR_INFO ("delete: %U, %U", + format_vnet_sw_if_index_name, vnet_get_main (), + sw_if_index, format_ip46_address, ip, type); + + const ip_neighbor_key_t key = { + .ipnk_ip = *ip, + .ipnk_sw_if_index = sw_if_index, + .ipnk_type = type, + }; + + ipn = ip_neighbor_db_find (&key); + + if (NULL == ipn) + return (VNET_API_ERROR_NO_SUCH_ENTRY); + + ip_neighbor_free (ipn); + + return (0); +} + +void +ip_neighbor_update (vnet_main_t * vnm, adj_index_t ai) +{ + ip_neighbor_t *ipn; + ip_adjacency_t *adj; + + adj = adj_get (ai); + + ip_neighbor_key_t key = { + .ipnk_ip = adj->sub_type.nbr.next_hop, + .ipnk_type = fib_proto_to_ip46 (adj->ia_nh_proto), + .ipnk_sw_if_index = adj->rewrite_header.sw_if_index, + }; + ipn = ip_neighbor_db_find (&key); + + switch (adj->lookup_next_index) + { + case IP_LOOKUP_NEXT_ARP: + if (NULL != ipn) + { + adj_nbr_walk_nh (adj->rewrite_header.sw_if_index, + adj->ia_nh_proto, + &ipn->ipn_key->ipnk_ip, + ip_neighbor_mk_complete_walk, ipn); + } + else + { + /* + * no matching ARP entry. + * construct the rewrite required to for an ARP packet, and stick + * that in the adj's pipe to smoke. + */ + adj_nbr_update_rewrite + (ai, + ADJ_NBR_REWRITE_FLAG_INCOMPLETE, + ethernet_build_rewrite + (vnm, + adj->rewrite_header.sw_if_index, + VNET_LINK_ARP, + VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST)); + + /* + * since the FIB has added this adj for a route, it makes sense it + * may want to forward traffic sometime soon. Let's send a + * speculative ARP. just one. If we were to do periodically that + * wouldn't be bad either, but that's more code than i'm prepared to + * write at this time for relatively little reward. + */ + ip_neighbor_probe (adj); + } + break; + case IP_LOOKUP_NEXT_GLEAN: + case IP_LOOKUP_NEXT_BCAST: + case IP_LOOKUP_NEXT_MCAST: + case IP_LOOKUP_NEXT_DROP: + case IP_LOOKUP_NEXT_PUNT: + case IP_LOOKUP_NEXT_LOCAL: + case IP_LOOKUP_NEXT_REWRITE: + case IP_LOOKUP_NEXT_MCAST_MIDCHAIN: + case IP_LOOKUP_NEXT_MIDCHAIN: + case IP_LOOKUP_NEXT_ICMP_ERROR: + case IP_LOOKUP_N_NEXT: + ASSERT (0); + break; + } +} + +void +ip_neighbor_learn (const ip_neighbor_learn_t * l) +{ + ip_neighbor_add (&l->ip, l->type, &l->mac, l->sw_if_index, + IP_NEIGHBOR_FLAG_DYNAMIC, NULL); +} + +static clib_error_t * +ip_neighbor_cmd (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) +{ + ip46_address_t ip = ip46_address_initializer; + mac_address_t mac = ZERO_MAC_ADDRESS; + vnet_main_t *vnm = vnet_get_main (); + ip_neighbor_flags_t flags; + u32 sw_if_index = ~0; + int is_add = 1; + int count = 1; + + flags = IP_NEIGHBOR_FLAG_DYNAMIC; + + while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT) + { + /* set ip arp TenGigE1/1/0/1 1.2.3.4 aa:bb:... or aabb.ccdd... */ + if (unformat (input, "%U %U %U", + unformat_vnet_sw_interface, vnm, &sw_if_index, + unformat_ip46_address, &ip, IP46_TYPE_ANY, + unformat_mac_address_t, &mac)) + ; + else if (unformat (input, "delete") || unformat (input, "del")) + is_add = 0; + else if (unformat (input, "static")) + { + flags |= IP_NEIGHBOR_FLAG_STATIC; + flags &= ~IP_NEIGHBOR_FLAG_DYNAMIC; + } + else if (unformat (input, "no-fib-entry")) + flags |= IP_NEIGHBOR_FLAG_NO_FIB_ENTRY; + else if (unformat (input, "count %d", &count)) + ; + else + break; + } + + if (sw_if_index == ~0 || + ip46_address_is_zero (&ip) || mac_address_is_zero (&mac)) + return clib_error_return (0, + "specify interface, IP address and MAC: `%U'", + format_unformat_error, input); + + while (count) + { + if (is_add) + ip_neighbor_add (&ip, ip46_address_get_type (&ip), &mac, sw_if_index, + flags, NULL); + else + ip_neighbor_del (&ip, ip46_address_get_type (&ip), sw_if_index); + + ip46_address_increment (ip46_address_get_type (&ip), &ip); + mac_address_increment (&mac); + + --count; + } + + return NULL; +} + +/* *INDENT-OFF* */ +/*? + * Add or delete IPv4 ARP cache entries. + * + * @note 'set ip neighbor' options (e.g. delete, static, 'fib-id <id>', + * 'count <number>', 'interface ip4_addr mac_addr') can be added in + * any order and combination. + * + * @cliexpar + * @parblock + * Add or delete IPv4 ARP cache entries as follows. MAC Address can be in + * either aa:bb:cc:dd:ee:ff format or aabb.ccdd.eeff format. + * @cliexcmd{set ip neighbor GigabitEthernet2/0/0 6.0.0.3 dead.beef.babe} + * @cliexcmd{set ip neighbor delete GigabitEthernet2/0/0 6.0.0.3 de:ad:be:ef:ba:be} + * + * To add or delete an IPv4 ARP cache entry to or from a specific fib + * table: + * @cliexcmd{set ip neighbor fib-id 1 GigabitEthernet2/0/0 6.0.0.3 dead.beef.babe} + * @cliexcmd{set ip neighbor fib-id 1 delete GigabitEthernet2/0/0 6.0.0.3 dead.beef.babe} + * + * Add or delete IPv4 static ARP cache entries as follows: + * @cliexcmd{set ip neighbor static GigabitEthernet2/0/0 6.0.0.3 dead.beef.babe} + * @cliexcmd{set ip neighbor static delete GigabitEthernet2/0/0 6.0.0.3 dead.beef.babe} + * + * For testing / debugging purposes, the 'set ip neighbor' command can add or + * delete multiple entries. Supply the 'count N' parameter: + * @cliexcmd{set ip neighbor count 10 GigabitEthernet2/0/0 6.0.0.3 dead.beef.babe} + * @endparblock + ?*/ +VLIB_CLI_COMMAND (ip_neighbor_command, static) = { + .path = "set ip neighbor", + .short_help = + "set ip neighbor [del] <intfc> <ip-address> <mac-address> [static] [no-fib-entry] [count <count>] [fib-id <fib-id>] [proxy <lo-addr> - <hi-addr>]", + .function = ip_neighbor_cmd, +}; +VLIB_CLI_COMMAND (ip_neighbor_command2, static) = { + .path = "ip neighbor", + .short_help = + "ip neighbor [del] <intfc> <ip-address> <mac-address> [static] [no-fib-entry] [count <count>] [fib-id <fib-id>] [proxy <lo-addr> - <hi-addr>]", + .function = ip_neighbor_cmd, +}; +/* *INDENT-ON* */ + +static int +ip_neighbor_sort (void *a1, void *a2) +{ + index_t *ipni1 = a1, *ipni2 = a2; + ip_neighbor_t *ipn1, *ipn2; + int cmp; + + ipn1 = ip_neighbor_get (*ipni1); + ipn2 = ip_neighbor_get (*ipni2); + + cmp = vnet_sw_interface_compare (vnet_get_main (), + ipn1->ipn_key->ipnk_sw_if_index, + ipn2->ipn_key->ipnk_sw_if_index); + if (!cmp) + cmp = ip46_address_cmp (&ipn1->ipn_key->ipnk_ip, &ipn2->ipn_key->ipnk_ip); + return cmp; +} + +static index_t * +ip_neighbor_entries (u32 sw_if_index, ip46_type_t type) +{ + index_t *ipnis = NULL; + ip_neighbor_t *ipn; + + /* *INDENT-OFF* */ + pool_foreach (ipn, ip_neighbor_pool, + ({ + if (sw_if_index != ~0 && + ipn->ipn_key->ipnk_sw_if_index != sw_if_index && + (IP46_TYPE_ANY == type || + (ipn->ipn_key->ipnk_type == type))) + continue; + vec_add1 (ipnis, ip_neighbor_get_index(ipn)); + })); + + /* *INDENT-ON* */ + + if (ipnis) + vec_sort_with_function (ipnis, ip_neighbor_sort); + return ipnis; +} + +static clib_error_t * +ip_neighbor_show_sorted_i (vlib_main_t * vm, + unformat_input_t * input, + vlib_cli_command_t * cmd, ip46_type_t type) +{ + ip_neighbor_elt_t *elt, *head; + + head = pool_elt_at_index (ip_neighbor_elt_pool, + ip_neighbor_list_head[type]); + + + vlib_cli_output (vm, "%=12s%=40s%=6s%=20s%=24s", "Time", "IP", + "Flags", "Ethernet", "Interface"); + + /* *INDENT-OFF*/ + /* the list is time sorted, newest first, so start from the back + * and work forwards. Stop when we get to one that is alive */ + clib_llist_foreach_reverse(ip_neighbor_elt_pool, + ipne_anchor, head, elt, + ({ + vlib_cli_output (vm, "%U", format_ip_neighbor, elt->ipne_index); + })); + /* *INDENT-ON*/ + + return (NULL); +} + +static clib_error_t * +ip_neighbor_show_i (vlib_main_t * vm, + unformat_input_t * input, + vlib_cli_command_t * cmd, ip46_type_t type) +{ + index_t *ipni, *ipnis = NULL; + u32 sw_if_index; + + /* Filter entries by interface if given. */ + sw_if_index = ~0; + (void) unformat_user (input, unformat_vnet_sw_interface, vnet_get_main (), + &sw_if_index); + + ipnis = ip_neighbor_entries (sw_if_index, type); + + if (ipnis) + vlib_cli_output (vm, "%=12s%=40s%=6s%=20s%=24s", "Time", "IP", + "Flags", "Ethernet", "Interface"); + + vec_foreach (ipni, ipnis) + { + vlib_cli_output (vm, "%U", format_ip_neighbor, *ipni); + } + vec_free (ipnis); + + return (NULL); +} + +static clib_error_t * +ip_neighbor_show (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) +{ + return (ip_neighbor_show_i (vm, input, cmd, IP46_TYPE_ANY)); +} + +static clib_error_t * +ip6_neighbor_show (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) +{ + return (ip_neighbor_show_i (vm, input, cmd, IP46_TYPE_IP6)); +} + +static clib_error_t * +ip4_neighbor_show (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) +{ + return (ip_neighbor_show_i (vm, input, cmd, IP46_TYPE_IP4)); +} + +static clib_error_t * +ip6_neighbor_show_sorted (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) +{ + return (ip_neighbor_show_sorted_i (vm, input, cmd, IP46_TYPE_IP6)); +} + +static clib_error_t * +ip4_neighbor_show_sorted (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) +{ + return (ip_neighbor_show_sorted_i (vm, input, cmd, IP46_TYPE_IP4)); +} + +/*? + * Display all the IP neighbor entries. + * + * @cliexpar + * Example of how to display the IPv4 ARP table: + * @cliexstart{show ip neighbor} + * Time FIB IP4 Flags Ethernet Interface + * 346.3028 0 6.1.1.3 de:ad:be:ef:ba:be GigabitEthernet2/0/0 + * 3077.4271 0 6.1.1.4 S de:ad:be:ef:ff:ff GigabitEthernet2/0/0 + * 2998.6409 1 6.2.2.3 de:ad:be:ef:00:01 GigabitEthernet2/0/0 + * Proxy arps enabled for: + * Fib_index 0 6.0.0.1 - 6.0.0.11 + * @cliexend + ?*/ +/* *INDENT-OFF* */ +VLIB_CLI_COMMAND (show_ip_neighbors_cmd_node, static) = { + .path = "show ip neighbors", + .function = ip_neighbor_show, + .short_help = "show ip neighbors [interface]", +}; +VLIB_CLI_COMMAND (show_ip4_neighbors_cmd_node, static) = { + .path = "show ip4 neighbors", + .function = ip4_neighbor_show, + .short_help = "show ip4 neighbors [interface]", +}; +VLIB_CLI_COMMAND (show_ip6_neighbors_cmd_node, static) = { + .path = "show ip6 neighbors", + .function = ip6_neighbor_show, + .short_help = "show ip6 neighbors [interface]", +}; +VLIB_CLI_COMMAND (show_ip_neighbor_cmd_node, static) = { + .path = "show ip neighbor", + .function = ip_neighbor_show, + .short_help = "show ip neighbor [interface]", +}; +VLIB_CLI_COMMAND (show_ip4_neighbor_cmd_node, static) = { + .path = "show ip4 neighbor", + .function = ip4_neighbor_show, + .short_help = "show ip4 neighbor [interface]", +}; +VLIB_CLI_COMMAND (show_ip6_neighbor_cmd_node, static) = { + .path = "show ip6 neighbor", + .function = ip6_neighbor_show, + .short_help = "show ip6 neighbor [interface]", +}; +VLIB_CLI_COMMAND (show_ip4_neighbor_sorted_cmd_node, static) = { + .path = "show ip4 neighbor-sorted", + .function = ip4_neighbor_show_sorted, + .short_help = "show ip4 neighbor-sorted", +}; +VLIB_CLI_COMMAND (show_ip6_neighbor_sorted_cmd_node, static) = { + .path = "show ip6 neighbor-sorted", + .function = ip6_neighbor_show_sorted, + .short_help = "show ip6 neighbor-sorted", +}; +/* *INDENT-ON* */ + +static ip_neighbor_vft_t ip_nbr_vfts[IP46_N_TYPES]; + +void +ip_neighbor_register (ip46_type_t type, const ip_neighbor_vft_t * vft) +{ + ip_nbr_vfts[type] = *vft; +} + +void +ip_neighbor_probe_dst (const ip_adjacency_t * adj, const ip46_address_t * dst) +{ + if (!vnet_sw_interface_is_admin_up (vnet_get_main (), + adj->rewrite_header.sw_if_index)) + return; + + switch (adj->ia_nh_proto) + { + case FIB_PROTOCOL_IP6: + ip6_neighbor_probe_dst (adj, &dst->ip6); + break; + case FIB_PROTOCOL_IP4: + ip4_neighbor_probe_dst (adj, &dst->ip4); + break; + case FIB_PROTOCOL_MPLS: + ASSERT (0); + break; + } +} + +void +ip_neighbor_probe (const ip_adjacency_t * adj) +{ + ip_neighbor_probe_dst (adj, &adj->sub_type.nbr.next_hop); +} + +void +ip_neighbor_advertise (vlib_main_t * vm, + ip46_type_t type, + const ip46_address_t * addr, u32 sw_if_index) +{ + vnet_main_t *vnm = vnet_get_main (); + + if (type == IP46_TYPE_IP4 || type == IP46_TYPE_BOTH) + ip4_neighbor_advertise (vm, vnm, sw_if_index, &addr->ip4); + if (type == IP46_TYPE_IP6 || type == IP46_TYPE_BOTH) + ip6_neighbor_advertise (vm, vnm, sw_if_index, &addr->ip6); +} + +void +ip_neighbor_walk (ip46_type_t type, + u32 sw_if_index, ip_neighbor_walk_cb_t cb, void *ctx) +{ + ip_neighbor_key_t *key; + index_t ipni; + + if (~0 == sw_if_index) + { + uword **hash; + + vec_foreach (hash, ip_neighbor_db[type].ipndb_hash) + { + /* *INDENT-OFF* */ + hash_foreach (key, ipni, *hash, + ({ + cb (ipni, ctx); + })); + /* *INDENT-ON* */ + } + } + else + { + uword *hash; + + if (vec_len (ip_neighbor_db[type].ipndb_hash) <= sw_if_index) + return; + hash = ip_neighbor_db[type].ipndb_hash[sw_if_index]; + + /* *INDENT-OFF* */ + hash_foreach (key, ipni, hash, + ({ + cb (ipni, ctx); + })); + /* *INDENT-ON* */ + } +} + +int +ip4_neighbor_proxy_add (u32 fib_index, + const ip4_address_t * start, + const ip4_address_t * end) +{ + if (ip_nbr_vfts[IP46_TYPE_IP4].inv_proxy4_add) + { + return (ip_nbr_vfts[IP46_TYPE_IP4].inv_proxy4_add + (fib_index, start, end)); + } + + return (-1); +} + +int +ip4_neighbor_proxy_delete (u32 fib_index, + const ip4_address_t * start, + const ip4_address_t * end) +{ + if (ip_nbr_vfts[IP46_TYPE_IP4].inv_proxy4_del) + { + return (ip_nbr_vfts[IP46_TYPE_IP4].inv_proxy4_del + (fib_index, start, end)); + } + return -1; +} + +int +ip4_neighbor_proxy_enable (u32 sw_if_index) +{ + if (ip_nbr_vfts[IP46_TYPE_IP4].inv_proxy4_enable) + { + return (ip_nbr_vfts[IP46_TYPE_IP4].inv_proxy4_enable (sw_if_index)); + } + return -1; +} + +int +ip4_neighbor_proxy_disable (u32 sw_if_index) +{ + if (ip_nbr_vfts[IP46_TYPE_IP4].inv_proxy4_disable) + { + return (ip_nbr_vfts[IP46_TYPE_IP4].inv_proxy4_disable (sw_if_index)); + } + return -1; +} + +int +ip6_neighbor_proxy_add (u32 sw_if_index, const ip6_address_t * addr) +{ + if (ip_nbr_vfts[IP46_TYPE_IP6].inv_proxy6_add) + { + return (ip_nbr_vfts[IP46_TYPE_IP6].inv_proxy6_add (sw_if_index, addr)); + } + return -1; +} + +int +ip6_neighbor_proxy_del (u32 sw_if_index, const ip6_address_t * addr) +{ + if (ip_nbr_vfts[IP46_TYPE_IP6].inv_proxy6_del) + { + return (ip_nbr_vfts[IP46_TYPE_IP6].inv_proxy6_del (sw_if_index, addr)); + } + return -1; +} + +static void +ip_neighbor_ethernet_change_mac (ethernet_main_t * em, + u32 sw_if_index, uword opaque) +{ + ip_neighbor_t *ipn; + adj_index_t ai; + + IP_NEIGHBOR_DBG ("mac-change: %U", + format_vnet_sw_if_index_name, vnet_get_main (), + sw_if_index); + + /* *INDENT-OFF* */ + pool_foreach (ipn, ip_neighbor_pool, + ({ + if (ipn->ipn_key->ipnk_sw_if_index == sw_if_index) + adj_nbr_walk_nh (ipn->ipn_key->ipnk_sw_if_index, + fib_proto_from_ip46(ipn->ipn_key->ipnk_type), + &ipn->ipn_key->ipnk_ip, + ip_neighbor_mk_complete_walk, + ipn); + })); + /* *INDENT-ON* */ + + ai = adj_glean_get (FIB_PROTOCOL_IP4, sw_if_index); + + if (ADJ_INDEX_INVALID != ai) + adj_glean_update_rewrite (ai); +} + +void +ip_neighbor_populate (ip46_type_t type, u32 sw_if_index) +{ + index_t *ipnis = NULL, *ipni; + ip_neighbor_t *ipn; + + IP_NEIGHBOR_DBG ("populate: %U %U", + format_vnet_sw_if_index_name, vnet_get_main (), + sw_if_index, format_ip46_type, type); + + /* *INDENT-OFF* */ + pool_foreach (ipn, ip_neighbor_pool, + ({ + if (ipn->ipn_key->ipnk_type == type && + ipn->ipn_key->ipnk_sw_if_index == sw_if_index) + vec_add1 (ipnis, ipn - ip_neighbor_pool); + })); + /* *INDENT-ON* */ + + vec_foreach (ipni, ipnis) + { + ipn = ip_neighbor_get (*ipni); + + adj_nbr_walk_nh (ipn->ipn_key->ipnk_sw_if_index, + fib_proto_from_ip46 (ipn->ipn_key->ipnk_type), + &ipn->ipn_key->ipnk_ip, + ip_neighbor_mk_complete_walk, ipn); + } + vec_free (ipnis); +} + +void +ip_neighbor_flush (ip46_type_t type, u32 sw_if_index) +{ + index_t *ipnis = NULL, *ipni; + ip_neighbor_t *ipn; + + IP_NEIGHBOR_DBG ("flush: %U %U", + format_vnet_sw_if_index_name, vnet_get_main (), + sw_if_index, format_ip46_type, type); + + /* *INDENT-OFF* */ + pool_foreach (ipn, ip_neighbor_pool, + ({ + if (ipn->ipn_key->ipnk_type == type && + ipn->ipn_key->ipnk_sw_if_index == sw_if_index && + ip_neighbor_is_dynamic (ipn)) + vec_add1 (ipnis, ipn - ip_neighbor_pool); + })); + /* *INDENT-ON* */ + + vec_foreach (ipni, ipnis) ip_neighbor_free (ip_neighbor_get (*ipni)); + vec_free (ipnis); +} + +/* + * Remove any arp entries associated with the specified interface + */ +static clib_error_t * +ip_neighbor_interface_admin_change (vnet_main_t * vnm, + u32 sw_if_index, u32 flags) +{ + ip46_type_t type; + + IP_NEIGHBOR_DBG ("interface-admin: %U %s", + format_vnet_sw_if_index_name, vnet_get_main (), + sw_if_index, + (flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP ? "up" : "down")); + + if (flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP) + { + FOREACH_IP46_TYPE (type) ip_neighbor_populate (type, sw_if_index); + } + else + { + /* admin down, flush all neighbours */ + FOREACH_IP46_TYPE (type) ip_neighbor_flush (type, sw_if_index); + } + + return (NULL); +} + +VNET_SW_INTERFACE_ADMIN_UP_DOWN_FUNCTION (ip_neighbor_interface_admin_change); + +/* + * Remove any arp entries associated with the specified interface + */ +static clib_error_t * +ip_neighbor_delete_sw_interface (vnet_main_t * vnm, + u32 sw_if_index, u32 is_add) +{ + IP_NEIGHBOR_DBG ("interface-change: %U %s", + format_vnet_sw_if_index_name, vnet_get_main (), + sw_if_index, (is_add ? "add" : "del")); + + if (!is_add && sw_if_index != ~0) + { + ip46_type_t type; + + FOREACH_IP46_TYPE (type) ip_neighbor_flush (type, sw_if_index); + } + + return (NULL); +} + +VNET_SW_INTERFACE_ADD_DEL_FUNCTION (ip_neighbor_delete_sw_interface); + +typedef struct ip_neighbor_walk_covered_ctx_t_ +{ + ip46_type_t type; + ip46_address_t addr; + u32 length; + index_t *ipnis; +} ip_neighbor_walk_covered_ctx_t; + +static walk_rc_t +ip_neighbor_walk_covered (index_t ipni, void *arg) +{ + ip_neighbor_walk_covered_ctx_t *ctx = arg; + ip_neighbor_t *ipn; + + ipn = ip_neighbor_get (ipni); + + ASSERT (ipn->ipn_key->ipnk_type == ctx->type); + + if (IP46_TYPE_IP4 == ctx->type) + { + if (ip4_destination_matches_route (&ip4_main, + &ipn->ipn_key->ipnk_ip.ip4, + &ctx->addr.ip4, + ctx->length) && + ip_neighbor_is_dynamic (ipn)) + { + vec_add1 (ctx->ipnis, ip_neighbor_get_index (ipn)); + } + } + return (WALK_CONTINUE); +} + + +/* + * callback when an interface address is added or deleted + */ +static void +ip_neighbor_add_del_interface_address_v4 (ip4_main_t * im, + uword opaque, + u32 sw_if_index, + ip4_address_t * address, + u32 address_length, + u32 if_address_index, u32 is_del) +{ + /* + * Flush the ARP cache of all entries covered by the address + * that is being removed. + */ + IP_NEIGHBOR_DBG ("addr-%d: %U, %U/%d", + (is_del ? "del" : "add"), + format_vnet_sw_if_index_name, vnet_get_main (), + sw_if_index, format_ip4_address, address, address_length); + + if (is_del) + { + ip_neighbor_walk_covered_ctx_t ctx = { + .addr.ip4 = *address, + .type = IP46_TYPE_IP4, + .length = address_length, + }; + index_t *ipni; + + ip_neighbor_walk (IP46_TYPE_IP4, sw_if_index, + ip_neighbor_walk_covered, &ctx); + + vec_foreach (ipni, ctx.ipnis) + ip_neighbor_free (ip_neighbor_get (*ipni)); + + vec_free (ctx.ipnis); + } +} + +/* + * callback when an interface address is added or deleted + */ +static void +ip_neighbor_add_del_interface_address_v6 (ip6_main_t * im, + uword opaque, + u32 sw_if_index, + ip6_address_t * address, + u32 address_length, + u32 if_address_index, u32 is_del) +{ + /* + * Flush the ARP cache of all entries covered by the address + * that is being removed. + */ + IP_NEIGHBOR_DBG ("addr-change: %U, %U/%d %s", + format_vnet_sw_if_index_name, vnet_get_main (), + sw_if_index, format_ip6_address, address, address_length, + (is_del ? "del" : "add")); + + if (is_del) + { + ip_neighbor_walk_covered_ctx_t ctx = { + .addr.ip6 = *address, + .type = IP46_TYPE_IP6, + .length = address_length, + }; + index_t *ipni; + + ip_neighbor_walk (IP46_TYPE_IP6, sw_if_index, + ip_neighbor_walk_covered, &ctx); + + vec_foreach (ipni, ctx.ipnis) + ip_neighbor_free (ip_neighbor_get (*ipni)); + + vec_free (ctx.ipnis); + } +} + +typedef struct ip_neighbor_table_bind_ctx_t_ +{ + u32 new_fib_index; + u32 old_fib_index; +} ip_neighbor_table_bind_ctx_t; + +static walk_rc_t +ip_neighbor_walk_table_bind (index_t ipni, void *arg) +{ + ip_neighbor_table_bind_ctx_t *ctx = arg; + ip_neighbor_t *ipn; + + ipn = ip_neighbor_get (ipni); + ip_neighbor_adj_fib_remove (ipn, ctx->old_fib_index); + ip_neighbor_adj_fib_add (ipn, ctx->new_fib_index); + + return (WALK_CONTINUE); +} + +static void +ip_neighbor_table_bind_v4 (ip4_main_t * im, + uword opaque, + u32 sw_if_index, + u32 new_fib_index, u32 old_fib_index) +{ + ip_neighbor_table_bind_ctx_t ctx = { + .old_fib_index = old_fib_index, + .new_fib_index = new_fib_index, + }; + + ip_neighbor_walk (IP46_TYPE_IP4, sw_if_index, + ip_neighbor_walk_table_bind, &ctx); +} + +static void +ip_neighbor_table_bind_v6 (ip6_main_t * im, + uword opaque, + u32 sw_if_index, + u32 new_fib_index, u32 old_fib_index) +{ + ip_neighbor_table_bind_ctx_t ctx = { + .old_fib_index = old_fib_index, + .new_fib_index = new_fib_index, + }; + + ip_neighbor_walk (IP46_TYPE_IP6, sw_if_index, + ip_neighbor_walk_table_bind, &ctx); +} + +typedef enum ip_neighbor_age_state_t_ +{ + IP_NEIGHBOR_AGE_ALIVE, + IP_NEIGHBOR_AGE_PROBE, + IP_NEIGHBOR_AGE_DEAD, +} ip_neighbor_age_state_t; + +#define IP_NEIGHBOR_PROCESS_SLEEP_LONG (0) + +static ip_neighbor_age_state_t +ip_neighbour_age_out (index_t ipni, f64 now, f64 * wait) +{ + ip_neighbor_t *ipn; + f64 ttl; + + ipn = ip_neighbor_get (ipni); + ttl = now - ipn->ipn_time_last_updated; + *wait = IP_NEIGHBOR_PROCESS_SLEEP_LONG; + + if (ttl > ip_neighbor_db[ipn->ipn_key->ipnk_type].ipndb_age) + { + IP_NEIGHBOR_DBG ("aged: %U @%f - %f > %d", + format_ip_neighbor, ipni, now, + ipn->ipn_time_last_updated, + ip_neighbor_db[ipn->ipn_key->ipnk_type].ipndb_age); + if (ipn->ipn_n_probes > 2) + { + /* 3 strikes and yea-re out */ + IP_NEIGHBOR_DBG ("dead: %U", format_ip_neighbor, ipni); + return (IP_NEIGHBOR_AGE_DEAD); + } + else + { + adj_index_t ai; + + ai = adj_glean_get (fib_proto_from_ip46 (ipn->ipn_key->ipnk_type), + ip_neighbor_get_sw_if_index (ipn)); + + if (ADJ_INDEX_INVALID != ai) + ip_neighbor_probe_dst (adj_get (ai), ip_neighbor_get_ip (ipn)); + + ipn->ipn_n_probes++; + *wait = 1; + } + } + else + { + *wait = ttl; + return (IP_NEIGHBOR_AGE_ALIVE); + } + + return (IP_NEIGHBOR_AGE_PROBE); +} + +typedef enum ip_neighbor_process_event_t_ +{ + IP_NEIGHBOR_AGE_PROCESS_WAKEUP, +} ip_neighbor_process_event_t; + +static uword +ip_neighbor_age_loop (vlib_main_t * vm, + vlib_node_runtime_t * rt, + vlib_frame_t * f, ip46_type_t type) +{ + uword event_type, *event_data = NULL; + f64 timeout; + + /* Set the timeout to an effectively infinite value when the process starts */ + timeout = IP_NEIGHBOR_PROCESS_SLEEP_LONG; + + while (1) + { + f64 now; + + if (!timeout) + vlib_process_wait_for_event (vm); + else + vlib_process_wait_for_event_or_clock (vm, timeout); + + event_type = vlib_process_get_events (vm, &event_data); + vec_reset_length (event_data); + + now = vlib_time_now (vm); + + switch (event_type) + { + case ~0: + { + /* timer expired */ + ip_neighbor_elt_t *elt, *head; + f64 wait; + + timeout = 1e5; + head = pool_elt_at_index (ip_neighbor_elt_pool, + ip_neighbor_list_head[type]); + + /* *INDENT-OFF*/ + /* the list is time sorted, newest first, so start from the back + * and work forwards. Stop when we get to one that is alive */ + restart: + clib_llist_foreach_reverse(ip_neighbor_elt_pool, + ipne_anchor, head, elt, + ({ + ip_neighbor_age_state_t res; + + res = ip_neighbour_age_out(elt->ipne_index, now, &wait); + + if (IP_NEIGHBOR_AGE_ALIVE == res) { + /* the oldest neighbor has not yet expired, go back to sleep */ + break; + } + else if (IP_NEIGHBOR_AGE_DEAD == res) { + /* the oldest neighbor is dead, pop it, then restart the walk + * again from the back */ + ip_neighbor_free (ip_neighbor_get(elt->ipne_index)); + goto restart; + } + + timeout = clib_min (wait, timeout); + })); + /* *INDENT-ON* */ + break; + } + case IP_NEIGHBOR_AGE_PROCESS_WAKEUP: + { + + if (!ip_neighbor_db[type].ipndb_age) + { + /* aging has been disabled */ + timeout = 0; + break; + } + ip_neighbor_elt_t *elt, *head; + + head = pool_elt_at_index (ip_neighbor_elt_pool, + ip_neighbor_list_head[type]); + elt = clib_llist_prev (ip_neighbor_elt_pool, ipne_anchor, head); + + /* poke the oldset neighbour for aging, which returns how long we sleep for */ + if (IP_NEIGHBOR_AGE_PROBE == + ip_neighbour_age_out (elt->ipne_index, now, &timeout)) + /* we probed for the oldest entry, sleep for a short time to get to the next */ + timeout = 0.01; + break; + } + } + } + return 0; +} + +static uword +ip4_neighbor_age_process (vlib_main_t * vm, + vlib_node_runtime_t * rt, vlib_frame_t * f) +{ + return (ip_neighbor_age_loop (vm, rt, f, IP46_TYPE_IP4)); +} + +static uword +ip6_neighbor_age_process (vlib_main_t * vm, + vlib_node_runtime_t * rt, vlib_frame_t * f) +{ + return (ip_neighbor_age_loop (vm, rt, f, IP46_TYPE_IP6)); +} + +/* *INDENT-OFF* */ +VLIB_REGISTER_NODE (ip4_neighbor_age_process_node,static) = { + .function = ip4_neighbor_age_process, + .type = VLIB_NODE_TYPE_PROCESS, + .name = "ip4-neighbor-age-process", +}; +VLIB_REGISTER_NODE (ip6_neighbor_age_process_node,static) = { + .function = ip6_neighbor_age_process, + .type = VLIB_NODE_TYPE_PROCESS, + .name = "ip6-neighbor-age-process", +}; +/* *INDENT-ON* */ + +int +ip_neighbor_config (ip46_type_t type, u32 limit, u32 age, bool recycle) +{ + ip_neighbor_db[type].ipndb_limit = limit; + ip_neighbor_db[type].ipndb_recycle = recycle; + ip_neighbor_db[type].ipndb_age = age; + + vlib_process_signal_event (vlib_get_main (), + (IP46_TYPE_IP4 == type ? + ip4_neighbor_age_process_node.index : + ip6_neighbor_age_process_node.index), + IP_NEIGHBOR_AGE_PROCESS_WAKEUP, 0); + + return (0); +} + +static clib_error_t * +ip_neighbor_config_show (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) +{ + ip46_type_t type; + + /* *INDENT-OFF* */ + FOREACH_IP46_TYPE(type) { + vlib_cli_output (vm, "%U:", format_ip46_type, type); + vlib_cli_output (vm, " limit:%d, age:%d, recycle:%d", + ip_neighbor_db[type].ipndb_limit, + ip_neighbor_db[type].ipndb_age, + ip_neighbor_db[type].ipndb_recycle); + } + + /* *INDENT-ON* */ + return (NULL); +} + +/* *INDENT-OFF* */ +VLIB_CLI_COMMAND (show_ip_neighbor_cfg_cmd_node, static) = { + .path = "show ip neighbor-config", + .function = ip_neighbor_config_show, + .short_help = "show ip neighbor-config", +}; +/* *INDENT-ON* */ + +static clib_error_t * +ip_neighbor_init (vlib_main_t * vm) +{ + { + ip4_add_del_interface_address_callback_t cb = { + .function = ip_neighbor_add_del_interface_address_v4, + }; + vec_add1 (ip4_main.add_del_interface_address_callbacks, cb); + } + { + ip6_add_del_interface_address_callback_t cb = { + .function = ip_neighbor_add_del_interface_address_v6, + }; + vec_add1 (ip6_main.add_del_interface_address_callbacks, cb); + } + { + ip4_table_bind_callback_t cb = { + .function = ip_neighbor_table_bind_v4, + }; + vec_add1 (ip4_main.table_bind_callbacks, cb); + } + { + ip6_table_bind_callback_t cb = { + .function = ip_neighbor_table_bind_v6, + }; + vec_add1 (ip6_main.table_bind_callbacks, cb); + } + { + ethernet_address_change_ctx_t ctx = { + .function = ip_neighbor_ethernet_change_mac, + .function_opaque = 0, + }; + vec_add1 (ethernet_main.address_change_callbacks, ctx); + } + + ipn_logger = vlib_log_register_class ("ip", "neighbor"); + + ip46_type_t type; + + FOREACH_IP46_TYPE (type) + ip_neighbor_list_head[type] = + clib_llist_make_head (ip_neighbor_elt_pool, ipne_anchor); + + return (NULL); +} + +/* *INDENT-OFF* */ +VLIB_INIT_FUNCTION (ip_neighbor_init) = +{ + .runs_after = VLIB_INITS("ip_main_init"), +}; +/* *INDENT-ON* */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip-neighbor/ip_neighbor.h b/src/vnet/ip-neighbor/ip_neighbor.h new file mode 100644 index 00000000000..cb384c5e240 --- /dev/null +++ b/src/vnet/ip-neighbor/ip_neighbor.h @@ -0,0 +1,124 @@ +/* + * ip_neighboor.h: ip neighbor generic services + * + * Copyright (c) 2018 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef __INCLUDE_IP_NEIGHBOR_H__ +#define __INCLUDE_IP_NEIGHBOR_H__ + +#include <vnet/ip-neighbor/ip_neighbor_types.h> + +#include <vnet/adj/adj.h> + +void ip_neighbor_scan_enable_disable (ip_neighbor_scan_arg_t * arg); + + +/***** + * APIs external modules can invoke on the neighbor subsystem + */ + +extern ip_neighbor_t *ip_neighbor_get (index_t ipni); +extern int ip_neighbor_add (const ip46_address_t * ip, + ip46_type_t type, + const mac_address_t * mac, + u32 sw_if_index, + ip_neighbor_flags_t flags, u32 * stats_index); +extern int ip_neighbor_del (const ip46_address_t * ip, + ip46_type_t type, u32 sw_if_index); + +extern int ip_neighbor_config (ip46_type_t type, u32 limit, u32 age, + bool recycle); + +typedef walk_rc_t (*ip_neighbor_walk_cb_t) (index_t ipni, void *ctx); +extern void ip_neighbor_walk (ip46_type_t type, + u32 sw_if_index, + ip_neighbor_walk_cb_t fn, void *ctx); + +extern const ip46_address_t *ip_neighbor_get_ip (const ip_neighbor_t * ipn); +extern const mac_address_t *ip_neighbor_get_mac (const ip_neighbor_t * ipn); +extern const u32 ip_neighbor_get_sw_if_index (const ip_neighbor_t * ipn); + +extern void ip_neighbor_learn (const ip_neighbor_learn_t * l); + +extern void ip_neighbor_update (vnet_main_t * vnm, adj_index_t ai); + +extern void ip_neighbor_advertise (vlib_main_t * vm, + ip46_type_t tyoe, + const ip46_address_t * addr, + u32 sw_if_index); +extern void ip_neighbor_probe (const ip_adjacency_t * adj); +extern void ip_neighbor_probe_dst (const ip_adjacency_t * adj, + const ip46_address_t * ip); + +/** + * From the watcher to the API to publish a new neighbor + */ +extern void ip_neighbor_handle_event (const ip_neighbor_event_t * ipne); + +/** + * The set of function that vnet requires from the IP neighbour module. + * Note that an implementation of these functions will not exist + * if the ip-neighbour plugin is not loaded. so check the error codes! + */ +extern int ip4_neighbor_proxy_add (u32 fib_index, + const ip4_address_t * start, + const ip4_address_t * end); +extern int ip4_neighbor_proxy_delete (u32 fib_index, + const ip4_address_t * start, + const ip4_address_t * end); +extern int ip4_neighbor_proxy_enable (u32 sw_if_index); +extern int ip4_neighbor_proxy_disable (u32 sw_if_index); +extern int ip6_neighbor_proxy_add (u32 sw_if_index, + const ip6_address_t * addr); +extern int ip6_neighbor_proxy_del (u32 sw_if_index, + const ip6_address_t * addr); + +/** + * neighbor protocol implementation registration functions + * this are provided by ARP and IP-ND + */ +typedef int (*ip4_neighbor_proxy_addr_t) (u32 fib_index, + const ip4_address_t * start, + const ip4_address_t * end); +typedef int (*ip4_neighbor_proxy_cfg_t) (u32 sw_if_index); +typedef int (*ip6_neighbor_proxy_cfg_t) (u32 sw_if_index, + const ip6_address_t * addr); + +/** + * Virtual function Table for neighbor protocol implementations to register + */ +typedef struct ip_neighbor_vft_t_ +{ + ip4_neighbor_proxy_cfg_t inv_proxy4_enable; + ip4_neighbor_proxy_cfg_t inv_proxy4_disable; + ip4_neighbor_proxy_addr_t inv_proxy4_add; + ip4_neighbor_proxy_addr_t inv_proxy4_del; + ip6_neighbor_proxy_cfg_t inv_proxy6_add; + ip6_neighbor_proxy_cfg_t inv_proxy6_del; +} ip_neighbor_vft_t; + +extern void ip_neighbor_register (ip46_type_t type, + const ip_neighbor_vft_t * vft); + + +#endif /* __INCLUDE_IP_NEIGHBOR_H__ */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip-neighbor/ip_neighbor_api.c b/src/vnet/ip-neighbor/ip_neighbor_api.c new file mode 100644 index 00000000000..b066423afe5 --- /dev/null +++ b/src/vnet/ip-neighbor/ip_neighbor_api.c @@ -0,0 +1,298 @@ +/* + *------------------------------------------------------------------ + * ip_api.c - vnet ip api + * + * Copyright (c) 2016 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 <stddef.h> + +#include <vnet/ip-neighbor/ip_neighbor.h> +#include <vnet/ip-neighbor/ip_neighbor_watch.h> +#include <vnet/ip/ip_types_api.h> +#include <vnet/ethernet/ethernet_types_api.h> + +#include <vlibapi/api.h> +#include <vlibmemory/api.h> + +#include <vnet/ip-neighbor/ip_neighbor.api_enum.h> +#include <vnet/ip-neighbor/ip_neighbor.api_types.h> + +static u16 msg_id_base; +#define REPLY_MSG_ID_BASE msg_id_base + +#include <vlibapi/api_helper_macros.h> + +#include <vnet/format_fns.h> + + +static ip46_type_t +ip46_type_from_af (ip_address_family_t af) +{ + return (AF_IP4 == af ? IP46_TYPE_IP4 : IP46_TYPE_IP6); +} + +static vl_api_ip_neighbor_flags_t +ip_neighbor_flags_encode (ip_neighbor_flags_t f) +{ + vl_api_ip_neighbor_flags_t v = IP_API_NEIGHBOR_FLAG_NONE; + + if (f & IP_NEIGHBOR_FLAG_STATIC) + v |= IP_API_NEIGHBOR_FLAG_STATIC; + if (f & IP_NEIGHBOR_FLAG_NO_FIB_ENTRY) + v |= IP_API_NEIGHBOR_FLAG_NO_FIB_ENTRY; + + return (v); +} + +static void +ip_neighbor_encode (vl_api_ip_neighbor_t * api, const ip_neighbor_t * ipn) +{ + api->sw_if_index = htonl (ipn->ipn_key->ipnk_sw_if_index); + api->flags = ip_neighbor_flags_encode (ipn->ipn_flags); + + ip_address_encode (&ipn->ipn_key->ipnk_ip, + ipn->ipn_key->ipnk_type, &api->ip_address); + mac_address_encode (&ipn->ipn_mac, api->mac_address); +} + +void +ip_neighbor_handle_event (const ip_neighbor_event_t * ipne) +{ + vl_api_ip_neighbor_event_t *mp; + vl_api_registration_t *reg; + const ip_neighbor_t *ipn; + + ipn = ip_neighbor_get (ipne->ipne_index); + + if (NULL == ipn) + /* Client can cancel, die, etc. */ + return; + + /* Customer(s) requesting event for this neighbor */ + reg = vl_api_client_index_to_registration (ipne->ipne_watch.ipw_client); + if (!reg) + return; + + if (vl_api_can_send_msg (reg)) + { + mp = vl_msg_api_alloc (sizeof (*mp)); + clib_memset (mp, 0, sizeof (*mp)); + mp->_vl_msg_id = ntohs (VL_API_IP_NEIGHBOR_EVENT + REPLY_MSG_ID_BASE); + mp->client_index = ipne->ipne_watch.ipw_client; + mp->pid = ipne->ipne_watch.ipw_pid; + + ip_neighbor_encode (&mp->neighbor, ipn); + + vl_api_send_msg (reg, (u8 *) mp); + } + else + { + static f64 last_time; + /* + * Throttle syslog msgs. + * It's pretty tempting to just revoke the registration... + */ + if (vlib_time_now (vlib_get_main ()) > last_time + 10.0) + { + clib_warning ("ip6 nd event for %U to pid %d: queue stuffed!", + format_ip46_address, &ipn->ipn_key->ipnk_ip, + IP46_TYPE_ANY, ipne->ipne_watch.ipw_pid); + last_time = vlib_time_now (vlib_get_main ()); + } + } +} + +typedef struct ip_neighbor_dump_ctx_t_ +{ + vl_api_registration_t *reg; + u32 context; +} ip_neighbor_dump_ctx_t; + +static walk_rc_t +send_ip_neighbor_details (index_t ipni, void *arg) +{ + ip_neighbor_dump_ctx_t *ctx = arg; + vl_api_ip_neighbor_details_t *mp; + ip_neighbor_t *ipn; + + ipn = ip_neighbor_get (ipni); + mp = vl_msg_api_alloc (sizeof (*mp)); + clib_memset (mp, 0, sizeof (*mp)); + mp->_vl_msg_id = ntohs (VL_API_IP_NEIGHBOR_DETAILS + REPLY_MSG_ID_BASE); + mp->context = ctx->context; + ip_neighbor_encode (&mp->neighbor, ipn); + + vl_api_send_msg (ctx->reg, (u8 *) mp); + + return (WALK_CONTINUE); +} + +static void +vl_api_ip_neighbor_dump_t_handler (vl_api_ip_neighbor_dump_t * mp) +{ + vl_api_registration_t *reg; + ip_address_family_t af; + int rv; + + reg = vl_api_client_index_to_registration (mp->client_index); + if (!reg) + return; + + u32 sw_if_index = ntohl (mp->sw_if_index); + + rv = ip_address_family_decode (mp->af, &af); + + if (rv) + return; + + ip_neighbor_dump_ctx_t ctx = { + .reg = reg, + .context = mp->context, + }; + + // walk all neighbours on all interfaces + ip_neighbor_walk ((af == AF_IP4 ? + IP46_TYPE_IP4 : + IP46_TYPE_IP6), + sw_if_index, send_ip_neighbor_details, &ctx); +} + +static ip_neighbor_flags_t +ip_neighbor_flags_decode (vl_api_ip_neighbor_flags_t v) +{ + ip_neighbor_flags_t f = IP_NEIGHBOR_FLAG_NONE; + + if (v & IP_API_NEIGHBOR_FLAG_STATIC) + f |= IP_NEIGHBOR_FLAG_STATIC; + if (v & IP_API_NEIGHBOR_FLAG_NO_FIB_ENTRY) + f |= IP_NEIGHBOR_FLAG_NO_FIB_ENTRY; + + return (f); +} + +static void +vl_api_ip_neighbor_add_del_t_handler (vl_api_ip_neighbor_add_del_t * mp, + vlib_main_t * vm) +{ + vl_api_ip_neighbor_add_del_reply_t *rmp; + ip_neighbor_flags_t flags; + u32 stats_index = ~0; + ip46_address_t ip = ip46_address_initializer; + mac_address_t mac; + ip46_type_t type; + int rv; + + VALIDATE_SW_IF_INDEX ((&mp->neighbor)); + + flags = ip_neighbor_flags_decode (mp->neighbor.flags); + type = ip_address_decode (&mp->neighbor.ip_address, &ip); + mac_address_decode (mp->neighbor.mac_address, &mac); + + /* must be static or dynamic, default to dynamic */ + if (!(flags & IP_NEIGHBOR_FLAG_STATIC) && + !(flags & IP_NEIGHBOR_FLAG_DYNAMIC)) + flags |= IP_NEIGHBOR_FLAG_DYNAMIC; + + /* + * there's no validation here of the ND/ARP entry being added. + * The expectation is that the FIB will ensure that nothing bad + * will come of adding bogus entries. + */ + if (mp->is_add) + rv = ip_neighbor_add (&ip, type, &mac, + ntohl (mp->neighbor.sw_if_index), + flags, &stats_index); + else + rv = ip_neighbor_del (&ip, type, ntohl (mp->neighbor.sw_if_index)); + + BAD_SW_IF_INDEX_LABEL; + + /* *INDENT-OFF* */ + REPLY_MACRO2 (VL_API_IP_NEIGHBOR_ADD_DEL_REPLY, + ({ + rmp->stats_index = htonl (stats_index); + })); + /* *INDENT-ON* */ +} + +static void +vl_api_want_ip_neighbor_events_t_handler (vl_api_want_ip_neighbor_events_t * + mp) +{ + vl_api_want_ip_neighbor_events_reply_t *rmp; + ip46_address_t ip; + ip46_type_t itype; + int rv = 0; + + if (mp->sw_if_index != ~0) + VALIDATE_SW_IF_INDEX (mp); + itype = ip_address_decode (&mp->ip, &ip); + + ip_neighbor_watcher_t watch = { + .ipw_client = mp->client_index, + .ipw_pid = mp->pid, + }; + + if (mp->enable) + ip_neighbor_watch (&ip, itype, ntohl (mp->sw_if_index), &watch); + else + ip_neighbor_unwatch (&ip, itype, ntohl (mp->sw_if_index), &watch); + + BAD_SW_IF_INDEX_LABEL; + REPLY_MACRO (VL_API_WANT_IP_NEIGHBOR_EVENTS_REPLY); +} + +static void +vl_api_ip_neighbor_config_t_handler (vl_api_ip_neighbor_config_t * mp) +{ + vl_api_ip_neighbor_config_reply_t *rmp; + ip_address_family_t af; + int rv; + + rv = ip_address_family_decode (mp->af, &af); + + if (!rv) + rv = ip_neighbor_config (ip46_type_from_af (af), + ntohl (mp->max_number), + ntohl (mp->max_age), mp->recycle); + + REPLY_MACRO (VL_API_IP_NEIGHBOR_CONFIG_REPLY); +} + +#define vl_msg_name_crc_list +#include <vnet/ip-neighbor/ip_neighbor.api.h> +#undef vl_msg_name_crc_list + +#include <vnet/ip-neighbor/ip_neighbor.api.c> + +static clib_error_t * +ip_neighbor_api_init (vlib_main_t * vm) +{ + /* Ask for a correctly-sized block of API message decode slots */ + msg_id_base = setup_message_id_table (); + + return 0; +} + +VLIB_INIT_FUNCTION (ip_neighbor_api_init); + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip-neighbor/ip_neighbor_dp.c b/src/vnet/ip-neighbor/ip_neighbor_dp.c new file mode 100644 index 00000000000..6219a730af4 --- /dev/null +++ b/src/vnet/ip-neighbor/ip_neighbor_dp.c @@ -0,0 +1,39 @@ +/* + * ip_neighboor.h: ip neighbor generic services + * + * Copyright (c) 2018 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 <vlibmemory/api.h> + +#include <vnet/ip-neighbor/ip_neighbor_dp.h> +#include <vnet/ip-neighbor/ip_neighbor.h> + +/** + * APIs invoked by neighbor implementation (i.s. ARP and ND) that can be + * called from the DP when the protocol has resolved a neighbor + */ +void +ip_neighbor_learn_dp (const ip_neighbor_learn_t * l) +{ + vl_api_rpc_call_main_thread (ip_neighbor_learn, (u8 *) l, sizeof (*l)); +} + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip-neighbor/ip_neighbor_dp.h b/src/vnet/ip-neighbor/ip_neighbor_dp.h new file mode 100644 index 00000000000..b5664c2d195 --- /dev/null +++ b/src/vnet/ip-neighbor/ip_neighbor_dp.h @@ -0,0 +1,38 @@ +/* + * ip_neighboor.h: ip neighbor generic services + * + * Copyright (c) 2018 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef __INCLUDE_IP_NEIGHBOR_DP_H__ +#define __INCLUDE_IP_NEIGHBOR_DP_H__ + +#include <vnet/ip-neighbor/ip_neighbor_types.h> + +/** + * APIs invoked by neighbor implementation (i.s. ARP and ND) that can be + * called from the DP when the protocol has resolved a neighbor + */ + +extern void ip_neighbor_learn_dp (const ip_neighbor_learn_t * l); + +#endif /* __INCLUDE_IP_NEIGHBOR_H__ */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip-neighbor/ip_neighbor_types.c b/src/vnet/ip-neighbor/ip_neighbor_types.c new file mode 100644 index 00000000000..27262a5d62c --- /dev/null +++ b/src/vnet/ip-neighbor/ip_neighbor_types.c @@ -0,0 +1,83 @@ +/* + * ip_neighboor.h: ip neighbor generic services + * + * Copyright (c) 2018 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/ip-neighbor/ip_neighbor_types.h> + +u8 * +format_ip_neighbor_flags (u8 * s, va_list * args) +{ + ip_neighbor_flags_t flags = va_arg (*args, int); + + if (flags & IP_NEIGHBOR_FLAG_STATIC) + s = format (s, "S"); + + if (flags & IP_NEIGHBOR_FLAG_DYNAMIC) + s = format (s, "D"); + + if (flags & IP_NEIGHBOR_FLAG_NO_FIB_ENTRY) + s = format (s, "N"); + + return s; +} + + +u8 * +format_ip_neighbor_key (u8 * s, va_list * va) +{ + ip_neighbor_key_t *key = va_arg (*va, ip_neighbor_key_t *); + + return (format (s, "[%U, %U]", + format_vnet_sw_if_index_name, vnet_get_main (), + key->ipnk_sw_if_index, + format_ip46_address, &key->ipnk_ip, key->ipnk_type)); +} + +u8 * +format_ip_neighbor_watcher (u8 * s, va_list * va) +{ + ip_neighbor_watcher_t *watcher = va_arg (*va, ip_neighbor_watcher_t *); + + return (format (s, "[pid:%d, client:%d]", + clib_host_to_net_u32 (watcher->ipw_pid), + clib_host_to_net_u32 (watcher->ipw_client))); +} + +u8 * +format_ip_neighbor (u8 * s, va_list * va) +{ + index_t ipni = va_arg (*va, index_t); + ip_neighbor_t *ipn; + + ipn = ip_neighbor_get (ipni); + + return (format (s, "%=12U%=40U%=6U%=20U%U", + format_vlib_time, vlib_get_main (), + ipn->ipn_time_last_updated, + format_ip46_address, &ipn->ipn_key->ipnk_ip, IP46_TYPE_ANY, + format_ip_neighbor_flags, ipn->ipn_flags, + format_mac_address_t, &ipn->ipn_mac, + format_vnet_sw_if_index_name, vnet_get_main (), + ipn->ipn_key->ipnk_sw_if_index)); +} + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip/ip_neighbor.h b/src/vnet/ip-neighbor/ip_neighbor_types.h index d370ddeeee4..c6d4e104e8b 100644 --- a/src/vnet/ip/ip_neighbor.h +++ b/src/vnet/ip-neighbor/ip_neighbor_types.h @@ -15,8 +15,12 @@ * limitations under the License. */ -#ifndef included_ip_neighbor_h -#define included_ip_neighbor_h +#ifndef __INCLUDE_IP_NEIGHBOR_TYPES_H__ +#define __INCLUDE_IP_NEIGHBOR_TYPES_H__ + +#include <vnet/ip/ip6_packet.h> +#include <vnet/ethernet/mac_address.h> +#include <vnet/fib/fib_types.h> #define IP_SCAN_DISABLED 0 #define IP_SCAN_V4_NEIGHBORS (1 << 0) @@ -33,28 +37,86 @@ typedef struct u8 stale_threshold; /* Threshold in minutes to delete nei entry */ } ip_neighbor_scan_arg_t; -void ip_neighbor_scan_enable_disable (ip_neighbor_scan_arg_t * arg); - typedef enum ip_neighbor_flags_t_ { IP_NEIGHBOR_FLAG_NONE = 0, IP_NEIGHBOR_FLAG_STATIC = (1 << 0), IP_NEIGHBOR_FLAG_DYNAMIC = (1 << 1), IP_NEIGHBOR_FLAG_NO_FIB_ENTRY = (1 << 2), + IP_NEIGHBOR_FLAG_PENDING = (1 << 3), } __attribute__ ((packed)) ip_neighbor_flags_t; +typedef struct ip_neighbor_watcher_t_ +{ + u32 ipw_pid; + u32 ipw_client; +} ip_neighbor_watcher_t; + +extern u8 *format_ip_neighbor_watcher (u8 * s, va_list * args); + +typedef struct ip_neighbor_key_t_ +{ + ip46_address_t ipnk_ip; + ip46_type_t ipnk_type; + u32 ipnk_sw_if_index; +} ip_neighbor_key_t; + +/** + * A representation of an IP neighbour/peer + */ +typedef struct ip_neighbor_t_ +{ + /** + * The idempotent key + */ + ip_neighbor_key_t *ipn_key; + + /** + * The learned MAC address of the neighbour + */ + mac_address_t ipn_mac; + + /** + * Falgs for this object + */ + ip_neighbor_flags_t ipn_flags; + + /** + * Aging related data + * - last time the neighbour was probed + * - number of probes - 3 and it's dead + */ + f64 ipn_time_last_updated; + u8 ipn_n_probes; + index_t ipn_elt; + + /** + * The index of the adj fib created for this neighbour + */ + fib_node_index_t ipn_fib_entry_index; +} ip_neighbor_t; + extern u8 *format_ip_neighbor_flags (u8 * s, va_list * args); +extern u8 *format_ip_neighbor_key (u8 * s, va_list * args); +extern u8 *format_ip_neighbor (u8 * s, va_list * args); + +extern ip_neighbor_t *ip_neighbor_get (index_t ipni); -extern int ip_neighbor_add (const ip46_address_t * ip, - ip46_type_t type, - const mac_address_t * mac, - u32 sw_if_index, - ip_neighbor_flags_t flags, u32 * stats_index); +typedef struct ip_neighbor_learn_t_ +{ + ip46_address_t ip; + ip46_type_t type; + mac_address_t mac; + u32 sw_if_index; +} ip_neighbor_learn_t; -extern int ip_neighbor_del (const ip46_address_t * ip, - ip46_type_t type, u32 sw_if_index); +typedef struct ip_neighbor_event_t_ +{ + ip_neighbor_watcher_t ipne_watch; + index_t ipne_index; +} ip_neighbor_event_t; -#endif /* included_ip_neighbor_h */ +#endif /* __INCLUDE_IP_NEIGHBOR_H__ */ /* * fd.io coding-style-patch-verification: ON diff --git a/src/vnet/ip-neighbor/ip_neighbor_watch.c b/src/vnet/ip-neighbor/ip_neighbor_watch.c new file mode 100644 index 00000000000..7464ee62189 --- /dev/null +++ b/src/vnet/ip-neighbor/ip_neighbor_watch.c @@ -0,0 +1,283 @@ +/* + * ip_neighboor_watch.c; IP neighbor watching + * + * Copyright (c) 2019 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/ip-neighbor/ip_neighbor.h> +#include <vnet/ip/ip_types_api.h> +#include <vnet/ethernet/ethernet_types_api.h> + +#include <vnet/ip-neighbor/ip_neighbor.api_enum.h> +#include <vnet/ip-neighbor/ip_neighbor.api_types.h> + +#include <vlibmemory/api.h> + +/** + * Database of registered watchers + * The key for a watcher is {type, sw_if_index, addreess} + * interface=~0 / address=all-zeros imples any. + */ +typedef struct ip_neighbor_watch_db_t_ +{ + mhash_t ipnwdb_hash; +} ip_neighbor_watch_db_t; + +static ip_neighbor_watch_db_t ipnw_db; + +static uword +ip_neighbor_event_process (vlib_main_t * vm, + vlib_node_runtime_t * rt, vlib_frame_t * f) +{ + ip_neighbor_event_t *ipne, *ipnes = NULL; + uword event_type = ~0; + + while (1) + { + vlib_process_wait_for_event (vm); + + ipnes = vlib_process_get_event_data (vm, &event_type); + + switch (event_type) + { + default: + vec_foreach (ipne, ipnes) ip_neighbor_handle_event (ipne); + break; + + case ~0: + /* timeout - */ + break; + } + + vec_reset_length (ipnes); + } + return 0; +} + +/* *INDENT-OFF* */ +VLIB_REGISTER_NODE (ip_neighbor_event_process_node) = { + .function = ip_neighbor_event_process, + .type = VLIB_NODE_TYPE_PROCESS, + .name = "ip-neighbor-event", +}; +/* *INDENT-ON* */ + + +static clib_error_t * +want_ip_neighbor_events_reaper (u32 client_index) +{ + ip_neighbor_key_t *key, *empty_keys = NULL; + ip_neighbor_watcher_t *watchers; + uword *v; + i32 pos; + + /* walk the entire IP neighbour DB and removes the client's registrations */ + /* *INDENT-OFF* */ + mhash_foreach(key, v, &ipnw_db.ipnwdb_hash, + ({ + watchers = (ip_neighbor_watcher_t*) *v; + + vec_foreach_index_backwards (pos, watchers) { + if (watchers[pos].ipw_client == client_index) + vec_del1(watchers, pos); + } + + if (vec_len(watchers) == 0) + vec_add1 (empty_keys, *key); + })); + /* *INDENT-OFF* */ + + vec_foreach (key, empty_keys) + mhash_unset (&ipnw_db.ipnwdb_hash, key, NULL); + vec_free (empty_keys); + return (NULL); +} + +VL_MSG_API_REAPER_FUNCTION (want_ip_neighbor_events_reaper); + +static int +ip_neighbor_watch_cmp (const ip_neighbor_watcher_t * w1, + const ip_neighbor_watcher_t * w2) +{ + return (0 == clib_memcmp (w1, w2, sizeof(*w1))); +} + +void +ip_neighbor_watch (const ip46_address_t * ip, + ip46_type_t type, + u32 sw_if_index, + const ip_neighbor_watcher_t * watch) +{ + ip_neighbor_key_t key = { + .ipnk_ip = *ip, + .ipnk_sw_if_index = (sw_if_index == 0 ? ~0 : sw_if_index), + .ipnk_type = type, + }; + ip_neighbor_watcher_t *ipws = NULL; + uword *p; + + p = mhash_get (&ipnw_db.ipnwdb_hash, &key); + + if (p) + { + ipws = (ip_neighbor_watcher_t*) p[0]; + + if (~0 != vec_search_with_function (ipws, watch, + ip_neighbor_watch_cmp)) + /* duplicate */ + return; + } + + vec_add1 (ipws, *watch); + + mhash_set (&ipnw_db.ipnwdb_hash, &key, (uword) ipws, NULL); +} + +void +ip_neighbor_unwatch (const ip46_address_t * ip, + ip46_type_t type, + u32 sw_if_index, + const ip_neighbor_watcher_t * watch) +{ + ip_neighbor_key_t key = { + .ipnk_ip = *ip, + .ipnk_sw_if_index = (sw_if_index == 0 ? ~0 : sw_if_index), + .ipnk_type = type, + }; + ip_neighbor_watcher_t *ipws = NULL; + uword *p; + u32 pos; + + p = mhash_get (&ipnw_db.ipnwdb_hash, &key); + + if (!p) + return; + + ipws = (ip_neighbor_watcher_t*) p[0]; + + pos = vec_search_with_function (ipws, watch, ip_neighbor_watch_cmp); + + if (~0 == pos) + return; + + vec_del1 (ipws, pos); + + if (vec_len(ipws) == 0) + mhash_unset (&ipnw_db.ipnwdb_hash, &key, NULL); +} + +static void +ip_neighbor_signal (ip_neighbor_watcher_t *watchers, index_t ipni) +{ + ip_neighbor_watcher_t *watcher; + + vec_foreach (watcher, watchers) { + ip_neighbor_event_t *ipne; + + ipne = vlib_process_signal_event_data (vlib_get_main(), + ip_neighbor_event_process_node.index, + 0, 1, sizeof(*ipne)); + ipne->ipne_watch = *watcher; + ipne->ipne_index = ipni; + } +} + +void +ip_neighbor_publish (index_t ipni) +{ + const ip_neighbor_t *ipn; + ip_neighbor_key_t key; + uword *p; + + ipn = ip_neighbor_get (ipni); + + clib_memcpy (&key, ipn->ipn_key, sizeof (key)); + + /* Search the DB from longest to shortest key */ + p = mhash_get (&ipnw_db.ipnwdb_hash, &key); + + if (p) { + ip_neighbor_signal ((ip_neighbor_watcher_t*) p[0], ipni); + } + + ip46_address_reset (&key.ipnk_ip); + p = mhash_get (&ipnw_db.ipnwdb_hash, &key); + + if (p) { + ip_neighbor_signal ((ip_neighbor_watcher_t*) p[0], ipni); + } + + key.ipnk_sw_if_index = ~0; + p = mhash_get (&ipnw_db.ipnwdb_hash, &key); + + if (p) { + ip_neighbor_signal ((ip_neighbor_watcher_t*) p[0], ipni); + } +} + +static clib_error_t * +ip_neighbor_watchers_show (vlib_main_t * vm, + unformat_input_t * input, + vlib_cli_command_t * cmd) +{ + ip_neighbor_watcher_t *watchers, *watcher; + ip_neighbor_key_t *key; + uword *v; + + /* *INDENT-OFF* */ + mhash_foreach(key, v, &ipnw_db.ipnwdb_hash, + ({ + watchers = (ip_neighbor_watcher_t*) *v; + + ASSERT(vec_len(watchers)); + vlib_cli_output (vm, "Key: %U", format_ip_neighbor_key, key); + + vec_foreach (watcher, watchers) + vlib_cli_output (vm, " %U", format_ip_neighbor_watcher, watcher); + })); + /* *INDENT-ON* */ + return (NULL); +} + +/* *INDENT-OFF* */ +VLIB_CLI_COMMAND (show_ip_neighbor_watchers_cmd_node, static) = { + .path = "show ip neighbor-watcher", + .function = ip_neighbor_watchers_show, + .short_help = "show ip neighbors-watcher", +}; +/* *INDENT-ON* */ + +static clib_error_t * +ip_neighbor_watch_init (vlib_main_t * vm) +{ + mhash_init (&ipnw_db.ipnwdb_hash, + sizeof (ip_neighbor_watcher_t *), sizeof (ip_neighbor_key_t)); + return (NULL); +} + +/* *INDENT-OFF* */ +VLIB_INIT_FUNCTION (ip_neighbor_watch_init) = +{ + .runs_after = VLIB_INITS("ip_neighbor_init"), +}; +/* *INDENT-ON* */ + + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip-neighbor/ip_neighbor_watch.h b/src/vnet/ip-neighbor/ip_neighbor_watch.h new file mode 100644 index 00000000000..91d9f6fe12f --- /dev/null +++ b/src/vnet/ip-neighbor/ip_neighbor_watch.h @@ -0,0 +1,42 @@ +/* + * ip_neighboor_watch.h: ip neighbor event handling + * + * Copyright (c) 2018 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef __IP_NEIGHBOR_WATCH_H__ +#define __IP_NEIGHBOR_WATCH_H__ + +#include <vnet/ip-neighbor/ip_neighbor_types.h> + +extern void ip_neighbor_watch (const ip46_address_t * ip, + ip46_type_t type, + u32 sw_if_index, + const ip_neighbor_watcher_t * watch); +extern void ip_neighbor_unwatch (const ip46_address_t * ip, + ip46_type_t type, + u32 sw_if_index, + const ip_neighbor_watcher_t * watch); + +extern void ip_neighbor_publish (index_t ipni); + +#endif + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip/format.h b/src/vnet/ip/format.h index 952455c975a..6e9075cc877 100644 --- a/src/vnet/ip/format.h +++ b/src/vnet/ip/format.h @@ -58,8 +58,6 @@ typedef enum format_ip_adjacency_flags_t_ format_function_t format_ip_adjacency; format_function_t format_ip_adjacency_packet_data; -format_function_t format_ip46_address; - /* unformat_ip46_address expects arguments (ip46_address_t *, ip46_type_t) * The type argument is used to enforce a particular IP version. */ unformat_function_t unformat_ip46_address; diff --git a/src/vnet/ip/icmp46_packet.h b/src/vnet/ip/icmp46_packet.h index e761af8bf16..0545046fe60 100644 --- a/src/vnet/ip/icmp46_packet.h +++ b/src/vnet/ip/icmp46_packet.h @@ -187,13 +187,15 @@ typedef enum #undef _ } icmp6_code_t; +/* *INDENT-OFF* */ typedef CLIB_PACKED (struct - { - u8 type; - u8 code; - /* IP checksum of icmp header plus data which follows. */ - u16 checksum; - }) icmp46_header_t; +{ + u8 type; + u8 code; + /* IP checksum of icmp header plus data which follows. */ + u16 checksum; +}) icmp46_header_t; +/* *INDENT-ON* */ /* ip6 neighbor discovery */ #define foreach_icmp6_neighbor_discovery_option \ @@ -236,164 +238,125 @@ typedef enum icmp6_neighbor_discovery_option_type #undef _ } icmp6_neighbor_discovery_option_type_t; +/* *INDENT-OFF* */ typedef CLIB_PACKED (struct - { - /* Option type. */ - u8 type; - /* Length of this header plus option data in 8 byte units. */ - u8 n_data_u64s; - /* Option data follows. */ - u8 data[0]; - }) icmp6_neighbor_discovery_option_header_t; +{ + /* Option type. */ + u8 type; + /* Length of this header plus option data in 8 byte units. */ + u8 n_data_u64s; + /* Option data follows. */ + u8 data[0]; +}) icmp6_neighbor_discovery_option_header_t; typedef CLIB_PACKED (struct - { - icmp6_neighbor_discovery_option_header_t header; - u8 dst_address_length; - u8 flags; +{ + icmp6_neighbor_discovery_option_header_t header; + u8 dst_address_length; + u8 flags; #define ICMP6_NEIGHBOR_DISCOVERY_PREFIX_INFORMATION_FLAG_ON_LINK (1 << 7) #define ICMP6_NEIGHBOR_DISCOVERY_PREFIX_INFORMATION_AUTO (1 << 6) - u32 valid_time; - u32 preferred_time; - u32 unused; ip6_address_t dst_address; - }) icmp6_neighbor_discovery_prefix_information_option_t; + u32 valid_time; + u32 preferred_time; + u32 unused; ip6_address_t dst_address; +}) icmp6_neighbor_discovery_prefix_information_option_t; typedef CLIB_PACKED (struct - { - u8 type; - u8 aux_data_len_u32s; - u16 num_sources; - ip6_address_t mcast_addr; ip6_address_t source_addr[0]; - }) icmp6_multicast_address_record_t; - -typedef CLIB_PACKED (struct - { - ip6_hop_by_hop_ext_t ext_hdr; - ip6_router_alert_option_t alert; - ip6_padN_option_t pad; - icmp46_header_t icmp; - u16 rsvd; - u16 num_addr_records; - icmp6_multicast_address_record_t records[0]; - }) icmp6_multicast_listener_report_header_t; +{ + u8 type; + u8 aux_data_len_u32s; + u16 num_sources; + ip6_address_t mcast_addr; ip6_address_t source_addr[0]; +}) icmp6_multicast_address_record_t; typedef CLIB_PACKED (struct - { - icmp6_neighbor_discovery_option_header_t header; - u8 reserved[6]; - /* IP6 header plus payload follows. */ - u8 data[0]; - }) icmp6_neighbor_discovery_redirected_header_option_t; +{ + icmp6_neighbor_discovery_option_header_t header; + u8 reserved[6]; + /* IP6 header plus payload follows. */ + u8 data[0]; +}) icmp6_neighbor_discovery_redirected_header_option_t; typedef CLIB_PACKED (struct - { - icmp6_neighbor_discovery_option_header_t header; - u16 unused; u32 mtu; - }) icmp6_neighbor_discovery_mtu_option_t; +{ + icmp6_neighbor_discovery_option_header_t header; + u16 unused; u32 mtu; +}) icmp6_neighbor_discovery_mtu_option_t; typedef CLIB_PACKED (struct - { - icmp6_neighbor_discovery_option_header_t header; - u8 ethernet_address[6]; - }) - icmp6_neighbor_discovery_ethernet_link_layer_address_option_t; +{ + icmp6_neighbor_discovery_option_header_t header; + u8 ethernet_address[6]; +}) icmp6_neighbor_discovery_ethernet_link_layer_address_option_t; typedef CLIB_PACKED (struct - { - icmp6_neighbor_discovery_option_header_t header; - u8 max_l2_address[6 + 8]; - }) - icmp6_neighbor_discovery_max_link_layer_address_option_t; +{ + icmp6_neighbor_discovery_option_header_t header; + u8 max_l2_address[6 + 8]; +}) icmp6_neighbor_discovery_max_link_layer_address_option_t; /* Generic neighbor discover header. Used for router solicitations, etc. */ typedef CLIB_PACKED (struct - { - icmp46_header_t icmp; u32 reserved_must_be_zero; - }) icmp6_neighbor_discovery_header_t; +{ + icmp46_header_t icmp; u32 reserved_must_be_zero; +}) icmp6_neighbor_discovery_header_t; /* Router advertisement packet formats. */ typedef CLIB_PACKED (struct - { - icmp46_header_t icmp; - /* Current hop limit to use for outgoing packets. */ - u8 current_hop_limit; - u8 flags; +{ + icmp46_header_t icmp; + /* Current hop limit to use for outgoing packets. */ + u8 current_hop_limit; + u8 flags; #define ICMP6_ROUTER_DISCOVERY_FLAG_ADDRESS_CONFIG_VIA_DHCP (1 << 7) #define ICMP6_ROUTER_DISCOVERY_FLAG_OTHER_CONFIG_VIA_DHCP (1 << 6) - /* Zero means unspecified. */ - u16 router_lifetime_in_sec; - /* Zero means unspecified. */ - u32 neighbor_reachable_time_in_msec; - /* Zero means unspecified. */ - u32 - time_in_msec_between_retransmitted_neighbor_solicitations; - /* Options that may follow: source_link_layer_address, mtu, prefix_information. */ - }) icmp6_router_advertisement_header_t; + /* Zero means unspecified. */ + u16 router_lifetime_in_sec; + /* Zero means unspecified. */ + u32 neighbor_reachable_time_in_msec; + /* Zero means unspecified. */ + u32 + time_in_msec_between_retransmitted_neighbor_solicitations; + /* Options that may follow: source_link_layer_address, mtu, prefix_information. */ +}) icmp6_router_advertisement_header_t; /* Neighbor solicitation/advertisement header. */ typedef CLIB_PACKED (struct - { - icmp46_header_t icmp; - /* Zero for solicitation; flags for advertisement. */ - u32 advertisement_flags; - /* Set when sent by a router. */ +{ + icmp46_header_t icmp; + /* Zero for solicitation; flags for advertisement. */ + u32 advertisement_flags; + /* Set when sent by a router. */ #define ICMP6_NEIGHBOR_ADVERTISEMENT_FLAG_ROUTER (1 << 31) - /* Set when response to solicitation. */ + /* Set when response to solicitation. */ #define ICMP6_NEIGHBOR_ADVERTISEMENT_FLAG_SOLICITED (1 << 30) #define ICMP6_NEIGHBOR_ADVERTISEMENT_FLAG_OVERRIDE (1 << 29) - ip6_address_t target_address; - /* Options that may follow: source_link_layer_address - (for solicitation) target_link_layer_address (for advertisement). */ - }) icmp6_neighbor_solicitation_or_advertisement_header_t; + ip6_address_t target_address; + /* Options that may follow: source_link_layer_address + (for solicitation) target_link_layer_address (for advertisement). */ +}) icmp6_neighbor_solicitation_or_advertisement_header_t; typedef CLIB_PACKED (struct - { - icmp46_header_t icmp; - u32 reserved_must_be_zero; - /* Better next hop to use for given destination. */ - ip6_address_t better_next_hop_address; - ip6_address_t dst_address; - /* Options that may follow: target_link_layer_address, - redirected_header. */ - }) icmp6_redirect_header_t; +{ + icmp46_header_t icmp; + u32 reserved_must_be_zero; + /* Better next hop to use for given destination. */ + ip6_address_t better_next_hop_address; + ip6_address_t dst_address; + /* Options that may follow: target_link_layer_address, + redirected_header. */ +}) icmp6_redirect_header_t; /* Solicitation/advertisement packet format for ethernet. */ typedef CLIB_PACKED (struct - { - ip6_header_t ip; - icmp6_neighbor_solicitation_or_advertisement_header_t - neighbor; - icmp6_neighbor_discovery_ethernet_link_layer_address_option_t - link_layer_option; - }) icmp6_neighbor_solicitation_header_t; - -/* Router solicitation packet format for ethernet. */ -typedef CLIB_PACKED (struct - { - ip6_header_t ip; - icmp6_neighbor_discovery_header_t neighbor; - icmp6_neighbor_discovery_ethernet_link_layer_address_option_t - link_layer_option; - }) icmp6_router_solicitation_header_t; - -/* router advertisement packet format for ethernet. */ -typedef CLIB_PACKED (struct - { - ip6_header_t ip; - icmp6_router_advertisement_header_t router; - icmp6_neighbor_discovery_ethernet_link_layer_address_option_t - link_layer_option; - icmp6_neighbor_discovery_mtu_option_t mtu_option; - icmp6_neighbor_discovery_prefix_information_option_t - prefix[0]; - }) icmp6_router_advertisement_packet_t; - -/* multicast listener report packet format for ethernet. */ -typedef CLIB_PACKED (struct - { - ip6_header_t ip; - icmp6_multicast_listener_report_header_t report_hdr; - }) icmp6_multicast_listener_report_packet_t; +{ + ip6_header_t ip; + icmp6_neighbor_solicitation_or_advertisement_header_t + neighbor; + icmp6_neighbor_discovery_ethernet_link_layer_address_option_t + link_layer_option; +}) icmp6_neighbor_solicitation_header_t; +/* *INDENT-ON* */ #endif /* included_vnet_icmp46_packet_h */ diff --git a/src/vnet/ip/icmp6.c b/src/vnet/ip/icmp6.c index 95baa71344c..4bba430fadc 100644 --- a/src/vnet/ip/icmp6.c +++ b/src/vnet/ip/icmp6.c @@ -816,7 +816,7 @@ icmp6_init (vlib_main_t * vm) icmp6_register_type (vm, ICMP6_echo_request, ip6_icmp_echo_request_node.index); - return vlib_call_init_function (vm, ip6_neighbor_init); + return (NULL); } VLIB_INIT_FUNCTION (icmp6_init); diff --git a/src/vnet/ip/ip.api b/src/vnet/ip/ip.api index 0bdd1529b20..831b38ac53c 100644 --- a/src/vnet/ip/ip.api +++ b/src/vnet/ip/ip.api @@ -189,76 +189,6 @@ manual_endian manual_print define ip_route_details vl_api_ip_route_t route; }; -/** \brief IP neighbor flags - @param is_static - A static neighbor Entry - there are not flushed - If the interface goes down. - @param is_no_fib_entry - Do not create a corresponding entry in the FIB - table for the neighbor. -*/ -enum ip_neighbor_flags -{ - IP_API_NEIGHBOR_FLAG_NONE = 0, - IP_API_NEIGHBOR_FLAG_STATIC = 0x1, - IP_API_NEIGHBOR_FLAG_NO_FIB_ENTRY = 0x2, -}; - -/** \brief IP neighbor - @param sw_if_index - interface used to reach neighbor - @param mac_address - l2 address of the neighbor - @param ip_address - ip4 or ip6 address of the neighbor - @param flags - flags for the nieghbor -*/ -typedef ip_neighbor { - vl_api_interface_index_t sw_if_index; - vl_api_ip_neighbor_flags_t flags; - vl_api_mac_address_t mac_address; - vl_api_address_t ip_address; -}; - -/** \brief IP neighbor add / del request - @param client_index - opaque cookie to identify the sender - @param context - sender context, to match reply w/ request - @param is_add - 1 to add neighbor, 0 to delete - @param neighbor - the neighor to add/remove -*/ -define ip_neighbor_add_del -{ - u32 client_index; - u32 context; - /* 1 = add, 0 = delete */ - bool is_add; - vl_api_ip_neighbor_t neighbor; -}; -define ip_neighbor_add_del_reply -{ - u32 context; - i32 retval; - u32 stats_index; -}; - -/** \brief Dump IP neighboors - @param client_index - opaque cookie to identify the sender - @param context - sender context, to match reply w/ request - @param sw_if_index - the interface to dump neighboors, ~0 == all - @param is_ipv6 - [1|0] to indicate if address family is ipv[6|4] -*/ -define ip_neighbor_dump -{ - u32 client_index; - u32 context; - vl_api_interface_index_t sw_if_index; - bool is_ipv6; -}; - -/** \brief IP neighboors dump response - @param context - sender context which was passed in the request - @param neighbour - the neighbor -*/ -define ip_neighbor_details { - u32 context; - vl_api_ip_neighbor_t neighbor; -}; - /** \brief Set the ip flow hash config for a fib request @param client_index - opaque cookie to identify the sender @param context - sender context, to match reply w/ request @@ -287,150 +217,6 @@ autoreply define set_ip_flow_hash u8 symmetric; }; -/** \brief IPv6 router advertisement config request - @param client_index - opaque cookie to identify the sender - @param context - sender context, to match reply w/ request - @param suppress - - @param managed - - @param other - - @param ll_option - - @param send_unicast - - @param cease - - @param is_no - - @param default_router - - @param max_interval - - @param min_interval - - @param lifetime - - @param initial_count - - @param initial_interval - -*/ -autoreply define sw_interface_ip6nd_ra_config -{ - u32 client_index; - u32 context; - vl_api_interface_index_t sw_if_index; - u8 suppress; - u8 managed; - u8 other; - u8 ll_option; - u8 send_unicast; - u8 cease; - bool is_no; - u8 default_router; - u32 max_interval; - u32 min_interval; - u32 lifetime; - u32 initial_count; - u32 initial_interval; -}; - -/** \brief IPv6 router advertisement prefix config request - @param client_index - opaque cookie to identify the sender - @param context - sender context, to match reply w/ request - @param sw_if_index - The interface the RA prefix information is for - @param prefix - The prefix to advertise - @param use_default - Revert to default settings - @param no_advertise - Do not advertise this prefix - @param off_link - The prefix is off link (it is not configured on the interface) - Configures the L-flag, When set, indicates that this - prefix can be used for on-link determination. - @param no_autoconfig - Setting for the A-flag. When - set indicates that this prefix can be used for - stateless address configuration. - @param no_onlink - The prefix is not on link. Make sure this is consistent - with the off_link parameter else YMMV - @param is_no - add/delete - @param val_lifetime - The length of time in - seconds (relative to the time the packet is sent) - that the prefix is valid for the purpose of on-link - determination. A value of all one bits - (0xffffffff) represents infinity - @param pref_lifetime - The length of time in - seconds (relative to the time the packet is sent) - that addresses generated from the prefix via - stateless address autoconfiguration remain - preferred [ADDRCONF]. A value of all one bits - (0xffffffff) represents infinity. -*/ -autoreply define sw_interface_ip6nd_ra_prefix -{ - u32 client_index; - u32 context; - vl_api_interface_index_t sw_if_index; - vl_api_prefix_t prefix; - bool use_default; - bool no_advertise; - bool off_link; - bool no_autoconfig; - bool no_onlink; - bool is_no; - u32 val_lifetime; - u32 pref_lifetime; -}; - -/** \brief IPv6 ND proxy config - @param client_index - opaque cookie to identify the sender - @param context - sender context, to match reply w/ request - @param sw_if_index - The interface the host is on - @param address - The address of the host for which to proxy for - @param is_add - Adding or deleting -*/ -autoreply define ip6nd_proxy_add_del -{ - u32 client_index; - u32 context; - vl_api_interface_index_t sw_if_index; - bool is_del; - vl_api_ip6_address_t ip; -}; - -/** \brief IPv6 ND proxy details returned after request - @param context - sender context, to match reply w/ request - @param retval - return code for the request -*/ -define ip6nd_proxy_details -{ - u32 context; - vl_api_interface_index_t sw_if_index; - vl_api_ip6_address_t ip; -}; - -/** \brief IPv6 ND proxy dump request - @param context - sender context, to match reply w/ request - @param retval - return code for the request - @param sw_if_index - The interface the host is on - @param address - The address of the host for which to proxy for -*/ -define ip6nd_proxy_dump -{ - u32 client_index; - u32 context; -}; - -/** \brief Start / stop sending router solicitation - @param client_index - opaque cookie to identify the sender - @param context - sender context, to match reply w/ request - @param irt - initial retransmission time - @param mrt - maximum retransmission time - @param mrc - maximum retransmission count - @param mrd - maximum retransmission duration - @param sw_if_index - software interface index of interface - for sending router solicitation - @param stop - if non-zero then stop sending router solicitation, - otherwise start sending router solicitation -*/ -autoreply define ip6nd_send_router_solicitation -{ - u32 client_index; - u32 context; - u32 irt; - u32 mrt; - u32 mrc; - u32 mrd; - u32 sw_if_index; - bool stop; -}; - /** \brief IPv6 interface enable / disable request @param client_index - opaque cookie to identify the sender @param context - sender context, to match reply w/ request @@ -445,20 +231,6 @@ autoreply define sw_interface_ip6_enable_disable bool enable; /* set to true if enable */ }; -/** \brief IPv6 set link local address on interface request - @param client_index - opaque cookie to identify the sender - @param context - sender context, to match reply w/ request - @param sw_if_index - interface to set link local on - @param address[] - the new link local address -*/ -autoreply define sw_interface_ip6_set_link_local_address -{ - u32 client_index; - u32 context; - vl_api_interface_index_t sw_if_index; - vl_api_ip6_address_t address; -}; - /** \brief Dump IP multicast fib table @param client_index - opaque cookie to identify the sender */ @@ -752,279 +524,20 @@ autoreply define ip_source_check_interface_add_del vl_api_interface_index_t sw_if_index; }; -/** \brief Enable/disable periodic IP neighbor scan - @param client_index - opaque cookie to identify the sender - @param context - sender context, to match reply w/ request - @param mode - 0: disable, 1: IPv4, 2: IPv6, 3: both IPv4/v6 - @param scan_interval - neighbor scan interval in minutes, 0: default to 1 - @param max_proc_time - max processing time per run in usec, 0: default to 20 - @param max_update - max neighbor probe/delete per run, 0: default to 10 - @param scan_int_delay - delay in msec to resume scan if exceed max proc - time or update, 0: default to 1 - @param stale_threshold - threshold in minutes for neighbor deletion, - 0: default to 4*scan_interval -*/ -autoreply define ip_scan_neighbor_enable_disable -{ - u32 client_index; - u32 context; - u8 mode; - u8 scan_interval; - u8 max_proc_time; - u8 max_update; - u8 scan_int_delay; - u8 stale_threshold; -}; - -/** \brief IP probe neighbor address on an interface by sending an - ARP request (for IP4) or ICMP6 Neighbor Solicitation (for IP6) - @param client_index - opaque cookie to identify the sender - @param context - sender context, to match reply w/ request - @param sw_if_index - interface index - @param dst - target IP address to send IP addr resolution request -*/ -autoreply define ip_probe_neighbor -{ - u32 client_index; - u32 context; - vl_api_interface_index_t sw_if_index; - vl_api_address_t dst; -}; - -/** \brief Register for IP4 ARP resolution event on receing ARP reply or - MAC/IP info from ARP requests in L2 BDs +/** \brief IPv6 set link local address on interface request @param client_index - opaque cookie to identify the sender @param context - sender context, to match reply w/ request - @param enable_disable - 1 => register for events, 0 => cancel registration - @param pid - sender's pid - @param ip - exact IP4 address of interested arp resolution event, or - 0 to get MAC/IP info from ARP requests in BDs + @param sw_if_index - interface to set link local on + @param ip - the new link local address */ -autoreply define want_ip4_arp_events +autoreply define sw_interface_ip6_set_link_local_address { u32 client_index; u32 context; - bool enable_disable; - u32 pid; - vl_api_ip4_address_t ip; -}; - -/** \brief Tell client about an IP4 ARP resolution event or - MAC/IP info from ARP requests in L2 BDs - @param client_index - opaque cookie to identify the sender - @param ip - the exact ip4 address of interest - @param pid - client pid registered to receive notification - @param sw_if_index - interface which received ARP packet - @param mac - the new mac address - @param mac_ip - 0: ARP resolution event, 1: MAC/IP info from L2 BDs -*/ -define ip4_arp_event -{ - u32 client_index; - vl_api_ip4_address_t ip; - u32 pid; vl_api_interface_index_t sw_if_index; - vl_api_mac_address_t mac; - u8 mac_ip; -}; - -service { - rpc want_ip4_arp_events returns want_ip4_arp_events_reply - events ip4_arp_event; -}; - -/** \brief Register for IP6 ND resolution event on recieving NA reply - MAC/IP info from ICMP6 Neighbor Solicitation in L2 BDs - @param client_index - opaque cookie to identify the sender - @param context - sender context, to match reply w/ request - @param enable_disable - 1 => register for events, 0 => cancel registration - @param pid - sender's pid - @param ip - the exact IP6 address of interested ND resolution event, or - 0 to get MAC/IP info from ICMP6 NS in L2 BDs. -*/ -autoreply define want_ip6_nd_events -{ - u32 client_index; - u32 context; - bool enable_disable; - u32 pid; vl_api_ip6_address_t ip; }; -/** \brief Tell client about an IP6 ND resolution or - MAC/IP info from ICMP6 Neighbor Solicitation in L2 BDs. - @param client_index - opaque cookie to identify the sender - @param pid - client pid registered to receive notification - @param sw_if_index - interface which received ARP packet - @param ip - the exact ip6 address of interest - @param new_mac - the new mac address - @param mac_ip - 0: ND resolution event, 1: MAC/IP info from L2 BDs -*/ -define ip6_nd_event -{ - u32 client_index; - u32 pid; - vl_api_interface_index_t sw_if_index; - vl_api_ip6_address_t ip; - vl_api_mac_address_t mac; - u8 mac_ip; -}; - -service { - rpc want_ip6_ra_events returns want_ip6_ra_events_reply - events ip6_ra_event; -}; - -/** \brief Register for ip6 router advertisement events - @param client_index - opaque cookie to identify the sender - @param context - sender context, to match reply w/ request - @param enable_disable - 1 => register for events, 0 => cancel registration - @param pid - sender's pid -*/ -autoreply define want_ip6_ra_events -{ - u32 client_index; - u32 context; - bool enable_disable; - u32 pid; -}; - -/** \brief Struct representing RA prefix info - @param prefix - RA prefix info destination address - @param flags - RA prefix info flags - @param valid_time - RA prefix info valid time - @param preferred_time - RA prefix info preferred time -*/ -typedef ip6_ra_prefix_info -{ - vl_api_prefix_t prefix; - u8 flags; - u32 valid_time; - u32 preferred_time; -}; - -/** \brief Tell client about a router advertisement event - @param client_index - opaque cookie to identify the sender - @param pid - client pid registered to receive notification - @param current_hop_limit - RA current hop limit - @param flags - RA flags - @param router_lifetime_in_sec - RA lifetime in seconds - @param router_addr - The router's address - @param neighbor_reachable_time_in_msec - RA neighbor reachable time in msec - @param time_in_msec_between_retransmitted_neighbor_solicitations - - time in msec between retransmitted neighbor solicitations - @param n_prefixes - - @param prefixes - -*/ -define ip6_ra_event -{ - u32 client_index; - u32 pid; - vl_api_interface_index_t sw_if_index; - vl_api_ip6_address_t router_addr; - u8 current_hop_limit; - u8 flags; - u16 router_lifetime_in_sec; - u32 neighbor_reachable_time_in_msec; - u32 time_in_msec_between_retransmitted_neighbor_solicitations; - u32 n_prefixes; - vl_api_ip6_ra_prefix_info_t prefixes[n_prefixes]; -}; - -service { - rpc want_ip6_nd_events returns want_ip6_nd_events_reply - events ip6_nd_event; -}; - -/** \brief Proxy ARP configuration type - @param table_id - VRF / Fib table ID - @param low - Low address of the Proxy ARP range - @param hi - High address of the Proxy ARP range -*/ -typedef proxy_arp -{ - u32 table_id; - vl_api_ip4_address_t low; - vl_api_ip4_address_t hi; -}; - -/** \brief Proxy ARP add / del request - @param client_index - opaque cookie to identify the sender - @param context - sender context, to match reply w/ request - @param is_add - 1 if adding the Proxy ARP range, 0 if deleting - @param proxy - Proxy configuration -*/ -autoreply define proxy_arp_add_del -{ - u32 client_index; - u32 context; - bool is_add; - vl_api_proxy_arp_t proxy; -}; - -/** \brief Proxy ARP dump request - */ -define proxy_arp_dump -{ - u32 client_index; - u32 context; -}; - -/** \brief Proxy ARP dump details reply - * @param proxy - Same data as used to configure - */ -define proxy_arp_details -{ - u32 context; - vl_api_proxy_arp_t proxy; -}; - -/** \brief Proxy ARP add / del interface request - @param client_index - opaque cookie to identify the sender - @param context - sender context, to match reply w/ request - @param sw_if_index - Which interface to enable / disable Proxy Arp on - @param enable_disable - 1 to enable Proxy ARP on interface, 0 to disable -*/ -autoreply define proxy_arp_intfc_enable_disable -{ - u32 client_index; - u32 context; - vl_api_interface_index_t sw_if_index; - /* 1 = on, 0 = off */ - bool enable_disable; -}; - -/** \brief Proxy ARP interface dump request - */ -define proxy_arp_intfc_dump -{ - u32 client_index; - u32 context; -}; - -/** \brief Proxy ARP interface dump details reply - * @param sw_if_index The interface on which ARP proxy is enabled. - */ -define proxy_arp_intfc_details -{ - u32 context; - vl_api_interface_index_t sw_if_index; -}; - -/** \brief Set max allowed ARP or ip6 neighbor entries request - @param client_index - opaque cookie to identify the sender - @param context - sender context, to match reply w/ request - @param is_ipv6 - neighbor limit if non-zero, else ARP limit - @param arp_neighbor_limit - the new limit, defaults are ~ 50k -*/ -autoreply define set_arp_neighbor_limit -{ - u32 client_index; - u32 context; - bool is_ipv6; - u32 arp_neighbor_limit; -}; - /** \brief IOAM enable : Enable in-band OAM @param id - profile id @param seqno - To enable Seqno Processing diff --git a/src/vnet/ip/ip4.h b/src/vnet/ip/ip4.h index 339b84b1b62..a7e05a41f75 100644 --- a/src/vnet/ip/ip4.h +++ b/src/vnet/ip/ip4.h @@ -230,7 +230,7 @@ ip4_src_address_for_packet (ip_lookup_main_t * lm, /* Find interface address which matches destination. */ always_inline ip4_address_t * ip4_interface_address_matching_destination (ip4_main_t * im, - ip4_address_t * dst, + const ip4_address_t * dst, u32 sw_if_index, ip_interface_address_t ** result_ia) @@ -271,12 +271,6 @@ void ip4_sw_interface_enable_disable (u32 sw_if_index, u32 is_enable); int ip4_address_compare (ip4_address_t * a1, ip4_address_t * a2); -/* 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); - -clib_error_t *ip4_set_arp_limit (u32 arp_limit); - uword ip4_udp_register_listener (vlib_main_t * vm, u16 dst_port, u32 next_node_index); diff --git a/src/vnet/ip/ip46_address.c b/src/vnet/ip/ip46_address.c new file mode 100644 index 00000000000..184392e036d --- /dev/null +++ b/src/vnet/ip/ip46_address.c @@ -0,0 +1,76 @@ +/* + * Copyright (c) 2015-2019 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/ip/ip46_address.h> + +u8 * +format_ip46_type (u8 * s, va_list * args) +{ + ip46_type_t type = va_arg (*args, ip46_type_t); + + switch (type) + { + case IP46_TYPE_IP4: + return (format (s, "ip4")); + case IP46_TYPE_IP6: + return (format (s, "ip6")); + case IP46_TYPE_ANY: + return (format (s, "any")); + } + + return (format (s, "unknown")); +} + +void +ip4_address_increment (ip4_address_t * i) +{ + u32 t = clib_net_to_host_u32 (i->as_u32); + t++; + i->as_u32 = clib_net_to_host_u32 (t); +} + +void +ip6_address_increment (ip6_address_t * i) +{ + u64 tmp = clib_net_to_host_u64 (i->as_u64[1]); + + tmp++; + i->as_u64[1] = clib_host_to_net_u64 (tmp); + + if (!tmp) + { + tmp = clib_net_to_host_u64 (i->as_u64[0]); + tmp++; + i->as_u64[0] = clib_host_to_net_u64 (tmp); + } +} + +void +ip46_address_increment (ip46_type_t type, ip46_address_t * ip) +{ + if (IP46_TYPE_IP4 == type || + (IP46_TYPE_ANY == type && ip46_address_is_ip4 (ip))) + ip4_address_increment (&ip->ip4); + else + ip6_address_increment (&ip->ip6); +} + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip/ip46_address.h b/src/vnet/ip/ip46_address.h new file mode 100644 index 00000000000..f726178ee63 --- /dev/null +++ b/src/vnet/ip/ip46_address.h @@ -0,0 +1,172 @@ +/* + * Copyright (c) 2015-2019 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef included_ip46_address_h +#define included_ip46_address_h + +#include <vnet/ip/ip6_packet.h> +#include <vnet/ip/ip4_packet.h> + +typedef enum +{ + IP46_TYPE_ANY, + IP46_TYPE_BOTH = IP46_TYPE_ANY, + IP46_TYPE_IP4, + IP46_TYPE_IP6 +} ip46_type_t; + +#define IP46_N_TYPES (IP46_TYPE_IP6+2) + +#define FOREACH_IP46_TYPE(_type) \ + for (_type = IP46_TYPE_IP4; _type <= IP46_TYPE_IP6; _type++) + +extern u8 *format_ip46_type (u8 * s, va_list * args); + +/* *INDENT-OFF* */ +typedef CLIB_PACKED (union ip46_address_t_ { + struct { + u32 pad[3]; + ip4_address_t ip4; + }; + ip6_address_t ip6; + u8 as_u8[16]; + u64 as_u64[2]; +}) ip46_address_t; +/* *INDENT-ON* */ + + +format_function_t format_ip46_address; + +#define ip46_address_initializer {{{ 0 }}} + +always_inline u8 +ip46_address_is_ip4 (const ip46_address_t * ip46) +{ + return (((ip46)->pad[0] | (ip46)->pad[1] | (ip46)->pad[2]) == 0); +} + +always_inline void +ip46_address_mask_ip4 (ip46_address_t * ip46) +{ + ((ip46)->pad[0] = (ip46)->pad[1] = (ip46)->pad[2] = 0); +} + +always_inline void +ip46_address_set_ip4 (ip46_address_t * ip46, const ip4_address_t * ip) +{ + ip46_address_mask_ip4 (ip46); + ip46->ip4 = *ip; +} + +always_inline void +ip46_address_reset (ip46_address_t * ip46) +{ + ip46->as_u64[0] = ip46->as_u64[1] = 0; +} + +always_inline int +ip46_address_cmp (const ip46_address_t * ip46_1, + const ip46_address_t * ip46_2) +{ + return (memcmp (ip46_1, ip46_2, sizeof (*ip46_1))); +} + +always_inline u8 +ip46_address_is_zero (const ip46_address_t * ip46) +{ + return (ip46->as_u64[0] == 0 && ip46->as_u64[1] == 0); +} + +always_inline u8 +ip46_address_is_equal (const ip46_address_t * ip46_1, + const ip46_address_t * ip46_2) +{ + return ((ip46_1->as_u64[0] == ip46_2->as_u64[0]) && + (ip46_1->as_u64[1] == ip46_2->as_u64[1])); +} + +static_always_inline int +ip4_address_is_equal (const ip4_address_t * ip4_1, + const ip4_address_t * ip4_2) +{ + return (ip4_1->as_u32 == ip4_2->as_u32); +} + +static_always_inline int +ip46_address_is_equal_v4 (const ip46_address_t * ip46, + const ip4_address_t * ip4) +{ + return (ip46->ip4.as_u32 == ip4->as_u32); +} + +static_always_inline int +ip46_address_is_equal_v6 (const ip46_address_t * ip46, + const ip6_address_t * ip6) +{ + return ((ip46->ip6.as_u64[0] == ip6->as_u64[0]) && + (ip46->ip6.as_u64[1] == ip6->as_u64[1])); +} + +static_always_inline void +ip46_address_copy (ip46_address_t * dst, const ip46_address_t * src) +{ + dst->as_u64[0] = src->as_u64[0]; + dst->as_u64[1] = src->as_u64[1]; +} + +static_always_inline void +ip46_address_set_ip6 (ip46_address_t * dst, const ip6_address_t * src) +{ + dst->as_u64[0] = src->as_u64[0]; + dst->as_u64[1] = src->as_u64[1]; +} + +always_inline ip46_address_t +to_ip46 (u32 is_ipv6, u8 * buf) +{ + ip46_address_t ip; + if (is_ipv6) + ip.ip6 = *((ip6_address_t *) buf); + else + ip46_address_set_ip4 (&ip, (ip4_address_t *) buf); + return ip; +} + +always_inline ip46_type_t +ip46_address_get_type (const ip46_address_t * ip) +{ + return (ip46_address_is_ip4 (ip) ? IP46_TYPE_IP4 : IP46_TYPE_IP6); +} + +always_inline uword +ip46_address_is_multicast (const ip46_address_t * a) +{ + return ip46_address_is_ip4 (a) ? ip4_address_is_multicast (&a->ip4) : + ip6_address_is_multicast (&a->ip6); +} + +extern void ip4_address_increment (ip4_address_t * i); +extern void ip6_address_increment (ip6_address_t * i); +extern void ip46_address_increment (ip46_type_t type, ip46_address_t * ip); + +#endif /* included_ip46_address_h */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip/ip4_forward.c b/src/vnet/ip/ip4_forward.c index bb036c8e969..aa554eab5fb 100644 --- a/src/vnet/ip/ip4_forward.c +++ b/src/vnet/ip/ip4_forward.c @@ -1930,349 +1930,6 @@ VLIB_CLI_COMMAND (show_ip_local, static) = }; /* *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, diff --git a/src/vnet/ip/ip6.h b/src/vnet/ip/ip6.h index 28b1af749f6..575c6a0eec5 100644 --- a/src/vnet/ip/ip6.h +++ b/src/vnet/ip/ip6.h @@ -44,6 +44,7 @@ #include <vnet/ethernet/packet.h> #include <vnet/ethernet/mac_address.h> #include <vnet/ip/ip6_packet.h> +#include <vnet/ip/ip46_address.h> #include <vnet/ip/ip6_hop_by_hop_packet.h> #include <vnet/ip/lookup.h> #include <stdbool.h> @@ -217,9 +218,6 @@ typedef struct ip6_main_t /** Functions to call when interface to table biding changes. */ ip6_table_bind_callback_t *table_bind_callbacks; - /* Template used to generate IP6 neighbor solicitation packets. */ - vlib_packet_template_t discover_neighbor_packet_template; - /* ip6 lookup table config parameters */ u32 lookup_table_nbuckets; uword lookup_table_size; @@ -302,42 +300,10 @@ ip6_unaligned_destination_matches_route (ip6_main_t * im, return 1; } -extern int ip6_get_ll_address (u32 sw_if_index, ip6_address_t * addr); - -always_inline int -ip6_src_address_for_packet (ip_lookup_main_t * lm, - u32 sw_if_index, - const ip6_address_t * dst, ip6_address_t * src) -{ - if (ip6_address_is_link_local_unicast (dst)) - { - return ip6_get_ll_address (sw_if_index, src); - } - else - { - u32 if_add_index = - lm->if_address_pool_index_by_sw_if_index[sw_if_index]; - if (PREDICT_TRUE (if_add_index != ~0)) - { - ip_interface_address_t *if_add = - pool_elt_at_index (lm->if_address_pool, if_add_index); - ip6_address_t *if_ip = - ip_interface_address_get_address (lm, if_add); - *src = *if_ip; - return (!0); - } - } - - src->as_u64[0] = 0; - src->as_u64[1] = 0; - - return (0); -} - /* Find interface address which matches destination. */ always_inline ip6_address_t * ip6_interface_address_matching_destination (ip6_main_t * im, - ip6_address_t * dst, + const ip6_address_t * dst, u32 sw_if_index, ip_interface_address_t ** result_ia) @@ -376,9 +342,6 @@ ip6_address_t *ip6_interface_first_address (ip6_main_t * im, u32 sw_if_index); int ip6_address_compare (ip6_address_t * a1, ip6_address_t * a2); -clib_error_t *ip6_probe_neighbor (vlib_main_t * vm, ip6_address_t * dst, - u32 sw_if_index, u8 refresh); - uword ip6_udp_register_listener (vlib_main_t * vm, u16 dst_port, u32 next_node_index); @@ -393,74 +356,9 @@ void ip6_local_hop_by_hop_register_protocol (u32 protocol, u32 node_index); serialize_function_t serialize_vnet_ip6_main, unserialize_vnet_ip6_main; -void ip6_ethernet_update_adjacency (vnet_main_t * vnm, - u32 sw_if_index, u32 ai); - -always_inline void -ip6_link_local_address_from_ethernet_mac_address (ip6_address_t * ip, - u8 * mac) -{ - ip->as_u64[0] = clib_host_to_net_u64 (0xFE80000000000000ULL); - /* Invert the "u" bit */ - ip->as_u8[8] = mac[0] ^ (1 << 1); - ip->as_u8[9] = mac[1]; - ip->as_u8[10] = mac[2]; - ip->as_u8[11] = 0xFF; - ip->as_u8[12] = 0xFE; - ip->as_u8[13] = mac[3]; - ip->as_u8[14] = mac[4]; - ip->as_u8[15] = mac[5]; -} - -always_inline void -ip6_ethernet_mac_address_from_link_local_address (u8 * mac, - ip6_address_t * ip) -{ - /* Invert the previously inverted "u" bit */ - mac[0] = ip->as_u8[8] ^ (1 << 1); - mac[1] = ip->as_u8[9]; - mac[2] = ip->as_u8[10]; - mac[3] = ip->as_u8[13]; - mac[4] = ip->as_u8[14]; - mac[5] = ip->as_u8[15]; -} - int vnet_set_ip6_flow_hash (u32 table_id, flow_hash_config_t flow_hash_config); -clib_error_t *enable_ip6_interface (vlib_main_t * vm, u32 sw_if_index); - -clib_error_t *disable_ip6_interface (vlib_main_t * vm, u32 sw_if_index); - -int ip6_interface_enabled (vlib_main_t * vm, u32 sw_if_index); - -clib_error_t *set_ip6_link_local_address (vlib_main_t * vm, - u32 sw_if_index, - ip6_address_t * address); - -typedef int (*ip6_nd_change_event_cb_t) (u32 pool_index, - const mac_address_t * new_mac, - u32 sw_if_index, - const ip6_address_t * address); - -int vnet_add_del_ip6_nd_change_event (vnet_main_t * vnm, - ip6_nd_change_event_cb_t data_callback, - u32 pid, - void *address_arg, - uword node_index, - uword type_opaque, - uword data, int is_add); - -int vnet_ip6_nd_term (vlib_main_t * vm, - vlib_node_runtime_t * node, - vlib_buffer_t * p0, - ethernet_header_t * eth, - ip6_header_t * ip, u32 sw_if_index, u16 bd_index); - -void send_ip6_na (vlib_main_t * vm, u32 sw_if_index); -void send_ip6_na_w_addr (vlib_main_t * vm, - const ip6_address_t * addr, u32 sw_if_index); - u8 *format_ip6_forward_next_trace (u8 * s, va_list * args); u32 ip6_tcp_udp_icmp_validate_checksum (vlib_main_t * vm, vlib_buffer_t * p0); diff --git a/src/vnet/ip/ip6_forward.c b/src/vnet/ip/ip6_forward.c index 128204f2c2e..4cb7608d0b4 100644 --- a/src/vnet/ip/ip6_forward.c +++ b/src/vnet/ip/ip6_forward.c @@ -40,7 +40,7 @@ #include <vnet/vnet.h> #include <vnet/ip/ip.h> #include <vnet/ip/ip_frag.h> -#include <vnet/ip/ip6_neighbor.h> +#include <vnet/ip/ip6_link.h> #include <vnet/ethernet/ethernet.h> /* for ethernet_header_t */ #include <vnet/srp/srp.h> /* for srp_hw_interface_class */ #include <vppinfra/cache.h> @@ -297,7 +297,7 @@ ip6_add_del_interface_address (vlib_main_t * vm, clib_error_t *error; u32 if_address_index; ip6_address_fib_t ip6_af, *addr_fib = 0; - ip6_address_t ll_addr; + const ip6_address_t *ll_addr; /* local0 interface doesn't support IP addressing */ if (sw_if_index == 0) @@ -317,13 +317,20 @@ ip6_add_del_interface_address (vlib_main_t * vm, } if (!is_del) { - return ip6_neighbor_set_link_local_address (vm, sw_if_index, - address); + int rv; + + rv = ip6_set_link_local_address (sw_if_index, address); + + if (rv) + { + vnm->api_errno = rv; + return clib_error_create ("address not assignable"); + } } else { - ll_addr = ip6_neighbor_get_link_local_address (sw_if_index); - if (ip6_address_is_equal (&ll_addr, address)) + ll_addr = ip6_get_link_local_address (sw_if_index); + if (ip6_address_is_equal (ll_addr, address)) { vnm->api_errno = VNET_API_ERROR_ADDRESS_NOT_DELETABLE; return clib_error_create ("address not deletable"); @@ -408,6 +415,8 @@ ip6_add_del_interface_address (vlib_main_t * vm, } ip6_sw_interface_enable_disable (sw_if_index, !is_del); + if (!is_del) + ip6_link_enable (sw_if_index); /* intf addr routes are added/deleted on admin up/down */ if (vnet_sw_interface_is_admin_up (vnm, sw_if_index)) @@ -428,6 +437,8 @@ ip6_add_del_interface_address (vlib_main_t * vm, cb->function (im, cb->function_opaque, sw_if_index, address, address_length, if_address_index, is_del); } + if (is_del) + ip6_link_disable (sw_if_index); done: vec_free (addr_fib); @@ -620,7 +631,6 @@ ip6_sw_interface_add_del (vnet_main_t * vnm, u32 sw_if_index, u32 is_add) ip6_address_t *address; vlib_main_t *vm = vlib_get_main (); - ip6_neighbor_sw_interface_add_del (vnm, sw_if_index, 0 /* is_add */ ); vnet_sw_interface_update_unnumbered (sw_if_index, ~0, 0); /* *INDENT-OFF* */ foreach_ip_interface_address (lm6, ia, sw_if_index, 0, @@ -1564,9 +1574,7 @@ VNET_FEATURE_INIT (ip6_local_end_of_arc, static) = { #ifdef CLIB_MARCH_VARIANT extern vlib_node_registration_t ip6_local_node; - #else - void ip6_register_protocol (u32 protocol, u32 node_index) { @@ -1588,114 +1596,6 @@ ip6_unregister_protocol (u32 protocol) ASSERT (protocol < ARRAY_LEN (lm->local_next_by_ip_protocol)); lm->local_next_by_ip_protocol[protocol] = IP_LOCAL_NEXT_PUNT; } - -clib_error_t * -ip6_probe_neighbor (vlib_main_t * vm, ip6_address_t * dst, u32 sw_if_index, - u8 refresh) -{ - vnet_main_t *vnm = vnet_get_main (); - ip6_main_t *im = &ip6_main; - icmp6_neighbor_solicitation_header_t *h; - ip6_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; - int bogus_length; - - 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_ip6_address, dst, - format_vnet_sw_if_index_name, vnm, - sw_if_index); - } - - src = - ip6_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_ip6_address, dst, - format_vnet_sw_if_index_name, vnm, sw_if_index); - } - - h = - vlib_packet_template_get_packet (vm, - &im->discover_neighbor_packet_template, - &bi); - if (!h) - return clib_error_return (0, "ICMP6 NS packet allocation failed"); - - hi = vnet_get_sup_hw_interface (vnm, sw_if_index); - - /* Destination address is a solicited node multicast address. We need to fill in - the low 24 bits with low 24 bits of target's address. */ - h->ip.dst_address.as_u8[13] = dst->as_u8[13]; - h->ip.dst_address.as_u8[14] = dst->as_u8[14]; - h->ip.dst_address.as_u8[15] = dst->as_u8[15]; - - h->ip.src_address = src[0]; - h->neighbor.target_address = dst[0]; - - if (PREDICT_FALSE (!hi->hw_address)) - { - return clib_error_return (0, "%U: interface %U do not support ip probe", - format_ip6_address, dst, - format_vnet_sw_if_index_name, vnm, - sw_if_index); - } - - clib_memcpy_fast (h->link_layer_option.ethernet_address, hi->hw_address, - vec_len (hi->hw_address)); - - h->neighbor.icmp.checksum = - ip6_tcp_udp_icmp_compute_checksum (vm, 0, &h->ip, &bogus_length); - ASSERT (bogus_length == 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; - - /* Add encapsulation string for software interface (e.g. ethernet header). */ - ip46_address_t nh = { - .ip6 = *dst, - }; - - ai = adj_nbr_add_or_lock (FIB_PROTOCOL_IP6, - VNET_LINK_IP6, &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 && refresh == 0) - { - adj_unlock (ai); - ai = adj_glean_add_or_lock (FIB_PROTOCOL_IP6, - VNET_LINK_IP6, sw_if_index, &nh); - adj = adj_get (ai); - } - - vnet_rewrite_one_header (adj[0], h, sizeof (ethernet_header_t)); - 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 @@ -2852,79 +2752,11 @@ ip6_lookup_init (vlib_main_t * vm) /* Unless explicitly configured, don't process HBH options */ im->hbh_enabled = 0; - { - icmp6_neighbor_solicitation_header_t p; - - clib_memset (&p, 0, sizeof (p)); - - p.ip.ip_version_traffic_class_and_flow_label = - clib_host_to_net_u32 (0x6 << 28); - p.ip.payload_length = - clib_host_to_net_u16 (sizeof (p) - - STRUCT_OFFSET_OF - (icmp6_neighbor_solicitation_header_t, neighbor)); - p.ip.protocol = IP_PROTOCOL_ICMP6; - p.ip.hop_limit = 255; - ip6_set_solicited_node_multicast_address (&p.ip.dst_address, 0); - - p.neighbor.icmp.type = ICMP6_neighbor_solicitation; - - p.link_layer_option.header.type = - ICMP6_NEIGHBOR_DISCOVERY_OPTION_source_link_layer_address; - p.link_layer_option.header.n_data_u64s = - sizeof (p.link_layer_option) / sizeof (u64); - - vlib_packet_template_init (vm, - &im->discover_neighbor_packet_template, - &p, sizeof (p), - /* alloc chunk size */ 8, - "ip6 neighbor discovery"); - } - return error; } VLIB_INIT_FUNCTION (ip6_lookup_init); -static clib_error_t * -test_ip6_link_command_fn (vlib_main_t * vm, - unformat_input_t * input, vlib_cli_command_t * cmd) -{ - u8 mac[6]; - ip6_address_t _a, *a = &_a; - - if (unformat (input, "%U", unformat_ethernet_address, mac)) - { - ip6_link_local_address_from_ethernet_mac_address (a, mac); - vlib_cli_output (vm, "Link local address: %U", format_ip6_address, a); - ip6_ethernet_mac_address_from_link_local_address (mac, a); - vlib_cli_output (vm, "Original MAC address: %U", - format_ethernet_address, mac); - } - - return 0; -} - -/*? - * This command converts the given MAC Address into an IPv6 link-local - * address. - * - * @cliexpar - * Example of how to create an IPv6 link-local address: - * @cliexstart{test ip6 link 16:d9:e0:91:79:86} - * Link local address: fe80::14d9:e0ff:fe91:7986 - * Original MAC address: 16:d9:e0:91:79:86 - * @cliexend -?*/ -/* *INDENT-OFF* */ -VLIB_CLI_COMMAND (test_link_command, static) = -{ - .path = "test ip6 link", - .function = test_ip6_link_command_fn, - .short_help = "test ip6 link <mac-address>", -}; -/* *INDENT-ON* */ - #ifndef CLIB_MARCH_VARIANT int vnet_set_ip6_flow_hash (u32 table_id, u32 flow_hash_config) diff --git a/src/vnet/ip/ip6_link.c b/src/vnet/ip/ip6_link.c new file mode 100644 index 00000000000..b73fbe2125f --- /dev/null +++ b/src/vnet/ip/ip6_link.c @@ -0,0 +1,873 @@ +/* + * Copyright (c) 2019 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/ip/ip6_link.h> +#include <vnet/ip/ip6_ll_table.h> + +#include <vnet/ethernet/ethernet.h> +#include <vnet/mfib/ip6_mfib.h> +#include <vnet/adj/adj_mcast.h> + +typedef struct ip6_link_delegate_t_ +{ + u32 ild_sw_if_index; + ip6_link_delegate_id_t ild_type; + index_t ild_index; +} ip6_link_delegate_t; + +const static ip6_link_delegate_t ip6_link_delegate_uninit = { + .ild_sw_if_index = ~0, +}; + +typedef struct ip6_link_t_ +{ + /** interface ip6 is enabled on */ + u32 il_sw_if_index; + + /** link-local address - if unset that IP6 is disabled*/ + ip6_address_t il_ll_addr; + + /** list of delegates */ + ip6_link_delegate_t *il_delegates; + + /** multicast adjacency for this link */ + adj_index_t il_mcast_adj; + + /** number of references to IP6 enabled on this link */ + u32 il_locks; +} ip6_link_t; + +#define FOREACH_IP6_LINK_DELEGATE(_ild, _il, body) \ +{ \ + if (NULL != _il) { \ + vec_foreach (_ild, _il->il_delegates) { \ + if (ip6_link_delegate_is_init(_ild)) \ + body; \ + } \ + } \ +} + +#define FOREACH_IP6_LINK_DELEGATE_ID(_id) \ + for (_id = 0; _id < il_delegate_id; _id++) + +/** last used delegate ID */ +static ip6_link_delegate_id_t il_delegate_id; + +/** VFT registered per-delegate type */ +static ip6_link_delegate_vft_t *il_delegate_vfts; + +/** Per interface configs */ +static ip6_link_t *ip6_links; + +/** Randomizer */ +static u64 il_randomizer; + +/** Logging */ +static vlib_log_class_t ip6_link_logger; + +#define IP6_LINK_DBG(...) \ + vlib_log_debug (ip6_link_logger, __VA_ARGS__); + +#define IP6_LINK_INFO(...) \ + vlib_log_notice (ip6_link_logger, __VA_ARGS__); + +static bool +ip6_link_delegate_is_init (const ip6_link_delegate_t * ild) +{ + return (~0 != ild->ild_sw_if_index); +} + +static bool +ip6_link_is_enabled_i (const ip6_link_t * il) +{ + return (!ip6_address_is_zero (&il->il_ll_addr)); +} + +static void +ip6_link_local_address_from_mac (ip6_address_t * ip, const u8 * mac) +{ + ip->as_u64[0] = clib_host_to_net_u64 (0xFE80000000000000ULL); + /* Invert the "u" bit */ + ip->as_u8[8] = mac[0] ^ (1 << 1); + ip->as_u8[9] = mac[1]; + ip->as_u8[10] = mac[2]; + ip->as_u8[11] = 0xFF; + ip->as_u8[12] = 0xFE; + ip->as_u8[13] = mac[3]; + ip->as_u8[14] = mac[4]; + ip->as_u8[15] = mac[5]; +} + +static void +ip6_mac_address_from_link_local (u8 * mac, const ip6_address_t * ip) +{ + /* Invert the previously inverted "u" bit */ + mac[0] = ip->as_u8[8] ^ (1 << 1); + mac[1] = ip->as_u8[9]; + mac[2] = ip->as_u8[10]; + mac[3] = ip->as_u8[13]; + mac[4] = ip->as_u8[14]; + mac[5] = ip->as_u8[15]; +} + +static ip6_link_t * +ip6_link_get (u32 sw_if_index) +{ + ip6_link_t *il; + + if (sw_if_index >= vec_len (ip6_links)) + return (NULL); + + il = &ip6_links[sw_if_index]; + + if (!ip6_link_is_enabled_i (il)) + return (NULL); + + return (il); +} + +bool +ip6_link_is_enabled (u32 sw_if_index) +{ + return (NULL != ip6_link_get (sw_if_index)); +} + + +int +ip6_link_enable (u32 sw_if_index) +{ + ip6_link_t *il; + int rv; + + il = ip6_link_get (sw_if_index); + + if (NULL == il) + { + const vnet_sw_interface_t *sw, *sw_sup; + const ethernet_interface_t *eth; + vnet_main_t *vnm; + + vnm = vnet_get_main (); + + IP6_LINK_INFO ("enable: %U", + format_vnet_sw_if_index_name, vnm, sw_if_index); + + sw_sup = vnet_get_sup_sw_interface (vnm, sw_if_index); + if (sw_sup->type != VNET_SW_INTERFACE_TYPE_HARDWARE) + { + rv = VNET_API_ERROR_UNSUPPORTED; + goto out; + } + + eth = ethernet_get_interface (ðernet_main, sw_sup->hw_if_index); + + if (NULL == eth) + { + rv = VNET_API_ERROR_UNSUPPORTED; + goto out; + } + + vec_validate (ip6_links, sw_if_index); + + il = &ip6_links[sw_if_index]; + il->il_locks = 1; + il->il_sw_if_index = sw_if_index; + + sw = vnet_get_sup_sw_interface (vnm, sw_if_index); + + if (sw->type == VNET_SW_INTERFACE_TYPE_SUB || + sw->type == VNET_SW_INTERFACE_TYPE_PIPE || + sw->type == VNET_SW_INTERFACE_TYPE_P2P) + { + il->il_ll_addr.as_u64[0] = + clib_host_to_net_u64 (0xFE80000000000000ULL); + + /* make up an interface id */ + il->il_ll_addr.as_u64[1] = random_u64 (&il_randomizer); + + /* clear u bit */ + il->il_ll_addr.as_u8[8] &= 0xfd; + } + else + { + ip6_link_local_address_from_mac (&il->il_ll_addr, eth->address); + } + + { + ip6_ll_prefix_t ilp = { + .ilp_addr = il->il_ll_addr, + .ilp_sw_if_index = sw_if_index, + }; + + ip6_ll_table_entry_update (&ilp, FIB_ROUTE_PATH_LOCAL); + } + + /* essentially "enables" ipv6 on this interface */ + ip6_mfib_interface_enable_disable (sw_if_index, 1); + ip6_sw_interface_enable_disable (sw_if_index, 1); + + il->il_mcast_adj = adj_mcast_add_or_lock (FIB_PROTOCOL_IP6, + VNET_LINK_IP6, sw_if_index); + + /* inform all register clients */ + ip6_link_delegate_id_t id; + FOREACH_IP6_LINK_DELEGATE_ID (id) + { + if (NULL != il_delegate_vfts[id].ildv_enable) + il_delegate_vfts[id].ildv_enable (il->il_sw_if_index); + } + + rv = 0; + } + else + { + rv = VNET_API_ERROR_VALUE_EXIST; + } + +out: + return (rv); +} + +static void +ip6_link_delegate_flush (ip6_link_t * il) +{ + ip6_link_delegate_t *ild; + + /* *INDET-OFF* */ + FOREACH_IP6_LINK_DELEGATE (ild, il, ( + { + il_delegate_vfts[ild-> + ild_type].ildv_disable + (ild->ild_index); + })); + /* *INDET-ON* */ + + vec_free (il->il_delegates); + il->il_delegates = NULL; +} + +static void +ip6_link_last_lock_gone (ip6_link_t * il) +{ + ip6_ll_prefix_t ilp = { + .ilp_addr = il->il_ll_addr, + .ilp_sw_if_index = il->il_sw_if_index, + }; + + IP6_LINK_INFO ("last-lock: %U", + format_vnet_sw_if_index_name, + vnet_get_main (), il->il_sw_if_index); + + ip6_link_delegate_flush (il); + ip6_ll_table_entry_delete (&ilp); + + ip6_mfib_interface_enable_disable (il->il_sw_if_index, 0); + ip6_sw_interface_enable_disable (il->il_sw_if_index, 0); + + ip6_address_set_zero (&il->il_ll_addr); + adj_unlock (il->il_mcast_adj); + il->il_mcast_adj = ADJ_INDEX_INVALID; +} + +static void +ip6_link_unlock (ip6_link_t * il) +{ + if (NULL == il) + return; + + il->il_locks--; + + if (0 == il->il_locks) + ip6_link_last_lock_gone (il); +} + +int +ip6_link_disable (u32 sw_if_index) +{ + ip6_link_t *il; + + il = ip6_link_get (sw_if_index); + + if (NULL == il) + return (VNET_API_ERROR_IP6_NOT_ENABLED); + + IP6_LINK_INFO ("disable: %U", + format_vnet_sw_if_index_name, vnet_get_main (), sw_if_index); + + ip6_link_unlock (il); + + return (0); +} + +const ip6_address_t * +ip6_get_link_local_address (u32 sw_if_index) +{ + const ip6_link_t *il; + + vec_validate (ip6_links, sw_if_index); + + il = &ip6_links[sw_if_index]; + + return (&il->il_ll_addr); +} + +adj_index_t +ip6_link_get_mcast_adj (u32 sw_if_index) +{ + const ip6_link_t *il; + + il = &ip6_links[sw_if_index]; + + return (il->il_mcast_adj); +} + +int +ip6_src_address_for_packet (u32 sw_if_index, + const ip6_address_t * dst, ip6_address_t * src) +{ + ip_lookup_main_t *lm; + + lm = &ip6_main.lookup_main; + + if (ip6_address_is_link_local_unicast (dst)) + { + ip6_address_copy (src, ip6_get_link_local_address (sw_if_index)); + + return (!0); + } + else + { + u32 if_add_index = + lm->if_address_pool_index_by_sw_if_index[sw_if_index]; + if (PREDICT_TRUE (if_add_index != ~0)) + { + ip_interface_address_t *if_add = + pool_elt_at_index (lm->if_address_pool, if_add_index); + ip6_address_t *if_ip = + ip_interface_address_get_address (lm, if_add); + *src = *if_ip; + return (!0); + } + } + + src->as_u64[0] = 0; + src->as_u64[1] = 0; + + return (0); +} + +int +ip6_set_link_local_address (u32 sw_if_index, const ip6_address_t * address) +{ + ip6_link_delegate_t *ild; + ip6_link_t *il; + + il = ip6_link_get (sw_if_index); + + if (NULL == il) + return (VNET_API_ERROR_IP6_NOT_ENABLED); + + ip6_ll_prefix_t ilp = { + .ilp_addr = il->il_ll_addr, + .ilp_sw_if_index = sw_if_index, + }; + + IP6_LINK_INFO ("set-ll: %U -> %U", + format_vnet_sw_if_index_name, vnet_get_main (), sw_if_index, + format_ip6_address, address); + + ip6_ll_table_entry_delete (&ilp); + ip6_address_copy (&il->il_ll_addr, address); + ip6_address_copy (&ilp.ilp_addr, address); + ip6_ll_table_entry_update (&ilp, FIB_ROUTE_PATH_LOCAL); + + /* *INDENT-OFF* */ + FOREACH_IP6_LINK_DELEGATE (ild, il, + ({ + if (NULL != il_delegate_vfts[ild->ild_type].ildv_ll_change) + il_delegate_vfts[ild->ild_type].ildv_ll_change(ild->ild_index, + &il->il_ll_addr); + })); + /* *INDENT-ON* */ + + return (0); +} + +ip6_link_delegate_id_t +ip6_link_delegate_register (const ip6_link_delegate_vft_t * vft) +{ + ip6_link_delegate_id_t rc = il_delegate_id++; + + ASSERT (vft->ildv_disable); + + vec_validate (il_delegate_vfts, rc); + + il_delegate_vfts[rc] = *vft; + + return (rc); +} + +index_t +ip6_link_delegate_get (u32 sw_if_index, ip6_link_delegate_id_t id) +{ + ip6_link_t *il; + + il = ip6_link_get (sw_if_index); + + if (NULL == il) + return (INDEX_INVALID); + + vec_validate_init_empty (il->il_delegates, id, ip6_link_delegate_uninit); + + if (!ip6_link_delegate_is_init (&il->il_delegates[id])) + return (INDEX_INVALID); + + return (il->il_delegates[id].ild_index); +} + +bool +ip6_link_delegate_update (u32 sw_if_index, + ip6_link_delegate_id_t id, index_t ii) +{ + ip6_link_t *il; + + il = ip6_link_get (sw_if_index); + + if (NULL == il) + return (false); + + vec_validate_init_empty (il->il_delegates, id, ip6_link_delegate_uninit); + + il->il_delegates[id].ild_sw_if_index = sw_if_index; + il->il_delegates[id].ild_type = id; + il->il_delegates[id].ild_index = ii; + + return (true); +} + +void +ip6_link_delegate_remove (u32 sw_if_index, + ip6_link_delegate_id_t id, index_t ii) +{ + ip6_link_t *il; + + il = ip6_link_get (sw_if_index); + + if (NULL != il) + { + if (vec_len (il->il_delegates) > id) + { + clib_memcpy (&il->il_delegates[id], + &ip6_link_delegate_uninit, + sizeof (il->il_delegates[0])); + } + } +} + +static void +ip6_link_add_del_address (ip6_main_t * im, + uword opaque, + u32 sw_if_index, + ip6_address_t * address, + u32 address_length, + u32 if_address_index, u32 is_delete) +{ + const ip6_link_delegate_t *ild; + ip6_link_t *il; + + if (ip6_address_is_link_local_unicast (address)) + // only interested in global addresses here + return; + + IP6_LINK_INFO ("addr-%s: %U -> %U", + (is_delete ? "del" : "add"), + format_vnet_sw_if_index_name, vnet_get_main (), sw_if_index, + format_ip6_address, address); + + il = ip6_link_get (sw_if_index); + + if (NULL == il) + return; + + /* *INDENT-OFF* */ + FOREACH_IP6_LINK_DELEGATE (ild, il, + ({ + if (is_delete) + { + if (NULL != il_delegate_vfts[ild->ild_type].ildv_addr_del) + il_delegate_vfts[ild->ild_type].ildv_addr_del(ild->ild_index, + address, address_length); + } + else + { + if (NULL != il_delegate_vfts[ild->ild_type].ildv_addr_add) + il_delegate_vfts[ild->ild_type].ildv_addr_add(ild->ild_index, + address, address_length); + } + })); + /* *INDENT-ON* */ +} + +static clib_error_t * +ip6_link_interface_add_del (vnet_main_t * vnm, u32 sw_if_index, u32 is_add) +{ + if (!is_add) + { + ip6_link_t *il; + + il = ip6_link_get (sw_if_index); + + IP6_LINK_DBG ("link-del: %U", + format_vnet_sw_if_index_name, vnet_get_main (), + sw_if_index); + + if (NULL != il) + /* force cleanup */ + ip6_link_last_lock_gone (il); + } + + return (NULL); +} + +VNET_SW_INTERFACE_ADD_DEL_FUNCTION (ip6_link_interface_add_del); + +static clib_error_t * +ip6_link_init (vlib_main_t * vm) +{ + il_randomizer = clib_cpu_time_now (); + ip6_link_logger = vlib_log_register_class ("ip6", "link"); + + { + ip6_add_del_interface_address_callback_t cb = { + .function = ip6_link_add_del_address, + }; + vec_add1 (ip6_main.add_del_interface_address_callbacks, cb); + } + return (NULL); +} + +VLIB_INIT_FUNCTION (ip6_link_init); + + +static clib_error_t * +test_ip6_link_command_fn (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) +{ + u8 mac[6]; + ip6_address_t _a, *a = &_a; + + if (unformat (input, "%U", unformat_ethernet_address, mac)) + { + ip6_link_local_address_from_mac (a, mac); + vlib_cli_output (vm, "Link local address: %U", format_ip6_address, a); + ip6_mac_address_from_link_local (mac, a); + vlib_cli_output (vm, "Original MAC address: %U", + format_ethernet_address, mac); + } + + return 0; +} + +/*? + * This command converts the given MAC Address into an IPv6 link-local + * address. + * + * @cliexpar + * Example of how to create an IPv6 link-local address: + * @cliexstart{test ip6 link 16:d9:e0:91:79:86} + * Link local address: fe80::14d9:e0ff:fe91:7986 + * Original MAC address: 16:d9:e0:91:79:86 + * @cliexend +?*/ +/* *INDENT-OFF* */ +VLIB_CLI_COMMAND (test_link_command, static) = +{ + .path = "test ip6 link", + .function = test_ip6_link_command_fn, + .short_help = "test ip6 link <mac-address>", +}; +/* *INDENT-ON* */ + +static u8 * +ip6_print_addrs (u8 * s, u32 * addrs) +{ + ip_lookup_main_t *lm = &ip6_main.lookup_main; + u32 i; + + for (i = 0; i < vec_len (addrs); i++) + { + ip_interface_address_t *a = + pool_elt_at_index (lm->if_address_pool, addrs[i]); + ip6_address_t *address = ip_interface_address_get_address (lm, a); + + s = format (s, "%U%U/%d\n", + format_white_space, 4, + format_ip6_address, address, a->address_length); + } + + return (s); +} + +static u8 * +format_ip6_link (u8 * s, va_list * arg) +{ + const ip6_link_t *il = va_arg (*arg, ip6_link_t *); + ip_lookup_main_t *lm = &ip6_main.lookup_main; + vnet_main_t *vnm = vnet_get_main (); + + if (!ip6_link_is_enabled_i (il)) + return (s); + + s = format (s, "%U is admin %s\n", + format_vnet_sw_interface_name, vnm, + vnet_get_sw_interface (vnm, il->il_sw_if_index), + (vnet_sw_interface_is_admin_up (vnm, il->il_sw_if_index) ? + "up" : "down")); + + u32 ai; + u32 *link_scope = 0, *global_scope = 0; + u32 *local_scope = 0, *unknown_scope = 0; + ip_interface_address_t *a; + const ip6_link_delegate_t *ild; + + vec_validate_init_empty (lm->if_address_pool_index_by_sw_if_index, + il->il_sw_if_index, ~0); + ai = lm->if_address_pool_index_by_sw_if_index[il->il_sw_if_index]; + + while (ai != (u32) ~ 0) + { + a = pool_elt_at_index (lm->if_address_pool, ai); + ip6_address_t *address = ip_interface_address_get_address (lm, a); + + if (ip6_address_is_link_local_unicast (address)) + vec_add1 (link_scope, ai); + else if (ip6_address_is_global_unicast (address)) + vec_add1 (global_scope, ai); + else if (ip6_address_is_local_unicast (address)) + vec_add1 (local_scope, ai); + else + vec_add1 (unknown_scope, ai); + + ai = a->next_this_sw_interface; + } + + if (vec_len (link_scope)) + { + s = format (s, "%ULink-local address(es):\n", format_white_space, 2); + s = ip6_print_addrs (s, link_scope); + vec_free (link_scope); + } + + if (vec_len (local_scope)) + { + s = format (s, "%ULocal unicast address(es):\n", format_white_space, 2); + s = ip6_print_addrs (s, local_scope); + vec_free (local_scope); + } + + if (vec_len (global_scope)) + { + s = format (s, "%UGlobal unicast address(es):\n", + format_white_space, 2); + s = ip6_print_addrs (s, global_scope); + vec_free (global_scope); + } + + if (vec_len (unknown_scope)) + { + s = format (s, "%UOther-scope address(es):\n", format_white_space, 2); + s = ip6_print_addrs (s, unknown_scope); + vec_free (unknown_scope); + } + + s = format (s, "%ULink-local address(es):\n", format_white_space, 2); + s = format (s, "%U%U\n", + format_white_space, 4, format_ip6_address, &il->il_ll_addr); + + /* *INDENT-OFF* */ + FOREACH_IP6_LINK_DELEGATE(ild, il, + ({ + s = format (s, "%U", il_delegate_vfts[ild->ild_type].ildv_format, + ild->ild_index, 2); + })); + /* *INDENT-ON* */ + + return (s); +} + +static clib_error_t * +ip6_link_show (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) +{ + const ip6_link_t *il; + vnet_main_t *vnm; + u32 sw_if_index; + + vnm = vnet_get_main (); + sw_if_index = ~0; + + if (unformat_user (input, unformat_vnet_sw_interface, vnm, &sw_if_index)) + { + il = ip6_link_get (sw_if_index); + + if (NULL == il) + { + vlib_cli_output (vm, "IP6 disabled"); + return (NULL); + } + else + vlib_cli_output (vm, "%U", format_ip6_link, il); + } + else + { + vec_foreach (il, ip6_links) + vlib_cli_output (vm, "%U", format_ip6_link, il); + } + + return (NULL); +} + +/*? + * This command is used to display various IPv6 attributes on a given + * interface. + * + * @cliexpar + * Example of how to display IPv6 settings: + * @cliexstart{show ip6 interface GigabitEthernet2/0/0} + * GigabitEthernet2/0/0 is admin up + * Link-local address(es): + * fe80::ab8/64 + * Joined group address(es): + * ff02::1 + * ff02::2 + * ff02::16 + * ff02::1:ff00:ab8 + * Advertised Prefixes: + * prefix fe80::fe:28ff:fe9c:75b3, length 64 + * MTU is 1500 + * ICMP error messages are unlimited + * ICMP redirects are disabled + * ICMP unreachables are not sent + * ND DAD is disabled + * ND advertised reachable time is 0 + * ND advertised retransmit interval is 0 (msec) + * ND router advertisements are sent every 200 seconds (min interval is 150) + * ND router advertisements live for 600 seconds + * Hosts use stateless autoconfig for addresses + * ND router advertisements sent 19336 + * ND router solicitations received 0 + * ND router solicitations dropped 0 + * @cliexend + * Example of output if IPv6 is not enabled on the interface: + * @cliexstart{show ip6 interface GigabitEthernet2/0/0} + * show ip6 interface: IPv6 not enabled on interface + * @cliexend +?*/ +/* *INDENT-OFF* */ +VLIB_CLI_COMMAND (ip6_link_show_command, static) = +{ + .path = "show ip6 interface", + .function = ip6_link_show, + .short_help = "show ip6 interface <interface>", +}; +/* *INDENT-ON* */ + +static clib_error_t * +enable_ip6_interface_cmd (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) +{ + vnet_main_t *vnm = vnet_get_main (); + clib_error_t *error = NULL; + u32 sw_if_index; + + sw_if_index = ~0; + + if (unformat_user (input, unformat_vnet_sw_interface, vnm, &sw_if_index)) + { + if (ip6_link_enable (sw_if_index)) + error = clib_error_return (0, "Failed\n"); + } + else + { + error = clib_error_return (0, "unknown interface\n'", + format_unformat_error, input); + + } + return error; +} + +/*? + * This command is used to enable IPv6 on a given interface. + * + * @cliexpar + * Example of how enable IPv6 on a given interface: + * @cliexcmd{enable ip6 interface GigabitEthernet2/0/0} +?*/ +/* *INDENT-OFF* */ +VLIB_CLI_COMMAND (enable_ip6_interface_command, static) = +{ + .path = "enable ip6 interface", + .function = enable_ip6_interface_cmd, + .short_help = "enable ip6 interface <interface>", +}; +/* *INDENT-ON* */ + +static clib_error_t * +disable_ip6_interface_cmd (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) +{ + vnet_main_t *vnm = vnet_get_main (); + clib_error_t *error = NULL; + u32 sw_if_index; + + sw_if_index = ~0; + + if (unformat_user (input, unformat_vnet_sw_interface, vnm, &sw_if_index)) + { + if (ip6_link_disable (sw_if_index)) + error = clib_error_return (0, "Failed\n"); + } + else + { + error = clib_error_return (0, "unknown interface\n'", + format_unformat_error, input); + + } + return error; +} + +/*? + * This command is used to disable IPv6 on a given interface. + * + * @cliexpar + * Example of how disable IPv6 on a given interface: + * @cliexcmd{disable ip6 interface GigabitEthernet2/0/0} +?*/ +/* *INDENT-OFF* */ +VLIB_CLI_COMMAND (disable_ip6_interface_command, static) = +{ + .path = "disable ip6 interface", + .function = disable_ip6_interface_cmd, + .short_help = "disable ip6 interface <interface>", +}; +/* *INDENT-ON* */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip/ip6_link.h b/src/vnet/ip/ip6_link.h new file mode 100644 index 00000000000..01506bb4765 --- /dev/null +++ b/src/vnet/ip/ip6_link.h @@ -0,0 +1,82 @@ +/* + * Copyright (c) 2019 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/ip/ip.h> + +/** + * IPv6 Configuration on an interface + */ + +extern int ip6_link_enable (u32 sw_if_index); +extern bool ip6_link_is_enabled (u32 sw_if_index); +extern int ip6_link_disable (u32 sw_if_index); + +extern const ip6_address_t *ip6_get_link_local_address (u32 sw_if_index); + +extern int ip6_set_link_local_address (u32 sw_if_index, + const ip6_address_t * address); +extern adj_index_t ip6_link_get_mcast_adj (u32 sw_if_index); + +extern int +ip6_src_address_for_packet (u32 sw_if_index, + const ip6_address_t * dst, ip6_address_t * src); + +/** + * Delegates for the interfaces + * + * delegates are a means for a external component to 'extend' the config + * object by adding their own objects (aka delegates). + */ + +/** + * Callback functions for handling actions on the link + */ +typedef void (*ip6_link_enable_fn_t) (u32 sw_if_index); +typedef void (*ip6_link_disable_fn_t) (index_t ildi); +typedef void (*ip6_link_ll_change_fn_t) (u32 ildi, const ip6_address_t * a); +typedef void (*ip6_link_address_change_fn_t) (u32 ildi, + const ip6_address_t * a, + u8 address_oength); + +typedef struct ip6_link_delegate_vft_t_ +{ + ip6_link_enable_fn_t ildv_enable; + ip6_link_disable_fn_t ildv_disable; + ip6_link_ll_change_fn_t ildv_ll_change; + ip6_link_address_change_fn_t ildv_addr_add; + ip6_link_address_change_fn_t ildv_addr_del; + format_function_t *ildv_format; +} ip6_link_delegate_vft_t; + +typedef u32 ip6_link_delegate_id_t; + + +extern ip6_link_delegate_id_t ip6_link_delegate_register (const + ip6_link_delegate_vft_t + * vft); +extern index_t ip6_link_delegate_get (u32 sw_if_index, + ip6_link_delegate_id_t id); +extern bool ip6_link_delegate_update (u32 sw_if_index, + ip6_link_delegate_id_t id, index_t ii); +extern void ip6_link_delegate_remove (u32 sw_if_index, + ip6_link_delegate_id_t id, index_t ii); + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip/ip6_neighbor.c b/src/vnet/ip/ip6_neighbor.c deleted file mode 100644 index 5b8085d556d..00000000000 --- a/src/vnet/ip/ip6_neighbor.c +++ /dev/null @@ -1,5034 +0,0 @@ -/* - * ip/ip6_neighbor.c: IP6 neighbor handling - * - * Copyright (c) 2010 Cisco and/or its affiliates. - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at: - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -#include <vnet/ip/ip.h> -#include <vnet/ip/ip6_neighbor.h> -#include <vnet/ethernet/ethernet.h> -#include <vppinfra/mhash.h> -#include <vnet/adj/adj.h> -#include <vnet/adj/adj_mcast.h> -#include <vnet/fib/fib_table.h> -#include <vnet/fib/ip6_fib.h> -#include <vnet/mfib/ip6_mfib.h> -#include <vnet/ip/ip6_ll_table.h> -#include <vnet/l2/l2_input.h> -#include <vlibmemory/api.h> - -/** - * @file - * @brief IPv6 Neighbor Adjacency and Neighbor Discovery. - * - * The files contains the API and CLI code for managing IPv6 neighbor - * adjacency tables and neighbor discovery logic. - */ - -/* can't use sizeof link_layer_address, that's 8 */ -#define ETHER_MAC_ADDR_LEN 6 - -/* advertised prefix option */ -typedef struct -{ - /* basic advertised information */ - ip6_address_t prefix; - u8 prefix_len; - int adv_on_link_flag; - int adv_autonomous_flag; - u32 adv_valid_lifetime_in_secs; - u32 adv_pref_lifetime_in_secs; - - /* advertised values are computed from these times if decrementing */ - f64 valid_lifetime_expires; - f64 pref_lifetime_expires; - - /* local information */ - int enabled; - int deprecated_prefix_flag; - int decrement_lifetime_flag; - -#define MIN_ADV_VALID_LIFETIME 7203 /* seconds */ -#define DEF_ADV_VALID_LIFETIME 2592000 -#define DEF_ADV_PREF_LIFETIME 604800 - - /* extensions are added here, mobile, DNS etc.. */ -} ip6_radv_prefix_t; - - -typedef struct -{ - /* group information */ - u8 type; - ip6_address_t mcast_address; - u16 num_sources; - ip6_address_t *mcast_source_address_pool; -} ip6_mldp_group_t; - -/* configured router advertisement information per ipv6 interface */ -typedef struct -{ - - /* advertised config information, zero means unspecified */ - u8 curr_hop_limit; - int adv_managed_flag; - int adv_other_flag; - u16 adv_router_lifetime_in_sec; - u32 adv_neighbor_reachable_time_in_msec; - u32 adv_time_in_msec_between_retransmitted_neighbor_solicitations; - - /* mtu option */ - u32 adv_link_mtu; - - /* source link layer option */ - u8 link_layer_address[8]; - u8 link_layer_addr_len; - - /* prefix option */ - ip6_radv_prefix_t *adv_prefixes_pool; - - /* Hash table mapping address to index in interface advertised prefix pool. */ - mhash_t address_to_prefix_index; - - /* MLDP group information */ - ip6_mldp_group_t *mldp_group_pool; - - /* Hash table mapping address to index in mldp address pool. */ - mhash_t address_to_mldp_index; - - /* local information */ - u32 sw_if_index; - int send_radv; /* radv on/off on this interface - set by config */ - int cease_radv; /* we are ceasing to send - set byf config */ - int send_unicast; - int adv_link_layer_address; - int prefix_option; - int failed_device_check; - int all_routers_mcast; - u32 seed; - u64 randomizer; - int ref_count; - adj_index_t mcast_adj_index; - - /* timing information */ -#define DEF_MAX_RADV_INTERVAL 200 -#define DEF_MIN_RADV_INTERVAL .75 * DEF_MAX_RADV_INTERVAL -#define DEF_CURR_HOP_LIMIT 64 -#define DEF_DEF_RTR_LIFETIME 3 * DEF_MAX_RADV_INTERVAL -#define MAX_DEF_RTR_LIFETIME 9000 - -#define MAX_INITIAL_RTR_ADVERT_INTERVAL 16 /* seconds */ -#define MAX_INITIAL_RTR_ADVERTISEMENTS 3 /*transmissions */ -#define MIN_DELAY_BETWEEN_RAS 3 /* seconds */ -#define MAX_DELAY_BETWEEN_RAS 1800 /* seconds */ -#define MAX_RA_DELAY_TIME .5 /* seconds */ - - f64 max_radv_interval; - f64 min_radv_interval; - f64 min_delay_between_radv; - f64 max_delay_between_radv; - f64 max_rtr_default_lifetime; - - f64 last_radv_time; - f64 last_multicast_time; - f64 next_multicast_time; - - - u32 initial_adverts_count; - f64 initial_adverts_interval; - u32 initial_adverts_sent; - - /* stats */ - u32 n_advertisements_sent; - u32 n_solicitations_rcvd; - u32 n_solicitations_dropped; - - /* Link local address to use (defaults to underlying physical for logical interfaces */ - ip6_address_t link_local_address; - - /* router solicitations sending state */ - u8 keep_sending_rs; /* when true then next fields are valid */ - icmp6_send_router_solicitation_params_t params; - f64 sleep_interval; - f64 due_time; - u32 n_left; - f64 start_time; - vlib_buffer_t *buffer; -} ip6_radv_t; - -typedef struct -{ - u32 next_index; - uword node_index; - uword type_opaque; - uword data; - /* Used for nd event notification only */ - ip6_nd_change_event_cb_t data_callback; - u32 pid; -} pending_resolution_t; - - -typedef struct -{ - /* Hash tables mapping name to opcode. */ - uword *opcode_by_name; - - /* lite beer "glean" adjacency handling */ - mhash_t pending_resolutions_by_address; - pending_resolution_t *pending_resolutions; - - /* Mac address change notification */ - mhash_t mac_changes_by_address; - pending_resolution_t *mac_changes; - - u32 *neighbor_input_next_index_by_hw_if_index; - - ip6_neighbor_t *neighbor_pool; - - mhash_t neighbor_index_by_key; - - u32 *if_radv_pool_index_by_sw_if_index; - - ip6_radv_t *if_radv_pool; - - /* Neighbor attack mitigation */ - u32 limit_neighbor_cache_size; - u32 neighbor_delete_rotor; - - /* Wildcard nd report publisher */ - uword wc_ip6_nd_publisher_node; - uword wc_ip6_nd_publisher_et; - - /* Router advertisement report publisher */ - uword ip6_ra_publisher_node; - uword ip6_ra_publisher_et; -} ip6_neighbor_main_t; - -/* ipv6 neighbor discovery - timer/event types */ -typedef enum -{ - ICMP6_ND_EVENT_INIT, -} ip6_icmp_neighbor_discovery_event_type_t; - -typedef union -{ - u32 add_del_swindex; - struct - { - u32 up_down_swindex; - u32 fib_index; - } up_down_event; -} ip6_icmp_neighbor_discovery_event_data_t; - -static ip6_neighbor_main_t ip6_neighbor_main; -ip6_neighbor_public_main_t ip6_neighbor_public_main; -static ip6_address_t ip6a_zero; /* ip6 address 0 */ - -static void wc_nd_signal_report (wc_nd_report_t * r); -static void ra_signal_report (ra_report_t * r); - -ip6_address_t -ip6_neighbor_get_link_local_address (u32 sw_if_index) -{ - static ip6_address_t empty_address = { {0} }; - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_radv_t *radv_info; - u32 ri = ~0; - - if (vec_len (nm->if_radv_pool_index_by_sw_if_index) > sw_if_index) - ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - if (ri == ~0) - { - clib_warning ("IPv6 is not enabled for sw_if_index %d", sw_if_index); - return empty_address; - } - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - if (radv_info == NULL) - { - clib_warning ("Internal error"); - return empty_address; - } - return radv_info->link_local_address; -} - -/** - * @brief publish wildcard arp event - * @param sw_if_index The interface on which the ARP entries are acted - */ -static int -vnet_nd_wc_publish (u32 sw_if_index, - const mac_address_t * mac, const ip6_address_t * ip6) -{ - wc_nd_report_t r = { - .sw_if_index = sw_if_index, - .ip6 = *ip6, - .mac = *mac, - }; - - vl_api_rpc_call_main_thread (wc_nd_signal_report, (u8 *) & r, sizeof r); - return 0; -} - -static void -wc_nd_signal_report (wc_nd_report_t * r) -{ - vlib_main_t *vm = vlib_get_main (); - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - uword ni = nm->wc_ip6_nd_publisher_node; - uword et = nm->wc_ip6_nd_publisher_et; - - if (ni == (uword) ~ 0) - return; - wc_nd_report_t *q = - vlib_process_signal_event_data (vm, ni, et, 1, sizeof *q); - - *q = *r; -} - -void -wc_nd_set_publisher_node (uword node_index, uword event_type) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - nm->wc_ip6_nd_publisher_node = node_index; - nm->wc_ip6_nd_publisher_et = event_type; -} - -static int -ra_publish (ra_report_t * r) -{ - void vl_api_rpc_call_main_thread (void *fp, u8 * data, u32 data_length); - vl_api_rpc_call_main_thread (ra_signal_report, (u8 *) r, sizeof *r); - return 0; -} - -static void -ra_signal_report (ra_report_t * r) -{ - vlib_main_t *vm = vlib_get_main (); - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - uword ni = nm->ip6_ra_publisher_node; - uword et = nm->ip6_ra_publisher_et; - - if (ni == (uword) ~ 0) - return; - ra_report_t *q = vlib_process_signal_event_data (vm, ni, et, 1, sizeof *q); - - *q = *r; -} - -void -ra_set_publisher_node (uword node_index, uword event_type) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - nm->ip6_ra_publisher_node = node_index; - nm->ip6_ra_publisher_et = event_type; -} - -static u8 * -format_ip6_neighbor_ip6_entry (u8 * s, va_list * va) -{ - vlib_main_t *vm = va_arg (*va, vlib_main_t *); - ip6_neighbor_t *n = va_arg (*va, ip6_neighbor_t *); - vnet_main_t *vnm = vnet_get_main (); - vnet_sw_interface_t *si; - - if (!n) - return format (s, "%=12s%=45s%=6s%=20s%=40s", "Time", "Address", "Flags", - "Link layer", "Interface"); - - si = vnet_get_sw_interface (vnm, n->key.sw_if_index); - - return format (s, "%=12U%=45U%=6U%=20U%=40U", - format_vlib_time, vm, n->time_last_updated, - format_ip6_address, &n->key.ip6_address, - format_ip_neighbor_flags, n->flags, - format_mac_address_t, &n->mac, - format_vnet_sw_interface_name, vnm, si); -} - -static void -ip6_neighbor_adj_fib_remove (ip6_neighbor_t * n, u32 fib_index) -{ - if (FIB_NODE_INDEX_INVALID != n->fib_entry_index) - { - if (ip6_address_is_link_local_unicast (&n->key.ip6_address)) - { - ip6_ll_prefix_t pfx = { - .ilp_addr = n->key.ip6_address, - .ilp_sw_if_index = n->key.sw_if_index, - }; - ip6_ll_table_entry_delete (&pfx); - } - else - { - fib_prefix_t pfx = { - .fp_len = 128, - .fp_proto = FIB_PROTOCOL_IP6, - .fp_addr.ip6 = n->key.ip6_address, - }; - fib_table_entry_path_remove (fib_index, - &pfx, - FIB_SOURCE_ADJ, - DPO_PROTO_IP6, - &pfx.fp_addr, - n->key.sw_if_index, ~0, - 1, FIB_ROUTE_PATH_FLAG_NONE); - } - } -} - -typedef struct -{ - u8 is_add; - ip_neighbor_flags_t flags; - mac_address_t mac; - u32 sw_if_index; - ip6_address_t addr; -} ip6_neighbor_set_unset_rpc_args_t; - -static void ip6_neighbor_set_unset_rpc_callback - (ip6_neighbor_set_unset_rpc_args_t * a); - -static void set_unset_ip6_neighbor_rpc - (vlib_main_t * vm, - u32 sw_if_index, - const ip6_address_t * a, - const mac_address_t * mac, int is_add, ip_neighbor_flags_t flags) -{ - ip6_neighbor_set_unset_rpc_args_t args; - void vl_api_rpc_call_main_thread (void *fp, u8 * data, u32 data_length); - - args.sw_if_index = sw_if_index; - args.is_add = is_add; - args.flags = flags; - ip6_address_copy (&args.addr, a); - mac_address_copy (&args.mac, mac); - - vl_api_rpc_call_main_thread (ip6_neighbor_set_unset_rpc_callback, - (u8 *) & args, sizeof (args)); -} - -static void -ip6_nbr_probe (ip_adjacency_t * adj) -{ - icmp6_neighbor_solicitation_header_t *h; - vnet_main_t *vnm = vnet_get_main (); - ip6_main_t *im = &ip6_main; - ip_interface_address_t *ia; - ip6_address_t *dst, *src; - vnet_hw_interface_t *hi; - vnet_sw_interface_t *si; - vlib_buffer_t *b; - int bogus_length; - vlib_main_t *vm; - u32 bi = 0; - - vm = vlib_get_main (); - - si = vnet_get_sw_interface (vnm, adj->rewrite_header.sw_if_index); - dst = &adj->sub_type.nbr.next_hop.ip6; - - if (!(si->flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP)) - { - return; - } - src = ip6_interface_address_matching_destination (im, dst, - adj->rewrite_header. - sw_if_index, &ia); - if (!src) - { - return; - } - - h = vlib_packet_template_get_packet (vm, - &im->discover_neighbor_packet_template, - &bi); - if (!h) - return; - - hi = vnet_get_sup_hw_interface (vnm, adj->rewrite_header.sw_if_index); - - h->ip.dst_address.as_u8[13] = dst->as_u8[13]; - h->ip.dst_address.as_u8[14] = dst->as_u8[14]; - h->ip.dst_address.as_u8[15] = dst->as_u8[15]; - h->ip.src_address = src[0]; - h->neighbor.target_address = dst[0]; - - clib_memcpy (h->link_layer_option.ethernet_address, - hi->hw_address, vec_len (hi->hw_address)); - - h->neighbor.icmp.checksum = - ip6_tcp_udp_icmp_compute_checksum (vm, 0, &h->ip, &bogus_length); - ASSERT (bogus_length == 0); - - b = vlib_get_buffer (vm, bi); - vnet_buffer (b)->sw_if_index[VLIB_RX] = - vnet_buffer (b)->sw_if_index[VLIB_TX] = adj->rewrite_header.sw_if_index; - - /* Add encapsulation string for software interface (e.g. ethernet header). */ - vnet_rewrite_one_header (adj[0], h, sizeof (ethernet_header_t)); - 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); - } -} - -static void -ip6_nd_mk_complete (adj_index_t ai, ip6_neighbor_t * nbr) -{ - adj_nbr_update_rewrite (ai, ADJ_NBR_REWRITE_FLAG_COMPLETE, - ethernet_build_rewrite (vnet_get_main (), - nbr->key.sw_if_index, - adj_get_link_type (ai), - nbr->mac.bytes)); -} - -static void -ip6_nd_mk_incomplete (adj_index_t ai) -{ - ip_adjacency_t *adj = adj_get (ai); - - adj_nbr_update_rewrite (ai, - ADJ_NBR_REWRITE_FLAG_INCOMPLETE, - ethernet_build_rewrite (vnet_get_main (), - adj->rewrite_header. - sw_if_index, - adj_get_link_type (ai), - VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST)); -} - -#define IP6_NBR_MK_KEY(k, sw_if_index, addr) \ -{ \ - k.sw_if_index = sw_if_index; \ - k.ip6_address = *addr; \ - k.pad = 0; \ -} - -static ip6_neighbor_t * -ip6_nd_find (u32 sw_if_index, const ip6_address_t * addr) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_neighbor_t *n = NULL; - ip6_neighbor_key_t k; - uword *p; - - IP6_NBR_MK_KEY (k, sw_if_index, addr); - - p = mhash_get (&nm->neighbor_index_by_key, &k); - if (p) - { - n = pool_elt_at_index (nm->neighbor_pool, p[0]); - } - - return (n); -} - -static adj_walk_rc_t -ip6_nd_mk_complete_walk (adj_index_t ai, void *ctx) -{ - ip6_neighbor_t *nbr = ctx; - - ip6_nd_mk_complete (ai, nbr); - - return (ADJ_WALK_RC_CONTINUE); -} - -static adj_walk_rc_t -ip6_nd_mk_incomplete_walk (adj_index_t ai, void *ctx) -{ - ip6_nd_mk_incomplete (ai); - - return (ADJ_WALK_RC_CONTINUE); -} - -static clib_error_t * -ip6_neighbor_sw_interface_up_down (vnet_main_t * vnm, - u32 sw_if_index, u32 flags) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_neighbor_t *n; - u32 i, *to_delete = 0; - - /* *INDENT-OFF* */ - pool_foreach (n, nm->neighbor_pool, - ({ - if (n->key.sw_if_index == sw_if_index) - vec_add1 (to_delete, n - nm->neighbor_pool); - })); - /* *INDENT-ON* */ - - if (flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP) - { - for (i = 0; i < vec_len (to_delete); i++) - { - n = pool_elt_at_index (nm->neighbor_pool, to_delete[i]); - adj_nbr_walk_nh6 (n->key.sw_if_index, &n->key.ip6_address, - ip6_nd_mk_complete_walk, n); - } - } - else - { - for (i = 0; i < vec_len (to_delete); i++) - { - n = pool_elt_at_index (nm->neighbor_pool, to_delete[i]); - adj_nbr_walk_nh6 (n->key.sw_if_index, &n->key.ip6_address, - ip6_nd_mk_incomplete_walk, NULL); - if (n->flags & IP_NEIGHBOR_FLAG_STATIC) - continue; - ip6_neighbor_adj_fib_remove (n, - ip6_fib_table_get_index_for_sw_if_index - (n->key.sw_if_index)); - mhash_unset (&nm->neighbor_index_by_key, &n->key, 0); - pool_put (nm->neighbor_pool, n); - } - } - - vec_free (to_delete); - return 0; -} - -VNET_SW_INTERFACE_ADMIN_UP_DOWN_FUNCTION (ip6_neighbor_sw_interface_up_down); - -void -ip6_ethernet_update_adjacency (vnet_main_t * vnm, u32 sw_if_index, u32 ai) -{ - ip6_neighbor_t *nbr; - ip_adjacency_t *adj; - - adj = adj_get (ai); - - nbr = ip6_nd_find (sw_if_index, &adj->sub_type.nbr.next_hop.ip6); - - switch (adj->lookup_next_index) - { - case IP_LOOKUP_NEXT_GLEAN: - adj_glean_update_rewrite (ai); - break; - case IP_LOOKUP_NEXT_ARP: - if (NULL != nbr) - { - adj_nbr_walk_nh6 (sw_if_index, &nbr->key.ip6_address, - ip6_nd_mk_complete_walk, nbr); - } - else - { - /* - * no matching ND entry. - * construct the rewrite required to for an ND packet, and stick - * that in the adj's pipe to smoke. - */ - adj_nbr_update_rewrite (ai, - ADJ_NBR_REWRITE_FLAG_INCOMPLETE, - ethernet_build_rewrite (vnm, - sw_if_index, - VNET_LINK_IP6, - VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST)); - - /* - * since the FIB has added this adj for a route, it makes sense it may - * want to forward traffic sometime soon. Let's send a speculative ND. - * just one. If we were to do periodically that wouldn't be bad either, - * but that's more code than i'm prepared to write at this time for - * relatively little reward. - */ - ip6_nbr_probe (adj); - } - break; - case IP_LOOKUP_NEXT_BCAST: - adj_nbr_update_rewrite (ai, - ADJ_NBR_REWRITE_FLAG_COMPLETE, - ethernet_build_rewrite (vnm, - sw_if_index, - VNET_LINK_IP6, - VNET_REWRITE_FOR_SW_INTERFACE_ADDRESS_BROADCAST)); - break; - case IP_LOOKUP_NEXT_MCAST: - { - /* - * Construct a partial rewrite from the known ethernet mcast dest MAC - */ - u8 *rewrite; - u8 offset; - - rewrite = ethernet_build_rewrite (vnm, - sw_if_index, - adj->ia_link, - ethernet_ip6_mcast_dst_addr ()); - - /* - * Complete the remaining fields of the adj's rewrite to direct the - * complete of the rewrite at switch time by copying in the IP - * dst address's bytes. - * Ofset is 2 bytes into the destintation address. - */ - offset = vec_len (rewrite) - 2; - adj_mcast_update_rewrite (ai, rewrite, offset); - - break; - } - case IP_LOOKUP_NEXT_DROP: - case IP_LOOKUP_NEXT_PUNT: - case IP_LOOKUP_NEXT_LOCAL: - case IP_LOOKUP_NEXT_REWRITE: - case IP_LOOKUP_NEXT_MCAST_MIDCHAIN: - case IP_LOOKUP_NEXT_MIDCHAIN: - case IP_LOOKUP_NEXT_ICMP_ERROR: - case IP_LOOKUP_N_NEXT: - ASSERT (0); - break; - } -} - - -static void -ip6_neighbor_adj_fib_add (ip6_neighbor_t * n, u32 fib_index) -{ - if (ip6_address_is_link_local_unicast (&n->key.ip6_address)) - { - ip6_ll_prefix_t pfx = { - .ilp_addr = n->key.ip6_address, - .ilp_sw_if_index = n->key.sw_if_index, - }; - n->fib_entry_index = - ip6_ll_table_entry_update (&pfx, FIB_ROUTE_PATH_FLAG_NONE); - } - else - { - fib_prefix_t pfx = { - .fp_len = 128, - .fp_proto = FIB_PROTOCOL_IP6, - .fp_addr.ip6 = n->key.ip6_address, - }; - - n->fib_entry_index = - fib_table_entry_path_add (fib_index, &pfx, FIB_SOURCE_ADJ, - FIB_ENTRY_FLAG_ATTACHED, - DPO_PROTO_IP6, &pfx.fp_addr, - n->key.sw_if_index, ~0, 1, NULL, - FIB_ROUTE_PATH_FLAG_NONE); - } -} - -static ip6_neighbor_t * -force_reuse_neighbor_entry (void) -{ - ip6_neighbor_t *n; - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - u32 count = 0; - u32 index = pool_next_index (nm->neighbor_pool, nm->neighbor_delete_rotor); - if (index == ~0) /* Try again from elt 0 */ - index = pool_next_index (nm->neighbor_pool, index); - - /* Find a non-static random entry to free up for reuse */ - do - { - if ((count++ == 100) || (index == ~0)) - return NULL; /* give up after 100 entries */ - n = pool_elt_at_index (nm->neighbor_pool, index); - nm->neighbor_delete_rotor = index; - index = pool_next_index (nm->neighbor_pool, index); - } - while (n->flags & IP_NEIGHBOR_FLAG_STATIC); - - /* Remove ARP entry from its interface and update fib */ - adj_nbr_walk_nh6 (n->key.sw_if_index, - &n->key.ip6_address, ip6_nd_mk_incomplete_walk, NULL); - ip6_neighbor_adj_fib_remove - (n, ip6_fib_table_get_index_for_sw_if_index (n->key.sw_if_index)); - mhash_unset (&nm->neighbor_index_by_key, &n->key, 0); - - return n; -} - -int -vnet_set_ip6_ethernet_neighbor (vlib_main_t * vm, - u32 sw_if_index, - const ip6_address_t * a, - const mac_address_t * mac, - ip_neighbor_flags_t flags) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_neighbor_key_t k; - ip6_neighbor_t *n = 0; - int make_new_nd_cache_entry = 1; - uword *p; - u32 next_index; - pending_resolution_t *pr, *mc; - - if (vlib_get_thread_index ()) - { - set_unset_ip6_neighbor_rpc (vm, sw_if_index, a, mac, 1, flags); - return 0; - } - - k.sw_if_index = sw_if_index; - k.ip6_address = a[0]; - k.pad = 0; - - p = mhash_get (&nm->neighbor_index_by_key, &k); - if (p) - { - n = pool_elt_at_index (nm->neighbor_pool, p[0]); - /* Refuse to over-write static neighbor entry. */ - if (!(flags & IP_NEIGHBOR_FLAG_STATIC) && - (n->flags & IP_NEIGHBOR_FLAG_STATIC)) - { - /* if MAC address match, still check to send event */ - if (0 == mac_address_cmp (&n->mac, mac)) - goto check_customers; - return -2; - } - make_new_nd_cache_entry = 0; - } - - if (make_new_nd_cache_entry) - { - if (nm->limit_neighbor_cache_size && - pool_elts (nm->neighbor_pool) >= nm->limit_neighbor_cache_size) - { - n = force_reuse_neighbor_entry (); - if (NULL == n) - return -2; - } - else - { - pool_get (nm->neighbor_pool, n); - memset (n, 0, sizeof (*n)); - } - - mhash_set (&nm->neighbor_index_by_key, &k, n - nm->neighbor_pool, - /* old value */ 0); - n->key = k; - n->fib_entry_index = FIB_NODE_INDEX_INVALID; - - mac_address_copy (&n->mac, mac); - - /* - * create the adj-fib. the entry in the FIB table for and to the peer. - */ - if (!(flags & IP_NEIGHBOR_FLAG_NO_FIB_ENTRY)) - { - ip6_neighbor_adj_fib_add - (n, ip6_fib_table_get_index_for_sw_if_index (n->key.sw_if_index)); - } - else - { - n->flags |= IP_NEIGHBOR_FLAG_NO_FIB_ENTRY; - } - } - else - { - /* - * prevent a DoS attack from the data-plane that - * spams us with no-op updates to the MAC address - */ - if (0 == mac_address_cmp (&n->mac, mac)) - { - n->time_last_updated = vlib_time_now (vm); - goto check_customers; - } - - mac_address_copy (&n->mac, mac); - } - - /* Update time stamp and flags. */ - n->time_last_updated = vlib_time_now (vm); - if (flags & IP_NEIGHBOR_FLAG_STATIC) - { - n->flags |= IP_NEIGHBOR_FLAG_STATIC; - n->flags &= ~IP_NEIGHBOR_FLAG_DYNAMIC; - } - else - { - n->flags |= IP_NEIGHBOR_FLAG_DYNAMIC; - n->flags &= ~IP_NEIGHBOR_FLAG_STATIC; - } - - adj_nbr_walk_nh6 (sw_if_index, - &n->key.ip6_address, ip6_nd_mk_complete_walk, n); - -check_customers: - /* Customer(s) waiting for this address to be resolved? */ - p = mhash_get (&nm->pending_resolutions_by_address, a); - if (p) - { - next_index = p[0]; - - while (next_index != (u32) ~ 0) - { - pr = pool_elt_at_index (nm->pending_resolutions, next_index); - vlib_process_signal_event (vm, pr->node_index, - pr->type_opaque, pr->data); - next_index = pr->next_index; - pool_put (nm->pending_resolutions, pr); - } - - mhash_unset (&nm->pending_resolutions_by_address, (void *) a, 0); - } - - /* Customer(s) requesting ND event for this address? */ - p = mhash_get (&nm->mac_changes_by_address, a); - if (p) - { - next_index = p[0]; - - while (next_index != (u32) ~ 0) - { - int rv = 1; - - mc = pool_elt_at_index (nm->mac_changes, next_index); - - /* Call the user's data callback, return 1 to suppress dup events */ - if (mc->data_callback) - rv = (mc->data_callback) (mc->data, mac, sw_if_index, &ip6a_zero); - /* - * Signal the resolver process, as long as the user - * says they want to be notified - */ - if (rv == 0) - vlib_process_signal_event (vm, mc->node_index, - mc->type_opaque, mc->data); - next_index = mc->next_index; - } - } - - return 0; -} - -int -vnet_unset_ip6_ethernet_neighbor (vlib_main_t * vm, - u32 sw_if_index, const ip6_address_t * a) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_neighbor_key_t k; - ip6_neighbor_t *n; - uword *p; - int rv = 0; - - if (vlib_get_thread_index ()) - { - set_unset_ip6_neighbor_rpc (vm, sw_if_index, a, NULL, 0, - IP_NEIGHBOR_FLAG_NONE); - return 0; - } - - k.sw_if_index = sw_if_index; - k.ip6_address = a[0]; - k.pad = 0; - - p = mhash_get (&nm->neighbor_index_by_key, &k); - if (p == 0) - { - rv = -1; - goto out; - } - - n = pool_elt_at_index (nm->neighbor_pool, p[0]); - - adj_nbr_walk_nh6 (sw_if_index, - &n->key.ip6_address, ip6_nd_mk_incomplete_walk, NULL); - ip6_neighbor_adj_fib_remove - (n, ip6_fib_table_get_index_for_sw_if_index (sw_if_index)); - - mhash_unset (&nm->neighbor_index_by_key, &n->key, 0); - pool_put (nm->neighbor_pool, n); - -out: - return rv; -} - -static void ip6_neighbor_set_unset_rpc_callback - (ip6_neighbor_set_unset_rpc_args_t * a) -{ - vlib_main_t *vm = vlib_get_main (); - if (a->is_add) - vnet_set_ip6_ethernet_neighbor (vm, a->sw_if_index, &a->addr, - &a->mac, a->flags); - else - vnet_unset_ip6_ethernet_neighbor (vm, a->sw_if_index, &a->addr); -} - -static int -ip6_neighbor_sort (void *a1, void *a2) -{ - vnet_main_t *vnm = vnet_get_main (); - ip6_neighbor_t *n1 = a1, *n2 = a2; - int cmp; - cmp = vnet_sw_interface_compare (vnm, n1->key.sw_if_index, - n2->key.sw_if_index); - if (!cmp) - cmp = ip6_address_compare (&n1->key.ip6_address, &n2->key.ip6_address); - return cmp; -} - -ip6_neighbor_t * -ip6_neighbors_pool (void) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - return nm->neighbor_pool; -} - -ip6_neighbor_t * -ip6_neighbors_entries (u32 sw_if_index) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_neighbor_t *n, *ns = NULL; - - /* *INDENT-OFF* */ - pool_foreach (n, nm->neighbor_pool, - ({ - if (sw_if_index != ~0 && n->key.sw_if_index != sw_if_index) - continue; - vec_add1 (ns, n[0]); - })); - /* *INDENT-ON* */ - - if (ns) - vec_sort_with_function (ns, ip6_neighbor_sort); - return ns; -} - -static clib_error_t * -show_ip6_neighbors (vlib_main_t * vm, - unformat_input_t * input, vlib_cli_command_t * cmd) -{ - vnet_main_t *vnm = vnet_get_main (); - ip6_neighbor_t *n, *ns; - clib_error_t *error = 0; - u32 sw_if_index; - int verbose = 0; - - if (unformat (input, "verbose")) - verbose = 1; - - /* Filter entries by interface if given. */ - sw_if_index = ~0; - (void) unformat_user (input, unformat_vnet_sw_interface, vnm, &sw_if_index); - - ns = ip6_neighbors_entries (sw_if_index); - if (ns) - { - /* - * Show the entire table if it's not too big, otherwise just - * show the size of the table. - */ - if (vec_len (ns) < 50) - verbose = 1; - if (verbose) - { - vlib_cli_output (vm, "%U", format_ip6_neighbor_ip6_entry, vm, 0); - vec_foreach (n, ns) - { - vlib_cli_output (vm, "%U", format_ip6_neighbor_ip6_entry, vm, n); - } - } - else - vlib_cli_output - (vm, "There are %u ip6 neighbors, " - "'show ip6 neighbors verbose' to display the entire table...", - vec_len (ns)); - vec_free (ns); - } - else - vlib_cli_output (vm, "No ip6 neighbors"); - - return error; -} - -/*? - * This command is used to display the adjacent IPv6 hosts found via - * neighbor discovery. Optionally, limit the output to the specified - * interface. - * - * @cliexpar - * Example of how to display the IPv6 neighbor adjacency table: - * @cliexstart{show ip6 neighbors} - * Time Address Flags Link layer Interface - * 34.0910 ::a:1:1:0:7 02:fe:6a:07:39:6f GigabitEthernet2/0/0 - * 173.2916 ::b:5:1:c:2 02:fe:50:62:3a:94 GigabitEthernet2/0/0 - * 886.6654 ::1:1:c:0:9 S 02:fe:e4:45:27:5b GigabitEthernet3/0/0 - * @cliexend - * Example of how to display the IPv6 neighbor adjacency table for given interface: - * @cliexstart{show ip6 neighbors GigabitEthernet2/0/0} - * Time Address Flags Link layer Interface - * 34.0910 ::a:1:1:0:7 02:fe:6a:07:39:6f GigabitEthernet2/0/0 - * 173.2916 ::b:5:1:c:2 02:fe:50:62:3a:94 GigabitEthernet2/0/0 - * @cliexend -?*/ -/* *INDENT-OFF* */ -VLIB_CLI_COMMAND (show_ip6_neighbors_command, static) = { - .path = "show ip6 neighbors", - .function = show_ip6_neighbors, - .short_help = "show ip6 neighbors [<interface>]", -}; -/* *INDENT-ON* */ - -static clib_error_t * -set_ip6_neighbor (vlib_main_t * vm, - unformat_input_t * input, vlib_cli_command_t * cmd) -{ - ip_neighbor_flags_t flags = IP_NEIGHBOR_FLAG_NONE; - vnet_main_t *vnm = vnet_get_main (); - ip6_address_t addr; - mac_address_t mac; - int addr_valid = 0; - int is_del = 0; - u32 sw_if_index; - - while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT) - { - /* intfc, ip6-address, mac-address */ - if (unformat (input, "%U %U %U", - unformat_vnet_sw_interface, vnm, &sw_if_index, - unformat_ip6_address, &addr, - unformat_mac_address_t, &mac)) - addr_valid = 1; - - else if (unformat (input, "delete") || unformat (input, "del")) - is_del = 1; - else if (unformat (input, "static")) - flags |= IP_NEIGHBOR_FLAG_STATIC; - else if (unformat (input, "no-fib-entry")) - flags |= IP_NEIGHBOR_FLAG_NO_FIB_ENTRY; - else - break; - } - - if (!addr_valid) - return clib_error_return (0, "Missing interface, ip6 or hw address"); - - if (!is_del) - vnet_set_ip6_ethernet_neighbor (vm, sw_if_index, &addr, &mac, flags); - else - vnet_unset_ip6_ethernet_neighbor (vm, sw_if_index, &addr); - return 0; -} - -/*? - * This command is used to manually add an entry to the IPv6 neighbor - * adjacency table. Optionally, the entry can be added as static. It is - * also used to remove an entry from the table. Use the '<em>show ip6 - * neighbors</em>' command to display all learned and manually entered entries. - * - * @cliexpar - * Example of how to add a static entry to the IPv6 neighbor adjacency table: - * @cliexcmd{set ip6 neighbor GigabitEthernet2/0/0 ::1:1:c:0:9 02:fe:e4:45:27:5b static} - * Example of how to delete an entry from the IPv6 neighbor adjacency table: - * @cliexcmd{set ip6 neighbor del GigabitEthernet2/0/0 ::1:1:c:0:9 02:fe:e4:45:27:5b} -?*/ -/* *INDENT-OFF* */ -VLIB_CLI_COMMAND (set_ip6_neighbor_command, static) = -{ - .path = "set ip6 neighbor", - .function = set_ip6_neighbor, - .short_help = "set ip6 neighbor [del] <interface> <ip6-address> <mac-address> [static]", -}; -/* *INDENT-ON* */ - -typedef enum -{ - ICMP6_NEIGHBOR_SOLICITATION_NEXT_DROP, - ICMP6_NEIGHBOR_SOLICITATION_NEXT_REPLY, - ICMP6_NEIGHBOR_SOLICITATION_N_NEXT, -} icmp6_neighbor_solicitation_or_advertisement_next_t; - -static_always_inline uword -icmp6_neighbor_solicitation_or_advertisement (vlib_main_t * vm, - vlib_node_runtime_t * node, - vlib_frame_t * frame, - uword is_solicitation) -{ - vnet_main_t *vnm = vnet_get_main (); - ip6_main_t *im = &ip6_main; - uword n_packets = frame->n_vectors; - u32 *from, *to_next; - u32 n_left_from, n_left_to_next, next_index, n_advertisements_sent; - icmp6_neighbor_discovery_option_type_t option_type; - vlib_node_runtime_t *error_node = - vlib_node_get_runtime (vm, ip6_icmp_input_node.index); - int bogus_length; - - from = vlib_frame_vector_args (frame); - n_left_from = n_packets; - next_index = node->cached_next_index; - - if (node->flags & VLIB_NODE_FLAG_TRACE) - vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors, - /* stride */ 1, - sizeof (icmp6_input_trace_t)); - - option_type = - (is_solicitation - ? ICMP6_NEIGHBOR_DISCOVERY_OPTION_source_link_layer_address - : ICMP6_NEIGHBOR_DISCOVERY_OPTION_target_link_layer_address); - n_advertisements_sent = 0; - - 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) - { - vlib_buffer_t *p0; - ip6_header_t *ip0; - icmp6_neighbor_solicitation_or_advertisement_header_t *h0; - icmp6_neighbor_discovery_ethernet_link_layer_address_option_t *o0; - u32 bi0, options_len0, sw_if_index0, next0, error0; - u32 ip6_sadd_link_local, ip6_sadd_unspecified; - int is_rewrite0; - u32 ni0; - - bi0 = to_next[0] = from[0]; - - from += 1; - to_next += 1; - n_left_from -= 1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer (vm, bi0); - ip0 = vlib_buffer_get_current (p0); - h0 = ip6_next_header (ip0); - options_len0 = - clib_net_to_host_u16 (ip0->payload_length) - sizeof (h0[0]); - - error0 = ICMP6_ERROR_NONE; - sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX]; - ip6_sadd_link_local = - ip6_address_is_link_local_unicast (&ip0->src_address); - ip6_sadd_unspecified = - ip6_address_is_unspecified (&ip0->src_address); - - /* Check that source address is unspecified, link-local or else on-link. */ - if (!ip6_sadd_unspecified && !ip6_sadd_link_local) - { - u32 src_adj_index0 = ip6_src_lookup_for_packet (im, p0, ip0); - - if (ADJ_INDEX_INVALID != src_adj_index0) - { - ip_adjacency_t *adj0 = adj_get (src_adj_index0); - - /* Allow all realistic-looking rewrite adjacencies to pass */ - ni0 = adj0->lookup_next_index; - is_rewrite0 = (ni0 >= IP_LOOKUP_NEXT_ARP) && - (ni0 < IP6_LOOKUP_N_NEXT); - - error0 = ((adj0->rewrite_header.sw_if_index != sw_if_index0 - || !is_rewrite0) - ? - ICMP6_ERROR_NEIGHBOR_SOLICITATION_SOURCE_NOT_ON_LINK - : error0); - } - else - { - error0 = - ICMP6_ERROR_NEIGHBOR_SOLICITATION_SOURCE_NOT_ON_LINK; - } - } - - o0 = (void *) (h0 + 1); - o0 = ((options_len0 == 8 && o0->header.type == option_type - && o0->header.n_data_u64s == 1) ? o0 : 0); - - /* If src address unspecified or link local, donot learn neighbor MAC */ - if (PREDICT_TRUE (error0 == ICMP6_ERROR_NONE && o0 != 0 && - !ip6_sadd_unspecified)) - { - vnet_set_ip6_ethernet_neighbor (vm, sw_if_index0, - is_solicitation ? - &ip0->src_address : - &h0->target_address, - (mac_address_t *) - o0->ethernet_address, - IP_NEIGHBOR_FLAG_NONE); - } - - if (is_solicitation && error0 == ICMP6_ERROR_NONE) - { - /* Check that target address is local to this router. */ - fib_node_index_t fei; - u32 fib_index; - - fib_index = - ip6_fib_table_get_index_for_sw_if_index (sw_if_index0); - - if (~0 == fib_index) - { - error0 = ICMP6_ERROR_NEIGHBOR_SOLICITATION_SOURCE_UNKNOWN; - } - else - { - if (ip6_address_is_link_local_unicast (&h0->target_address)) - { - fei = ip6_fib_table_lookup_exact_match - (ip6_ll_fib_get (sw_if_index0), - &h0->target_address, 128); - } - else - { - fei = ip6_fib_table_lookup_exact_match (fib_index, - &h0->target_address, - 128); - } - - if (FIB_NODE_INDEX_INVALID == fei) - { - /* The target address is not in the FIB */ - error0 = - ICMP6_ERROR_NEIGHBOR_SOLICITATION_SOURCE_UNKNOWN; - } - else - { - if (FIB_ENTRY_FLAG_LOCAL & - fib_entry_get_flags_for_source (fei, - FIB_SOURCE_INTERFACE)) - { - /* It's an address that belongs to one of our interfaces - * that's good. */ - } - else - if (fib_entry_is_sourced - (fei, FIB_SOURCE_IP6_ND_PROXY) || - fib_entry_is_sourced (fei, FIB_SOURCE_IP6_ND)) - { - /* The address was added by IPv6 Proxy ND config. - * We should only respond to these if the NS arrived on - * the link that has a matching covering prefix */ - } - else - { - error0 = - ICMP6_ERROR_NEIGHBOR_SOLICITATION_SOURCE_UNKNOWN; - } - } - } - } - - if (is_solicitation) - next0 = (error0 != ICMP6_ERROR_NONE - ? ICMP6_NEIGHBOR_SOLICITATION_NEXT_DROP - : ICMP6_NEIGHBOR_SOLICITATION_NEXT_REPLY); - else - { - next0 = 0; - error0 = error0 == ICMP6_ERROR_NONE ? - ICMP6_ERROR_NEIGHBOR_ADVERTISEMENTS_RX : error0; - } - - if (is_solicitation && error0 == ICMP6_ERROR_NONE) - { - vnet_sw_interface_t *sw_if0; - ethernet_interface_t *eth_if0; - ethernet_header_t *eth0; - - /* dst address is either source address or the all-nodes mcast addr */ - if (!ip6_sadd_unspecified) - ip0->dst_address = ip0->src_address; - else - ip6_set_reserved_multicast_address (&ip0->dst_address, - IP6_MULTICAST_SCOPE_link_local, - IP6_MULTICAST_GROUP_ID_all_hosts); - - ip0->src_address = h0->target_address; - ip0->hop_limit = 255; - h0->icmp.type = ICMP6_neighbor_advertisement; - - sw_if0 = vnet_get_sup_sw_interface (vnm, sw_if_index0); - ASSERT (sw_if0->type == VNET_SW_INTERFACE_TYPE_HARDWARE); - eth_if0 = - ethernet_get_interface (ðernet_main, sw_if0->hw_if_index); - if (eth_if0 && o0) - { - clib_memcpy (o0->ethernet_address, eth_if0->address, 6); - o0->header.type = - ICMP6_NEIGHBOR_DISCOVERY_OPTION_target_link_layer_address; - } - - h0->advertisement_flags = clib_host_to_net_u32 - (ICMP6_NEIGHBOR_ADVERTISEMENT_FLAG_SOLICITED - | ICMP6_NEIGHBOR_ADVERTISEMENT_FLAG_OVERRIDE); - - h0->icmp.checksum = 0; - h0->icmp.checksum = - ip6_tcp_udp_icmp_compute_checksum (vm, p0, ip0, - &bogus_length); - ASSERT (bogus_length == 0); - - /* Reuse current MAC header, copy SMAC to DMAC and - * interface MAC to SMAC */ - vlib_buffer_advance (p0, -ethernet_buffer_header_size (p0)); - eth0 = vlib_buffer_get_current (p0); - clib_memcpy (eth0->dst_address, eth0->src_address, 6); - if (eth_if0) - clib_memcpy (eth0->src_address, eth_if0->address, 6); - - /* Setup input and output sw_if_index for packet */ - ASSERT (vnet_buffer (p0)->sw_if_index[VLIB_RX] == sw_if_index0); - vnet_buffer (p0)->sw_if_index[VLIB_TX] = sw_if_index0; - vnet_buffer (p0)->sw_if_index[VLIB_RX] = - vnet_main.local_interface_sw_if_index; - - n_advertisements_sent++; - } - - p0->error = error_node->errors[error0]; - - vlib_validate_buffer_enqueue_x1 (vm, node, next_index, - to_next, n_left_to_next, - bi0, next0); - } - - vlib_put_next_frame (vm, node, next_index, n_left_to_next); - } - - /* Account for advertisements sent. */ - vlib_error_count (vm, error_node->node_index, - ICMP6_ERROR_NEIGHBOR_ADVERTISEMENTS_TX, - n_advertisements_sent); - - return frame->n_vectors; -} - -/* for "syslogging" - use elog for now */ -#define foreach_log_level \ - _ (DEBUG, "DEBUG") \ - _ (INFO, "INFORMATION") \ - _ (NOTICE, "NOTICE") \ - _ (WARNING, "WARNING") \ - _ (ERR, "ERROR") \ - _ (CRIT, "CRITICAL") \ - _ (ALERT, "ALERT") \ - _ (EMERG, "EMERGENCY") - -typedef enum -{ -#define _(f,s) LOG_##f, - foreach_log_level -#undef _ -} log_level_t; - -static char *log_level_strings[] = { -#define _(f,s) s, - foreach_log_level -#undef _ -}; - -static int logmask = 1 << LOG_DEBUG; - -static void -ip6_neighbor_syslog (vlib_main_t * vm, int priority, char *fmt, ...) -{ - /* just use elog for now */ - u8 *what; - va_list va; - - if ((priority > LOG_EMERG) || !(logmask & (1 << priority))) - return; - - va_start (va, fmt); - if (fmt) - { - what = va_format (0, fmt, &va); - - ELOG_TYPE_DECLARE (e) = - { - .format = "ip6 nd: (%s): %s",.format_args = "T4T4",}; - struct - { - u32 s[2]; - } *ed; - ed = ELOG_DATA (&vm->elog_main, e); - ed->s[0] = elog_string (&vm->elog_main, log_level_strings[priority]); - ed->s[1] = elog_string (&vm->elog_main, (char *) what); - } - va_end (va); - return; -} - -clib_error_t * -call_ip6_neighbor_callbacks (void *data, - _vnet_ip6_neighbor_function_list_elt_t * elt) -{ - clib_error_t *error = 0; - - while (elt) - { - error = elt->fp (data); - if (error) - return error; - elt = elt->next_ip6_neighbor_function; - } - - return error; -} - -/* ipv6 neighbor discovery - router advertisements */ -typedef enum -{ - ICMP6_ROUTER_SOLICITATION_NEXT_DROP, - ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_RW, - ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_TX, - ICMP6_ROUTER_SOLICITATION_N_NEXT, -} icmp6_router_solicitation_or_advertisement_next_t; - -static_always_inline uword -icmp6_router_solicitation (vlib_main_t * vm, - vlib_node_runtime_t * node, vlib_frame_t * frame) -{ - vnet_main_t *vnm = vnet_get_main (); - ip6_main_t *im = &ip6_main; - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - uword n_packets = frame->n_vectors; - u32 *from, *to_next; - u32 n_left_from, n_left_to_next, next_index; - u32 n_advertisements_sent = 0; - int bogus_length; - - icmp6_neighbor_discovery_option_type_t option_type; - - vlib_node_runtime_t *error_node = - vlib_node_get_runtime (vm, ip6_icmp_input_node.index); - - from = vlib_frame_vector_args (frame); - n_left_from = n_packets; - next_index = node->cached_next_index; - - if (node->flags & VLIB_NODE_FLAG_TRACE) - vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors, - /* stride */ 1, - sizeof (icmp6_input_trace_t)); - - /* source may append his LL address */ - option_type = ICMP6_NEIGHBOR_DISCOVERY_OPTION_source_link_layer_address; - - 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) - { - vlib_buffer_t *p0; - ip6_header_t *ip0; - ip6_radv_t *radv_info = 0; - - icmp6_neighbor_discovery_header_t *h0; - icmp6_neighbor_discovery_ethernet_link_layer_address_option_t *o0; - - u32 bi0, options_len0, sw_if_index0, next0, error0; - u32 is_solicitation = 1, is_dropped = 0; - u32 is_unspecified, is_link_local; - - bi0 = to_next[0] = from[0]; - - from += 1; - to_next += 1; - n_left_from -= 1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer (vm, bi0); - ip0 = vlib_buffer_get_current (p0); - h0 = ip6_next_header (ip0); - options_len0 = - clib_net_to_host_u16 (ip0->payload_length) - sizeof (h0[0]); - is_unspecified = ip6_address_is_unspecified (&ip0->src_address); - is_link_local = - ip6_address_is_link_local_unicast (&ip0->src_address); - - error0 = ICMP6_ERROR_NONE; - sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX]; - - /* check if solicitation (not from nd_timer node) */ - if (ip6_address_is_unspecified (&ip0->dst_address)) - is_solicitation = 0; - - /* Check that source address is unspecified, link-local or else on-link. */ - if (!is_unspecified && !is_link_local) - { - u32 src_adj_index0 = ip6_src_lookup_for_packet (im, p0, ip0); - - if (ADJ_INDEX_INVALID != src_adj_index0) - { - ip_adjacency_t *adj0 = adj_get (src_adj_index0); - - error0 = (adj0->rewrite_header.sw_if_index != sw_if_index0 - ? - ICMP6_ERROR_ROUTER_SOLICITATION_SOURCE_NOT_ON_LINK - : error0); - } - else - { - error0 = ICMP6_ERROR_ROUTER_SOLICITATION_SOURCE_NOT_ON_LINK; - } - } - - /* check for source LL option and process */ - o0 = (void *) (h0 + 1); - o0 = ((options_len0 == 8 - && o0->header.type == option_type - && o0->header.n_data_u64s == 1) ? o0 : 0); - - /* if src address unspecified IGNORE any options */ - if (PREDICT_TRUE (error0 == ICMP6_ERROR_NONE && o0 != 0 && - !is_unspecified && !is_link_local)) - { - vnet_set_ip6_ethernet_neighbor - (vm, sw_if_index0, - &ip0->src_address, - (mac_address_t *) o0->ethernet_address, - IP_NEIGHBOR_FLAG_NONE); - } - - /* default is to drop */ - next0 = ICMP6_ROUTER_SOLICITATION_NEXT_DROP; - - if (error0 == ICMP6_ERROR_NONE) - { - vnet_sw_interface_t *sw_if0; - ethernet_interface_t *eth_if0; - u32 adj_index0; - - sw_if0 = vnet_get_sup_sw_interface (vnm, sw_if_index0); - ASSERT (sw_if0->type == VNET_SW_INTERFACE_TYPE_HARDWARE); - eth_if0 = - ethernet_get_interface (ðernet_main, sw_if0->hw_if_index); - - /* only support ethernet interface type for now */ - error0 = - (!eth_if0) ? ICMP6_ERROR_ROUTER_SOLICITATION_UNSUPPORTED_INTF - : error0; - - if (error0 == ICMP6_ERROR_NONE) - { - u32 ri; - - /* adjust the sizeof the buffer to just include the ipv6 header */ - p0->current_length -= - (options_len0 + - sizeof (icmp6_neighbor_discovery_header_t)); - - /* look up the radv_t information for this interface */ - if (vec_len (nm->if_radv_pool_index_by_sw_if_index) > - sw_if_index0) - { - ri = - nm->if_radv_pool_index_by_sw_if_index[sw_if_index0]; - - if (ri != ~0) - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - } - - error0 = - ((!radv_info) ? - ICMP6_ERROR_ROUTER_SOLICITATION_RADV_NOT_CONFIG : - error0); - - if (error0 == ICMP6_ERROR_NONE) - { - f64 now = vlib_time_now (vm); - - /* for solicited adverts - need to rate limit */ - if (is_solicitation) - { - if (0 != radv_info->last_radv_time && - (now - radv_info->last_radv_time) < - MIN_DELAY_BETWEEN_RAS) - is_dropped = 1; - else - radv_info->last_radv_time = now; - } - - /* send now */ - icmp6_router_advertisement_header_t rh; - - rh.icmp.type = ICMP6_router_advertisement; - rh.icmp.code = 0; - rh.icmp.checksum = 0; - - rh.current_hop_limit = radv_info->curr_hop_limit; - rh.router_lifetime_in_sec = - clib_host_to_net_u16 - (radv_info->adv_router_lifetime_in_sec); - rh. - time_in_msec_between_retransmitted_neighbor_solicitations - = - clib_host_to_net_u32 (radv_info-> - adv_time_in_msec_between_retransmitted_neighbor_solicitations); - rh.neighbor_reachable_time_in_msec = - clib_host_to_net_u32 (radv_info-> - adv_neighbor_reachable_time_in_msec); - - rh.flags = - (radv_info->adv_managed_flag) ? - ICMP6_ROUTER_DISCOVERY_FLAG_ADDRESS_CONFIG_VIA_DHCP : - 0; - rh.flags |= - ((radv_info->adv_other_flag) ? - ICMP6_ROUTER_DISCOVERY_FLAG_OTHER_CONFIG_VIA_DHCP : - 0); - - - u16 payload_length = - sizeof (icmp6_router_advertisement_header_t); - - if (vlib_buffer_add_data - (vm, &bi0, (void *) &rh, - sizeof (icmp6_router_advertisement_header_t))) - { - /* buffer allocation failed, drop the pkt */ - error0 = ICMP6_ERROR_ALLOC_FAILURE; - goto drop0; - } - - if (radv_info->adv_link_layer_address) - { - icmp6_neighbor_discovery_ethernet_link_layer_address_option_t - h; - - h.header.type = - ICMP6_NEIGHBOR_DISCOVERY_OPTION_source_link_layer_address; - h.header.n_data_u64s = 1; - - /* copy ll address */ - clib_memcpy (&h.ethernet_address[0], - eth_if0->address, 6); - - if (vlib_buffer_add_data - (vm, &bi0, (void *) &h, - sizeof - (icmp6_neighbor_discovery_ethernet_link_layer_address_option_t))) - { - error0 = ICMP6_ERROR_ALLOC_FAILURE; - goto drop0; - } - - payload_length += - sizeof - (icmp6_neighbor_discovery_ethernet_link_layer_address_option_t); - } - - /* add MTU option */ - if (radv_info->adv_link_mtu) - { - icmp6_neighbor_discovery_mtu_option_t h; - - h.unused = 0; - h.mtu = - clib_host_to_net_u32 (radv_info->adv_link_mtu); - h.header.type = ICMP6_NEIGHBOR_DISCOVERY_OPTION_mtu; - h.header.n_data_u64s = 1; - - payload_length += - sizeof (icmp6_neighbor_discovery_mtu_option_t); - - if (vlib_buffer_add_data - (vm, &bi0, (void *) &h, - sizeof - (icmp6_neighbor_discovery_mtu_option_t))) - { - error0 = ICMP6_ERROR_ALLOC_FAILURE; - goto drop0; - } - } - - /* add advertised prefix options */ - ip6_radv_prefix_t *pr_info; - - /* *INDENT-OFF* */ - pool_foreach (pr_info, radv_info->adv_prefixes_pool, - ({ - if(pr_info->enabled && - (!pr_info->decrement_lifetime_flag - || (pr_info->pref_lifetime_expires >0))) - { - /* advertise this prefix */ - icmp6_neighbor_discovery_prefix_information_option_t h; - - h.header.type = ICMP6_NEIGHBOR_DISCOVERY_OPTION_prefix_information; - h.header.n_data_u64s = (sizeof(icmp6_neighbor_discovery_prefix_information_option_t) >> 3); - - h.dst_address_length = pr_info->prefix_len; - - h.flags = (pr_info->adv_on_link_flag) ? ICMP6_NEIGHBOR_DISCOVERY_PREFIX_INFORMATION_FLAG_ON_LINK : 0; - h.flags |= (pr_info->adv_autonomous_flag) ? ICMP6_NEIGHBOR_DISCOVERY_PREFIX_INFORMATION_AUTO : 0; - - if(radv_info->cease_radv && pr_info->deprecated_prefix_flag) - { - h.valid_time = clib_host_to_net_u32(MIN_ADV_VALID_LIFETIME); - h.preferred_time = 0; - } - else - { - if(pr_info->decrement_lifetime_flag) - { - pr_info->adv_valid_lifetime_in_secs = ((pr_info->valid_lifetime_expires > now)) ? - (pr_info->valid_lifetime_expires - now) : 0; - - pr_info->adv_pref_lifetime_in_secs = ((pr_info->pref_lifetime_expires > now)) ? - (pr_info->pref_lifetime_expires - now) : 0; - } - - h.valid_time = clib_host_to_net_u32(pr_info->adv_valid_lifetime_in_secs); - h.preferred_time = clib_host_to_net_u32(pr_info->adv_pref_lifetime_in_secs) ; - } - h.unused = 0; - - clib_warning ("Prefix %U valid %u preferred %u", - format_ip6_address, &pr_info->prefix, - ntohl(h.valid_time), - ntohl(h.preferred_time)); - - if (h.valid_time == 0) - clib_warning ("WARNING: valid_time 0!!!"); - - clib_memcpy(&h.dst_address, &pr_info->prefix, sizeof(ip6_address_t)); - - payload_length += sizeof( icmp6_neighbor_discovery_prefix_information_option_t); - - if (vlib_buffer_add_data - (vm, &bi0, (void *)&h, - sizeof(icmp6_neighbor_discovery_prefix_information_option_t))) - { - error0 = ICMP6_ERROR_ALLOC_FAILURE; - goto drop0; - } - - } - })); - /* *INDENT-ON* */ - - /* add additional options before here */ - - /* finish building the router advertisement... */ - if (!is_unspecified && radv_info->send_unicast) - { - ip0->dst_address = ip0->src_address; - } - else - { - /* target address is all-nodes mcast addr */ - ip6_set_reserved_multicast_address - (&ip0->dst_address, - IP6_MULTICAST_SCOPE_link_local, - IP6_MULTICAST_GROUP_ID_all_hosts); - } - - /* source address MUST be the link-local address */ - ip0->src_address = radv_info->link_local_address; - - ip0->hop_limit = 255; - ip0->payload_length = - clib_host_to_net_u16 (payload_length); - - icmp6_router_advertisement_header_t *rh0 = - (icmp6_router_advertisement_header_t *) (ip0 + 1); - rh0->icmp.checksum = - ip6_tcp_udp_icmp_compute_checksum (vm, p0, ip0, - &bogus_length); - ASSERT (bogus_length == 0); - - /* setup output if and adjacency */ - vnet_buffer (p0)->sw_if_index[VLIB_RX] = - vnet_main.local_interface_sw_if_index; - - if (is_solicitation) - { - ethernet_header_t *eth0; - /* Reuse current MAC header, copy SMAC to DMAC and - * interface MAC to SMAC */ - vlib_buffer_reset (p0); - eth0 = vlib_buffer_get_current (p0); - clib_memcpy (eth0->dst_address, eth0->src_address, - 6); - clib_memcpy (eth0->src_address, eth_if0->address, - 6); - next0 = - is_dropped ? next0 : - ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_TX; - vnet_buffer (p0)->sw_if_index[VLIB_TX] = - sw_if_index0; - } - else - { - adj_index0 = radv_info->mcast_adj_index; - if (adj_index0 == ADJ_INDEX_INVALID) - error0 = ICMP6_ERROR_DST_LOOKUP_MISS; - else - { - next0 = - is_dropped ? next0 : - ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_RW; - vnet_buffer (p0)->ip.adj_index[VLIB_TX] = - adj_index0; - } - } - p0->flags |= VNET_BUFFER_F_LOCALLY_ORIGINATED; - - radv_info->n_solicitations_dropped += is_dropped; - radv_info->n_solicitations_rcvd += is_solicitation; - - if ((error0 == ICMP6_ERROR_NONE) && !is_dropped) - { - radv_info->n_advertisements_sent++; - n_advertisements_sent++; - } - } - } - } - - drop0: - p0->error = error_node->errors[error0]; - - if (error0 != ICMP6_ERROR_NONE) - vlib_error_count (vm, error_node->node_index, error0, 1); - - vlib_validate_buffer_enqueue_x1 (vm, node, next_index, - to_next, n_left_to_next, - bi0, next0); - - } - - vlib_put_next_frame (vm, node, next_index, n_left_to_next); - } - - /* Account for router advertisements sent. */ - vlib_error_count (vm, error_node->node_index, - ICMP6_ERROR_ROUTER_ADVERTISEMENTS_TX, - n_advertisements_sent); - - return frame->n_vectors; -} - - /* validate advertised info for consistancy (see RFC-4861 section 6.2.7) - log any inconsistencies, packet will always be dropped */ -static_always_inline uword -icmp6_router_advertisement (vlib_main_t * vm, - vlib_node_runtime_t * node, vlib_frame_t * frame) -{ - vnet_main_t *vnm = vnet_get_main (); - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - uword n_packets = frame->n_vectors; - u32 *from, *to_next; - u32 n_left_from, n_left_to_next, next_index; - u32 n_advertisements_rcvd = 0; - - vlib_node_runtime_t *error_node = - vlib_node_get_runtime (vm, ip6_icmp_input_node.index); - - from = vlib_frame_vector_args (frame); - n_left_from = n_packets; - next_index = node->cached_next_index; - - if (node->flags & VLIB_NODE_FLAG_TRACE) - vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors, - /* stride */ 1, - sizeof (icmp6_input_trace_t)); - - 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) - { - vlib_buffer_t *p0; - ip6_header_t *ip0; - ip6_radv_t *radv_info = 0; - icmp6_router_advertisement_header_t *h0; - u32 bi0, options_len0, sw_if_index0, next0, error0; - - bi0 = to_next[0] = from[0]; - - from += 1; - to_next += 1; - n_left_from -= 1; - n_left_to_next -= 1; - - p0 = vlib_get_buffer (vm, bi0); - ip0 = vlib_buffer_get_current (p0); - h0 = ip6_next_header (ip0); - options_len0 = - clib_net_to_host_u16 (ip0->payload_length) - sizeof (h0[0]); - - error0 = ICMP6_ERROR_NONE; - sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX]; - - /* Check that source address is link-local */ - error0 = (!ip6_address_is_link_local_unicast (&ip0->src_address)) ? - ICMP6_ERROR_ROUTER_ADVERTISEMENT_SOURCE_NOT_LINK_LOCAL : error0; - - /* default is to drop */ - next0 = ICMP6_ROUTER_SOLICITATION_NEXT_DROP; - - n_advertisements_rcvd++; - - if (error0 == ICMP6_ERROR_NONE) - { - vnet_sw_interface_t *sw_if0; - ethernet_interface_t *eth_if0; - - sw_if0 = vnet_get_sup_sw_interface (vnm, sw_if_index0); - ASSERT (sw_if0->type == VNET_SW_INTERFACE_TYPE_HARDWARE); - eth_if0 = - ethernet_get_interface (ðernet_main, sw_if0->hw_if_index); - - /* only support ethernet interface type for now */ - error0 = - (!eth_if0) ? ICMP6_ERROR_ROUTER_SOLICITATION_UNSUPPORTED_INTF - : error0; - - if (error0 == ICMP6_ERROR_NONE) - { - u32 ri; - - /* look up the radv_t information for this interface */ - if (vec_len (nm->if_radv_pool_index_by_sw_if_index) > - sw_if_index0) - { - ri = - nm->if_radv_pool_index_by_sw_if_index[sw_if_index0]; - - if (ri != ~0) - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - } - - error0 = - ((!radv_info) ? - ICMP6_ERROR_ROUTER_SOLICITATION_RADV_NOT_CONFIG : - error0); - - if (error0 == ICMP6_ERROR_NONE) - { - radv_info->keep_sending_rs = 0; - - ra_report_t r; - - r.sw_if_index = sw_if_index0; - memcpy (&r.router_address, &ip0->src_address, 16); - r.current_hop_limit = h0->current_hop_limit; - r.flags = h0->flags; - r.router_lifetime_in_sec = - clib_net_to_host_u16 (h0->router_lifetime_in_sec); - r.neighbor_reachable_time_in_msec = - clib_net_to_host_u32 - (h0->neighbor_reachable_time_in_msec); - r.time_in_msec_between_retransmitted_neighbor_solicitations = clib_net_to_host_u32 (h0->time_in_msec_between_retransmitted_neighbor_solicitations); - r.prefixes = 0; - - /* validate advertised information */ - if ((h0->current_hop_limit && radv_info->curr_hop_limit) - && (h0->current_hop_limit != - radv_info->curr_hop_limit)) - { - ip6_neighbor_syslog (vm, LOG_WARNING, - "our AdvCurHopLimit on %U doesn't agree with %U", - format_vnet_sw_if_index_name, - vnm, sw_if_index0, - format_ip6_address, - &ip0->src_address); - } - - if ((h0->flags & - ICMP6_ROUTER_DISCOVERY_FLAG_ADDRESS_CONFIG_VIA_DHCP) - != radv_info->adv_managed_flag) - { - ip6_neighbor_syslog (vm, LOG_WARNING, - "our AdvManagedFlag on %U doesn't agree with %U", - format_vnet_sw_if_index_name, - vnm, sw_if_index0, - format_ip6_address, - &ip0->src_address); - } - - if ((h0->flags & - ICMP6_ROUTER_DISCOVERY_FLAG_OTHER_CONFIG_VIA_DHCP) - != radv_info->adv_other_flag) - { - ip6_neighbor_syslog (vm, LOG_WARNING, - "our AdvOtherConfigFlag on %U doesn't agree with %U", - format_vnet_sw_if_index_name, - vnm, sw_if_index0, - format_ip6_address, - &ip0->src_address); - } - - if ((h0-> - time_in_msec_between_retransmitted_neighbor_solicitations - && radv_info-> - adv_time_in_msec_between_retransmitted_neighbor_solicitations) - && (h0-> - time_in_msec_between_retransmitted_neighbor_solicitations - != - clib_host_to_net_u32 (radv_info-> - adv_time_in_msec_between_retransmitted_neighbor_solicitations))) - { - ip6_neighbor_syslog (vm, LOG_WARNING, - "our AdvRetransTimer on %U doesn't agree with %U", - format_vnet_sw_if_index_name, - vnm, sw_if_index0, - format_ip6_address, - &ip0->src_address); - } - - if ((h0->neighbor_reachable_time_in_msec && - radv_info->adv_neighbor_reachable_time_in_msec) && - (h0->neighbor_reachable_time_in_msec != - clib_host_to_net_u32 - (radv_info->adv_neighbor_reachable_time_in_msec))) - { - ip6_neighbor_syslog (vm, LOG_WARNING, - "our AdvReachableTime on %U doesn't agree with %U", - format_vnet_sw_if_index_name, - vnm, sw_if_index0, - format_ip6_address, - &ip0->src_address); - } - - /* check for MTU or prefix options or .. */ - u8 *opt_hdr = (u8 *) (h0 + 1); - while (options_len0 > 0) - { - icmp6_neighbor_discovery_option_header_t *o0 = - (icmp6_neighbor_discovery_option_header_t *) - opt_hdr; - int opt_len = o0->n_data_u64s << 3; - icmp6_neighbor_discovery_option_type_t option_type = - o0->type; - - if (options_len0 < 2) - { - ip6_neighbor_syslog (vm, LOG_ERR, - "malformed RA packet on %U from %U", - format_vnet_sw_if_index_name, - vnm, sw_if_index0, - format_ip6_address, - &ip0->src_address); - break; - } - - if (opt_len == 0) - { - ip6_neighbor_syslog (vm, LOG_ERR, - " zero length option in RA on %U from %U", - format_vnet_sw_if_index_name, - vnm, sw_if_index0, - format_ip6_address, - &ip0->src_address); - break; - } - else if (opt_len > options_len0) - { - ip6_neighbor_syslog (vm, LOG_ERR, - "option length in RA packet greater than total length on %U from %U", - format_vnet_sw_if_index_name, - vnm, sw_if_index0, - format_ip6_address, - &ip0->src_address); - break; - } - - options_len0 -= opt_len; - opt_hdr += opt_len; - - switch (option_type) - { - case ICMP6_NEIGHBOR_DISCOVERY_OPTION_source_link_layer_address: - { - icmp6_neighbor_discovery_ethernet_link_layer_address_option_t - * h = - (icmp6_neighbor_discovery_ethernet_link_layer_address_option_t - *) (o0); - - if (opt_len < sizeof (*h)) - break; - - memcpy (r.slla, h->ethernet_address, 6); - } - break; - - case ICMP6_NEIGHBOR_DISCOVERY_OPTION_mtu: - { - icmp6_neighbor_discovery_mtu_option_t *h = - (icmp6_neighbor_discovery_mtu_option_t - *) (o0); - - if (opt_len < sizeof (*h)) - break; - - r.mtu = clib_net_to_host_u32 (h->mtu); - - if ((h->mtu && radv_info->adv_link_mtu) && - (h->mtu != - clib_host_to_net_u32 - (radv_info->adv_link_mtu))) - { - ip6_neighbor_syslog (vm, LOG_WARNING, - "our AdvLinkMTU on %U doesn't agree with %U", - format_vnet_sw_if_index_name, - vnm, sw_if_index0, - format_ip6_address, - &ip0->src_address); - } - } - break; - - case ICMP6_NEIGHBOR_DISCOVERY_OPTION_prefix_information: - { - icmp6_neighbor_discovery_prefix_information_option_t - * h = - (icmp6_neighbor_discovery_prefix_information_option_t - *) (o0); - - /* validate advertised prefix options */ - ip6_radv_prefix_t *pr_info; - u32 preferred, valid; - - if (opt_len < sizeof (*h)) - break; - - vec_validate (r.prefixes, - vec_len (r.prefixes)); - ra_report_prefix_info_t *prefix = - vec_elt_at_index (r.prefixes, - vec_len (r.prefixes) - 1); - - preferred = - clib_net_to_host_u32 (h->preferred_time); - valid = clib_net_to_host_u32 (h->valid_time); - - prefix->preferred_time = preferred; - prefix->valid_time = valid; - prefix->flags = h->flags & 0xc0; - prefix->prefix.fp_len = h->dst_address_length; - prefix->prefix.fp_addr.ip6 = h->dst_address; - prefix->prefix.fp_proto = FIB_PROTOCOL_IP6; - - /* look for matching prefix - if we our advertising it, it better be consistant */ - /* *INDENT-OFF* */ - pool_foreach (pr_info, radv_info->adv_prefixes_pool, - ({ - - ip6_address_t mask; - ip6_address_mask_from_width(&mask, pr_info->prefix_len); - - if(pr_info->enabled && - (pr_info->prefix_len == h->dst_address_length) && - ip6_address_is_equal_masked (&pr_info->prefix, &h->dst_address, &mask)) - { - /* found it */ - if(!pr_info->decrement_lifetime_flag && - valid != pr_info->adv_valid_lifetime_in_secs) - { - ip6_neighbor_syslog(vm, LOG_WARNING, - "our ADV validlifetime on %U for %U does not agree with %U", - format_vnet_sw_if_index_name, vnm, sw_if_index0,format_ip6_address, &pr_info->prefix, - format_ip6_address, &h->dst_address); - } - if(!pr_info->decrement_lifetime_flag && - preferred != pr_info->adv_pref_lifetime_in_secs) - { - ip6_neighbor_syslog(vm, LOG_WARNING, - "our ADV preferredlifetime on %U for %U does not agree with %U", - format_vnet_sw_if_index_name, vnm, sw_if_index0,format_ip6_address, &pr_info->prefix, - format_ip6_address, &h->dst_address); - } - } - break; - })); - /* *INDENT-ON* */ - break; - } - default: - /* skip this one */ - break; - } - } - ra_publish (&r); - } - } - } - - p0->error = error_node->errors[error0]; - - if (error0 != ICMP6_ERROR_NONE) - vlib_error_count (vm, error_node->node_index, error0, 1); - - vlib_validate_buffer_enqueue_x1 (vm, node, next_index, - to_next, n_left_to_next, - bi0, next0); - } - - vlib_put_next_frame (vm, node, next_index, n_left_to_next); - } - - /* Account for router advertisements received. */ - vlib_error_count (vm, error_node->node_index, - ICMP6_ERROR_ROUTER_ADVERTISEMENTS_RX, - n_advertisements_rcvd); - - return frame->n_vectors; -} - -static inline f64 -random_f64_from_to (f64 from, f64 to) -{ - static u32 seed = 0; - static u8 seed_set = 0; - if (!seed_set) - { - seed = random_default_seed (); - seed_set = 1; - } - return random_f64 (&seed) * (to - from) + from; -} - -static inline u8 -get_mac_address (u32 sw_if_index, u8 * address) -{ - vnet_main_t *vnm = vnet_get_main (); - vnet_hw_interface_t *hw_if = vnet_get_sup_hw_interface (vnm, sw_if_index); - if (!hw_if->hw_address) - return 1; - clib_memcpy (address, hw_if->hw_address, 6); - return 0; -} - -static inline vlib_buffer_t * -create_buffer_for_rs (vlib_main_t * vm, ip6_radv_t * radv_info) -{ - u32 bi0; - vlib_buffer_t *p0; - icmp6_router_solicitation_header_t *rh; - u16 payload_length; - int bogus_length; - u32 sw_if_index; - - sw_if_index = radv_info->sw_if_index; - - if (vlib_buffer_alloc (vm, &bi0, 1) != 1) - { - clib_warning ("buffer allocation failure"); - return 0; - } - - p0 = vlib_get_buffer (vm, bi0); - VLIB_BUFFER_TRACE_TRAJECTORY_INIT (p0); - p0->flags |= VNET_BUFFER_F_LOCALLY_ORIGINATED; - - vnet_buffer (p0)->sw_if_index[VLIB_RX] = sw_if_index; - vnet_buffer (p0)->sw_if_index[VLIB_TX] = sw_if_index; - - vnet_buffer (p0)->ip.adj_index[VLIB_TX] = radv_info->mcast_adj_index; - - rh = vlib_buffer_get_current (p0); - p0->current_length = sizeof (*rh); - - rh->neighbor.icmp.type = ICMP6_router_solicitation; - rh->neighbor.icmp.code = 0; - rh->neighbor.icmp.checksum = 0; - rh->neighbor.reserved_must_be_zero = 0; - - rh->link_layer_option.header.type = - ICMP6_NEIGHBOR_DISCOVERY_OPTION_source_link_layer_address; - if (0 != get_mac_address (sw_if_index, - rh->link_layer_option.ethernet_address)) - { - clib_warning ("interface with sw_if_index %u has no mac address", - sw_if_index); - vlib_buffer_free (vm, &bi0, 1); - return 0; - } - rh->link_layer_option.header.n_data_u64s = 1; - - payload_length = sizeof (rh->neighbor) + sizeof (u64); - - rh->ip.ip_version_traffic_class_and_flow_label = - clib_host_to_net_u32 (0x6 << 28); - rh->ip.payload_length = clib_host_to_net_u16 (payload_length); - rh->ip.protocol = IP_PROTOCOL_ICMP6; - rh->ip.hop_limit = 255; - rh->ip.src_address = radv_info->link_local_address; - /* set address ff02::2 */ - rh->ip.dst_address.as_u64[0] = clib_host_to_net_u64 (0xff02ULL << 48); - rh->ip.dst_address.as_u64[1] = clib_host_to_net_u64 (2); - - rh->neighbor.icmp.checksum = ip6_tcp_udp_icmp_compute_checksum (vm, p0, - &rh->ip, - &bogus_length); - - return p0; -} - -static inline void -stop_sending_rs (vlib_main_t * vm, ip6_radv_t * ra) -{ - u32 bi0; - - ra->keep_sending_rs = 0; - if (ra->buffer) - { - bi0 = vlib_get_buffer_index (vm, ra->buffer); - vlib_buffer_free (vm, &bi0, 1); - ra->buffer = 0; - } -} - -static inline bool -check_send_rs (vlib_main_t * vm, ip6_radv_t * radv_info, f64 current_time, - f64 * due_time) -{ - vlib_buffer_t *p0; - vlib_frame_t *f; - u32 *to_next; - u32 next_index; - vlib_buffer_t *c0; - u32 ci0; - - icmp6_send_router_solicitation_params_t *params; - - if (!radv_info->keep_sending_rs) - return false; - - params = &radv_info->params; - - if (radv_info->due_time > current_time) - { - *due_time = radv_info->due_time; - return true; - } - - p0 = radv_info->buffer; - - next_index = ip6_rewrite_mcast_node.index; - - c0 = vlib_buffer_copy (vm, p0); - ci0 = vlib_get_buffer_index (vm, c0); - - f = vlib_get_frame_to_node (vm, next_index); - to_next = vlib_frame_vector_args (f); - to_next[0] = ci0; - f->n_vectors = 1; - vlib_put_frame_to_node (vm, next_index, f); - - if (params->mrc != 0 && --radv_info->n_left == 0) - stop_sending_rs (vm, radv_info); - else - { - radv_info->sleep_interval = - (2 + random_f64_from_to (-0.1, 0.1)) * radv_info->sleep_interval; - if (radv_info->sleep_interval > params->mrt) - radv_info->sleep_interval = - (1 + random_f64_from_to (-0.1, 0.1)) * params->mrt; - - radv_info->due_time = current_time + radv_info->sleep_interval; - - if (params->mrd != 0 - && current_time > radv_info->start_time + params->mrd) - stop_sending_rs (vm, radv_info); - else - *due_time = radv_info->due_time; - } - - return radv_info->keep_sending_rs; -} - -static uword -send_rs_process (vlib_main_t * vm, vlib_node_runtime_t * rt, - vlib_frame_t * f0) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_radv_t *radv_info; - uword *event_data = 0; - f64 sleep_time = 1e9; - f64 current_time; - f64 due_time; - f64 dt = 0; - - while (true) - { - vlib_process_wait_for_event_or_clock (vm, sleep_time); - vlib_process_get_events (vm, &event_data); - vec_reset_length (event_data); - - current_time = vlib_time_now (vm); - do - { - due_time = current_time + 1e9; - /* *INDENT-OFF* */ - pool_foreach (radv_info, nm->if_radv_pool, - ({ - if (check_send_rs (vm, radv_info, current_time, &dt) - && (dt < due_time)) - due_time = dt; - })); - /* *INDENT-ON* */ - current_time = vlib_time_now (vm); - } - while (due_time < current_time); - - sleep_time = due_time - current_time; - } - - return 0; -} - -/* *INDENT-OFF* */ -VLIB_REGISTER_NODE (send_rs_process_node) = { - .function = send_rs_process, - .type = VLIB_NODE_TYPE_PROCESS, - .name = "send-rs-process", -}; -/* *INDENT-ON* */ - -void -icmp6_send_router_solicitation (vlib_main_t * vm, u32 sw_if_index, u8 stop, - icmp6_send_router_solicitation_params_t * - params) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - u32 rai; - ip6_radv_t *ra = 0; - - ASSERT (~0 != sw_if_index); - - rai = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - ra = pool_elt_at_index (nm->if_radv_pool, rai); - - stop_sending_rs (vm, ra); - - if (!stop) - { - ra->keep_sending_rs = 1; - ra->params = *params; - ra->n_left = params->mrc; - ra->start_time = vlib_time_now (vm); - ra->sleep_interval = (1 + random_f64_from_to (-0.1, 0.1)) * params->irt; - ra->due_time = 0; /* send first packet ASAP */ - ra->buffer = create_buffer_for_rs (vm, ra); - if (!ra->buffer) - ra->keep_sending_rs = 0; - else - vlib_process_signal_event (vm, send_rs_process_node.index, 1, 0); - } -} - -/** - * @brief Add a multicast Address to the advertised MLD set - */ -static void -ip6_neighbor_add_mld_prefix (ip6_radv_t * radv_info, ip6_address_t * addr) -{ - ip6_mldp_group_t *mcast_group_info; - uword *p; - - /* lookup mldp info for this interface */ - p = mhash_get (&radv_info->address_to_mldp_index, addr); - mcast_group_info = - p ? pool_elt_at_index (radv_info->mldp_group_pool, p[0]) : 0; - - /* add address */ - if (!mcast_group_info) - { - /* add */ - u32 mi; - pool_get (radv_info->mldp_group_pool, mcast_group_info); - memset (mcast_group_info, 0, sizeof (*mcast_group_info)); - - mi = mcast_group_info - radv_info->mldp_group_pool; - mhash_set (&radv_info->address_to_mldp_index, addr, mi, /* old_value */ - 0); - - mcast_group_info->type = 4; - mcast_group_info->mcast_source_address_pool = 0; - mcast_group_info->num_sources = 0; - clib_memcpy (&mcast_group_info->mcast_address, addr, - sizeof (ip6_address_t)); - } -} - -/** - * @brief Delete a multicast Address from the advertised MLD set - */ -static void -ip6_neighbor_del_mld_prefix (ip6_radv_t * radv_info, ip6_address_t * addr) -{ - ip6_mldp_group_t *mcast_group_info; - uword *p; - - p = mhash_get (&radv_info->address_to_mldp_index, &addr); - mcast_group_info = - p ? pool_elt_at_index (radv_info->mldp_group_pool, p[0]) : 0; - - if (mcast_group_info) - { - mhash_unset (&radv_info->address_to_mldp_index, &addr, - /* old_value */ 0); - pool_put (radv_info->mldp_group_pool, mcast_group_info); - } -} - -/** - * @brief Add a multicast Address to the advertised MLD set - */ -static void -ip6_neighbor_add_mld_grp (ip6_radv_t * a, - ip6_multicast_address_scope_t scope, - ip6_multicast_link_local_group_id_t group) -{ - ip6_address_t addr; - - ip6_set_reserved_multicast_address (&addr, scope, group); - - ip6_neighbor_add_mld_prefix (a, &addr); -} - -/** - * @brief create and initialize router advertisement parameters with default - * values for this intfc - */ -u32 -ip6_neighbor_sw_interface_add_del (vnet_main_t * vnm, - u32 sw_if_index, u32 is_add) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_radv_t *a = 0; - u32 ri = ~0; - vnet_sw_interface_t *sw_if0; - ethernet_interface_t *eth_if0 = 0; - - /* lookup radv container - ethernet interfaces only */ - sw_if0 = vnet_get_sup_sw_interface (vnm, sw_if_index); - if (sw_if0->type == VNET_SW_INTERFACE_TYPE_HARDWARE) - eth_if0 = ethernet_get_interface (ðernet_main, sw_if0->hw_if_index); - - if (!eth_if0) - return ri; - - vec_validate_init_empty (nm->if_radv_pool_index_by_sw_if_index, sw_if_index, - ~0); - ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - - if (ri != ~0) - { - a = pool_elt_at_index (nm->if_radv_pool, ri); - - if (!is_add) - { - ip6_radv_prefix_t *p; - ip6_mldp_group_t *m; - - /* release the lock on the interface's mcast adj */ - adj_unlock (a->mcast_adj_index); - - /* clean up prefix and MDP pools */ - /* *INDENT-OFF* */ - pool_flush(p, a->adv_prefixes_pool, - ({ - mhash_unset (&a->address_to_prefix_index, &p->prefix, 0); - })); - pool_flush (m, a->mldp_group_pool, - ({ - mhash_unset (&a->address_to_mldp_index, &m->mcast_address, 0); - })); - /* *INDENT-ON* */ - - pool_free (a->mldp_group_pool); - pool_free (a->adv_prefixes_pool); - - mhash_free (&a->address_to_prefix_index); - mhash_free (&a->address_to_mldp_index); - - if (a->keep_sending_rs) - a->keep_sending_rs = 0; - - pool_put (nm->if_radv_pool, a); - nm->if_radv_pool_index_by_sw_if_index[sw_if_index] = ~0; - ri = ~0; - } - } - else - { - if (is_add) - { - pool_get (nm->if_radv_pool, a); - memset (a, 0, sizeof (*a)); - - ri = a - nm->if_radv_pool; - nm->if_radv_pool_index_by_sw_if_index[sw_if_index] = ri; - - /* initialize default values (most of which are zero) */ - clib_memset (a, 0, sizeof (a[0])); - - a->sw_if_index = sw_if_index; - a->max_radv_interval = DEF_MAX_RADV_INTERVAL; - a->min_radv_interval = DEF_MIN_RADV_INTERVAL; - a->curr_hop_limit = DEF_CURR_HOP_LIMIT; - a->adv_router_lifetime_in_sec = DEF_DEF_RTR_LIFETIME; - - /* send ll address source address option */ - a->adv_link_layer_address = 1; - - a->min_delay_between_radv = MIN_DELAY_BETWEEN_RAS; - a->max_delay_between_radv = MAX_DELAY_BETWEEN_RAS; - a->max_rtr_default_lifetime = MAX_DEF_RTR_LIFETIME; - a->seed = (u32) clib_cpu_time_now (); - (void) random_u32 (&a->seed); - a->randomizer = clib_cpu_time_now (); - (void) random_u64 (&a->randomizer); - - a->initial_adverts_count = MAX_INITIAL_RTR_ADVERTISEMENTS; - a->initial_adverts_sent = a->initial_adverts_count - 1; - a->initial_adverts_interval = MAX_INITIAL_RTR_ADVERT_INTERVAL; - - /* deafult is to send */ - a->send_radv = 1; - - /* fill in radv_info for this interface that will be needed later */ - a->adv_link_mtu = - vnet_sw_interface_get_mtu (vnm, sw_if_index, VNET_MTU_IP6); - - clib_memcpy (a->link_layer_address, eth_if0->address, 6); - - /* fill in default link-local address (this may be overridden) */ - ip6_link_local_address_from_ethernet_mac_address - (&a->link_local_address, eth_if0->address); - - mhash_init (&a->address_to_prefix_index, sizeof (uword), - sizeof (ip6_address_t)); - mhash_init (&a->address_to_mldp_index, sizeof (uword), - sizeof (ip6_address_t)); - - a->mcast_adj_index = adj_mcast_add_or_lock (FIB_PROTOCOL_IP6, - VNET_LINK_IP6, - sw_if_index); - - a->keep_sending_rs = 0; - - /* add multicast groups we will always be reporting */ - ip6_neighbor_add_mld_grp (a, - IP6_MULTICAST_SCOPE_link_local, - IP6_MULTICAST_GROUP_ID_all_hosts); - ip6_neighbor_add_mld_grp (a, - IP6_MULTICAST_SCOPE_link_local, - IP6_MULTICAST_GROUP_ID_all_routers); - ip6_neighbor_add_mld_grp (a, - IP6_MULTICAST_SCOPE_link_local, - IP6_MULTICAST_GROUP_ID_mldv2_routers); - } - } - return ri; -} - -/* send an mldpv2 report */ -static void -ip6_neighbor_send_mldpv2_report (u32 sw_if_index) -{ - vnet_main_t *vnm = vnet_get_main (); - vlib_main_t *vm = vnm->vlib_main; - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - vnet_sw_interface_t *sw_if0; - ethernet_interface_t *eth_if0; - u32 ri; - int bogus_length; - - ip6_radv_t *radv_info; - u16 payload_length; - vlib_buffer_t *b0; - ip6_header_t *ip0; - u32 *to_next; - vlib_frame_t *f; - u32 bo0; - u32 n_to_alloc = 1; - u32 n_allocated; - - icmp6_multicast_listener_report_header_t *rh0; - icmp6_multicast_listener_report_packet_t *rp0; - - sw_if0 = vnet_get_sup_sw_interface (vnm, sw_if_index); - ASSERT (sw_if0->type == VNET_SW_INTERFACE_TYPE_HARDWARE); - eth_if0 = ethernet_get_interface (ðernet_main, sw_if0->hw_if_index); - - if (!eth_if0 || !vnet_sw_interface_is_admin_up (vnm, sw_if_index)) - return; - - /* look up the radv_t information for this interface */ - vec_validate_init_empty (nm->if_radv_pool_index_by_sw_if_index, sw_if_index, - ~0); - - ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - - if (ri == ~0) - return; - - /* send report now - build a mldpv2 report packet */ - n_allocated = vlib_buffer_alloc (vm, &bo0, n_to_alloc); - if (PREDICT_FALSE (n_allocated == 0)) - { - alloc_fail: - clib_warning ("buffer allocation failure"); - return; - } - - b0 = vlib_get_buffer (vm, bo0); - - /* adjust the sizeof the buffer to just include the ipv6 header */ - b0->current_length = sizeof (icmp6_multicast_listener_report_packet_t); - - payload_length = sizeof (icmp6_multicast_listener_report_header_t); - - b0->error = ICMP6_ERROR_NONE; - - rp0 = vlib_buffer_get_current (b0); - ip0 = (ip6_header_t *) & rp0->ip; - rh0 = (icmp6_multicast_listener_report_header_t *) & rp0->report_hdr; - - clib_memset (rp0, 0x0, sizeof (icmp6_multicast_listener_report_packet_t)); - - ip0->ip_version_traffic_class_and_flow_label = - clib_host_to_net_u32 (0x6 << 28); - - ip0->protocol = IP_PROTOCOL_IP6_HOP_BY_HOP_OPTIONS; - /* for DEBUG - vnet driver won't seem to emit router alerts */ - /* ip0->protocol = IP_PROTOCOL_ICMP6; */ - ip0->hop_limit = 1; - - rh0->icmp.type = ICMP6_multicast_listener_report_v2; - - /* source address MUST be the link-local address */ - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - ip0->src_address = radv_info->link_local_address; - - /* destination is all mldpv2 routers */ - ip6_set_reserved_multicast_address (&ip0->dst_address, - IP6_MULTICAST_SCOPE_link_local, - IP6_MULTICAST_GROUP_ID_mldv2_routers); - - /* add reports here */ - ip6_mldp_group_t *m; - int num_addr_records = 0; - icmp6_multicast_address_record_t rr; - - /* fill in the hop-by-hop extension header (router alert) info */ - rh0->ext_hdr.next_hdr = IP_PROTOCOL_ICMP6; - rh0->ext_hdr.n_data_u64s = 0; - - rh0->alert.type = IP6_MLDP_ALERT_TYPE; - rh0->alert.len = 2; - rh0->alert.value = 0; - - rh0->pad.type = 1; - rh0->pad.len = 0; - - rh0->icmp.checksum = 0; - - /* *INDENT-OFF* */ - pool_foreach (m, radv_info->mldp_group_pool, - ({ - rr.type = m->type; - rr.aux_data_len_u32s = 0; - rr.num_sources = clib_host_to_net_u16 (m->num_sources); - clib_memcpy(&rr.mcast_addr, &m->mcast_address, sizeof(ip6_address_t)); - - num_addr_records++; - - if(vlib_buffer_add_data (vm, &bo0, (void *)&rr, - sizeof(icmp6_multicast_address_record_t))) - { - vlib_buffer_free (vm, &bo0, 1); - goto alloc_fail; - } - - payload_length += sizeof( icmp6_multicast_address_record_t); - })); - /* *INDENT-ON* */ - - rh0->rsvd = 0; - rh0->num_addr_records = clib_host_to_net_u16 (num_addr_records); - - /* update lengths */ - ip0->payload_length = clib_host_to_net_u16 (payload_length); - - rh0->icmp.checksum = ip6_tcp_udp_icmp_compute_checksum (vm, b0, ip0, - &bogus_length); - ASSERT (bogus_length == 0); - - /* - * OK to override w/ no regard for actual FIB, because - * ip6-rewrite only looks at the adjacency. - */ - vnet_buffer (b0)->sw_if_index[VLIB_RX] = - vnet_main.local_interface_sw_if_index; - - vnet_buffer (b0)->ip.adj_index[VLIB_TX] = radv_info->mcast_adj_index; - b0->flags |= VNET_BUFFER_F_LOCALLY_ORIGINATED; - - vlib_node_t *node = vlib_get_node_by_name (vm, (u8 *) "ip6-rewrite-mcast"); - - f = vlib_get_frame_to_node (vm, node->index); - to_next = vlib_frame_vector_args (f); - to_next[0] = bo0; - f->n_vectors = 1; - - vlib_put_frame_to_node (vm, node->index, f); - return; -} - -/* *INDENT-OFF* */ -VLIB_REGISTER_NODE (ip6_icmp_router_solicitation_node,static) = -{ - .function = icmp6_router_solicitation, - .name = "icmp6-router-solicitation", - - .vector_size = sizeof (u32), - - .format_trace = format_icmp6_input_trace, - - .n_next_nodes = ICMP6_ROUTER_SOLICITATION_N_NEXT, - .next_nodes = { - [ICMP6_ROUTER_SOLICITATION_NEXT_DROP] = "ip6-drop", - [ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_RW] = "ip6-rewrite-mcast", - [ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_TX] = "interface-output", - }, -}; -/* *INDENT-ON* */ - -/* send a RA or update the timer info etc.. */ -static uword -ip6_neighbor_process_timer_event (vlib_main_t * vm, - vlib_node_runtime_t * node, - vlib_frame_t * frame) -{ - vnet_main_t *vnm = vnet_get_main (); - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_radv_t *radv_info; - vlib_frame_t *f = 0; - u32 n_this_frame = 0; - u32 n_left_to_next = 0; - u32 *to_next = 0; - u32 bo0; - icmp6_router_solicitation_header_t *h0; - vlib_buffer_t *b0; - f64 now = vlib_time_now (vm); - - /* Interface ip6 radv info list */ - /* *INDENT-OFF* */ - pool_foreach (radv_info, nm->if_radv_pool, - ({ - if( !vnet_sw_interface_is_admin_up (vnm, radv_info->sw_if_index)) - { - radv_info->initial_adverts_sent = radv_info->initial_adverts_count-1; - radv_info->next_multicast_time = now; - radv_info->last_multicast_time = now; - radv_info->last_radv_time = 0; - radv_info->all_routers_mcast = 0; - continue; - } - - /* Make sure that we've joined the all-routers multicast group */ - if(!radv_info->all_routers_mcast) - { - /* send MDLP_REPORT_EVENT message */ - ip6_neighbor_send_mldpv2_report(radv_info->sw_if_index); - radv_info->all_routers_mcast = 1; - } - - /* is it time to send a multicast RA on this interface? */ - if(radv_info->send_radv && (now >= radv_info->next_multicast_time)) - { - u32 n_to_alloc = 1; - u32 n_allocated; - - f64 rfn = (radv_info->max_radv_interval - radv_info->min_radv_interval) * - random_f64 (&radv_info->seed) + radv_info->min_radv_interval; - - /* multicast send - compute next multicast send time */ - if( radv_info->initial_adverts_sent > 0) - { - radv_info->initial_adverts_sent--; - if(rfn > radv_info-> initial_adverts_interval) - rfn = radv_info-> initial_adverts_interval; - - /* check to see if we are ceasing to send */ - if( radv_info->initial_adverts_sent == 0) - if(radv_info->cease_radv) - radv_info->send_radv = 0; - } - - radv_info->next_multicast_time = rfn + now; - radv_info->last_multicast_time = now; - - /* send advert now - build a "solicted" router advert with unspecified source address */ - n_allocated = vlib_buffer_alloc (vm, &bo0, n_to_alloc); - - if (PREDICT_FALSE(n_allocated == 0)) - { - clib_warning ("buffer allocation failure"); - continue; - } - b0 = vlib_get_buffer (vm, bo0); - b0->current_length = sizeof( icmp6_router_solicitation_header_t); - b0->error = ICMP6_ERROR_NONE; - vnet_buffer (b0)->sw_if_index[VLIB_RX] = radv_info->sw_if_index; - - h0 = vlib_buffer_get_current (b0); - - clib_memset (h0, 0, sizeof (icmp6_router_solicitation_header_t)); - - h0->ip.ip_version_traffic_class_and_flow_label = clib_host_to_net_u32 (0x6 << 28); - h0->ip.payload_length = clib_host_to_net_u16 (sizeof (icmp6_router_solicitation_header_t) - - STRUCT_OFFSET_OF (icmp6_router_solicitation_header_t, neighbor)); - h0->ip.protocol = IP_PROTOCOL_ICMP6; - h0->ip.hop_limit = 255; - - /* set src/dst address as "unspecified" this marks this packet as internally generated rather than recieved */ - h0->ip.src_address.as_u64[0] = 0; - h0->ip.src_address.as_u64[1] = 0; - - h0->ip.dst_address.as_u64[0] = 0; - h0->ip.dst_address.as_u64[1] = 0; - - h0->neighbor.icmp.type = ICMP6_router_solicitation; - - if (PREDICT_FALSE(f == 0)) - { - f = vlib_get_frame_to_node (vm, ip6_icmp_router_solicitation_node.index); - to_next = vlib_frame_vector_args (f); - n_left_to_next = VLIB_FRAME_SIZE; - n_this_frame = 0; - } - - n_this_frame++; - n_left_to_next--; - to_next[0] = bo0; - to_next += 1; - - if (PREDICT_FALSE(n_left_to_next == 0)) - { - f->n_vectors = n_this_frame; - vlib_put_frame_to_node (vm, ip6_icmp_router_solicitation_node.index, f); - f = 0; - } - } - })); - /* *INDENT-ON* */ - - if (f) - { - ASSERT (n_this_frame); - f->n_vectors = n_this_frame; - vlib_put_frame_to_node (vm, ip6_icmp_router_solicitation_node.index, f); - } - return 0; -} - -static uword -ip6_icmp_neighbor_discovery_event_process (vlib_main_t * vm, - vlib_node_runtime_t * node, - vlib_frame_t * frame) -{ - uword event_type; - ip6_icmp_neighbor_discovery_event_data_t *event_data; - - /* init code here */ - - while (1) - { - vlib_process_wait_for_event_or_clock (vm, 1. /* seconds */ ); - - event_data = vlib_process_get_event_data (vm, &event_type); - - if (!event_data) - { - /* No events found: timer expired. */ - /* process interface list and send RAs as appropriate, update timer info */ - ip6_neighbor_process_timer_event (vm, node, frame); - } - else - { - switch (event_type) - { - - case ICMP6_ND_EVENT_INIT: - break; - - case ~0: - break; - - default: - ASSERT (0); - } - - if (event_data) - _vec_len (event_data) = 0; - } - } - return frame->n_vectors; -} - -/* *INDENT-OFF* */ -VLIB_REGISTER_NODE (ip6_icmp_router_advertisement_node,static) = -{ - .function = icmp6_router_advertisement, - .name = "icmp6-router-advertisement", - - .vector_size = sizeof (u32), - - .format_trace = format_icmp6_input_trace, - - .n_next_nodes = 1, - .next_nodes = { - [0] = "ip6-drop", - }, -}; -/* *INDENT-ON* */ - -vlib_node_registration_t ip6_icmp_neighbor_discovery_event_node = { - - .function = ip6_icmp_neighbor_discovery_event_process, - .name = "ip6-icmp-neighbor-discovery-event-process", - .type = VLIB_NODE_TYPE_PROCESS, -}; - -static uword -icmp6_neighbor_solicitation (vlib_main_t * vm, - vlib_node_runtime_t * node, vlib_frame_t * frame) -{ - return icmp6_neighbor_solicitation_or_advertisement (vm, node, frame, - /* is_solicitation */ - 1); -} - -static uword -icmp6_neighbor_advertisement (vlib_main_t * vm, - vlib_node_runtime_t * node, - vlib_frame_t * frame) -{ - return icmp6_neighbor_solicitation_or_advertisement (vm, node, frame, - /* is_solicitation */ - 0); -} - -/* *INDENT-OFF* */ -VLIB_REGISTER_NODE (ip6_icmp_neighbor_solicitation_node,static) = -{ - .function = icmp6_neighbor_solicitation, - .name = "icmp6-neighbor-solicitation", - - .vector_size = sizeof (u32), - - .format_trace = format_icmp6_input_trace, - - .n_next_nodes = ICMP6_NEIGHBOR_SOLICITATION_N_NEXT, - .next_nodes = { - [ICMP6_NEIGHBOR_SOLICITATION_NEXT_DROP] = "ip6-drop", - [ICMP6_NEIGHBOR_SOLICITATION_NEXT_REPLY] = "interface-output", - }, -}; -/* *INDENT-ON* */ - -/* *INDENT-OFF* */ -VLIB_REGISTER_NODE (ip6_icmp_neighbor_advertisement_node,static) = -{ - .function = icmp6_neighbor_advertisement, - .name = "icmp6-neighbor-advertisement", - - .vector_size = sizeof (u32), - - .format_trace = format_icmp6_input_trace, - - .n_next_nodes = 1, - .next_nodes = { - [0] = "ip6-drop", - }, -}; -/* *INDENT-ON* */ - -typedef enum -{ - IP6_DISCOVER_NEIGHBOR_NEXT_DROP, - IP6_DISCOVER_NEIGHBOR_NEXT_REPLY_TX, - IP6_DISCOVER_NEIGHBOR_N_NEXT, -} ip6_discover_neighbor_next_t; - -typedef enum -{ - IP6_DISCOVER_NEIGHBOR_ERROR_DROP, - IP6_DISCOVER_NEIGHBOR_ERROR_REQUEST_SENT, - IP6_DISCOVER_NEIGHBOR_ERROR_NO_SOURCE_ADDRESS, -} ip6_discover_neighbor_error_t; - -static uword -ip6_discover_neighbor_inline (vlib_main_t * vm, - vlib_node_runtime_t * node, - vlib_frame_t * frame, int is_glean) -{ - vnet_main_t *vnm = vnet_get_main (); - ip6_main_t *im = &ip6_main; - ip_lookup_main_t *lm = &im->lookup_main; - u32 *from, *to_next_drop; - uword n_left_from, n_left_to_next_drop; - u64 seed; - u32 thread_index = vm->thread_index; - int bogus_length; - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - - if (node->flags & VLIB_NODE_FLAG_TRACE) - ip6_forward_next_trace (vm, node, frame, VLIB_TX); - - seed = throttle_seed (&im->nd_throttle, thread_index, vlib_time_now (vm)); - - from = vlib_frame_vector_args (frame); - n_left_from = frame->n_vectors; - - while (n_left_from > 0) - { - vlib_get_next_frame (vm, node, IP6_DISCOVER_NEIGHBOR_NEXT_DROP, - to_next_drop, n_left_to_next_drop); - - while (n_left_from > 0 && n_left_to_next_drop > 0) - { - u32 pi0, adj_index0, sw_if_index0, drop0, r0, next0; - vnet_hw_interface_t *hw_if0; - ip6_radv_t *radv_info; - ip_adjacency_t *adj0; - vlib_buffer_t *p0; - ip6_header_t *ip0; - - pi0 = from[0]; - - p0 = vlib_get_buffer (vm, pi0); - - adj_index0 = vnet_buffer (p0)->ip.adj_index[VLIB_TX]; - - ip0 = vlib_buffer_get_current (p0); - - adj0 = adj_get (adj_index0); - - if (!is_glean) - { - ip0->dst_address.as_u64[0] = - adj0->sub_type.nbr.next_hop.ip6.as_u64[0]; - ip0->dst_address.as_u64[1] = - adj0->sub_type.nbr.next_hop.ip6.as_u64[1]; - } - - sw_if_index0 = adj0->rewrite_header.sw_if_index; - vnet_buffer (p0)->sw_if_index[VLIB_TX] = sw_if_index0; - - /* combine the address and interface for a hash */ - r0 = ip6_address_hash_to_u64 (&ip0->dst_address) ^ sw_if_index0; - - drop0 = throttle_check (&im->nd_throttle, thread_index, r0, seed); - - from += 1; - n_left_from -= 1; - to_next_drop[0] = pi0; - to_next_drop += 1; - n_left_to_next_drop -= 1; - - hw_if0 = vnet_get_sup_hw_interface (vnm, sw_if_index0); - - /* If the interface is link-down, drop the pkt */ - if (!(hw_if0->flags & VNET_HW_INTERFACE_FLAG_LINK_UP)) - drop0 = 1; - - if (vec_len (nm->if_radv_pool_index_by_sw_if_index) > sw_if_index0) - { - u32 ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index0]; - - if (ri != ~0) - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - else - drop0 = 1; - } - else - drop0 = 1; - - /* - * 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) - drop0 = 1; - - p0->error = - node->errors[drop0 ? IP6_DISCOVER_NEIGHBOR_ERROR_DROP - : IP6_DISCOVER_NEIGHBOR_ERROR_REQUEST_SENT]; - - if (drop0) - continue; - - { - u32 bi0 = 0; - icmp6_neighbor_solicitation_header_t *h0; - vlib_buffer_t *b0; - - h0 = vlib_packet_template_get_packet - (vm, &im->discover_neighbor_packet_template, &bi0); - if (!h0) - continue; - - /* copy the persistent fields from the original */ - b0 = vlib_get_buffer (vm, bi0); - clib_memcpy_fast (b0->opaque2, p0->opaque2, sizeof (p0->opaque2)); - - /* - * Build ethernet header. - * Choose source address based on destination lookup - * adjacency. - */ - if (!ip6_src_address_for_packet (lm, - sw_if_index0, - &ip0->dst_address, - &h0->ip.src_address)) - { - /* There is no address on the interface */ - p0->error = - node->errors[IP6_DISCOVER_NEIGHBOR_ERROR_NO_SOURCE_ADDRESS]; - vlib_buffer_free (vm, &bi0, 1); - continue; - } - - /* - * Destination address is a solicited node multicast address. - * We need to fill in - * the low 24 bits with low 24 bits of target's address. - */ - h0->ip.dst_address.as_u8[13] = ip0->dst_address.as_u8[13]; - h0->ip.dst_address.as_u8[14] = ip0->dst_address.as_u8[14]; - h0->ip.dst_address.as_u8[15] = ip0->dst_address.as_u8[15]; - - h0->neighbor.target_address = ip0->dst_address; - - clib_memcpy (h0->link_layer_option.ethernet_address, - hw_if0->hw_address, vec_len (hw_if0->hw_address)); - - /* $$$$ appears we need this; why is the checksum non-zero? */ - h0->neighbor.icmp.checksum = 0; - h0->neighbor.icmp.checksum = - ip6_tcp_udp_icmp_compute_checksum (vm, 0, &h0->ip, - &bogus_length); - - ASSERT (bogus_length == 0); - - vlib_buffer_copy_trace_flag (vm, p0, bi0); - vnet_buffer (b0)->sw_if_index[VLIB_TX] - = vnet_buffer (p0)->sw_if_index[VLIB_TX]; - - vnet_buffer (b0)->ip.adj_index[VLIB_TX] = - radv_info->mcast_adj_index; - - b0->flags |= VNET_BUFFER_F_LOCALLY_ORIGINATED; - next0 = IP6_DISCOVER_NEIGHBOR_NEXT_REPLY_TX; - - vlib_set_next_frame_buffer (vm, node, next0, bi0); - } - } - - vlib_put_next_frame (vm, node, IP6_DISCOVER_NEIGHBOR_NEXT_DROP, - n_left_to_next_drop); - } - - return frame->n_vectors; -} - -static uword -ip6_discover_neighbor (vlib_main_t * vm, - vlib_node_runtime_t * node, vlib_frame_t * frame) -{ - return (ip6_discover_neighbor_inline (vm, node, frame, 0)); -} - -static uword -ip6_glean (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * frame) -{ - return (ip6_discover_neighbor_inline (vm, node, frame, 1)); -} - -static char *ip6_discover_neighbor_error_strings[] = { - [IP6_DISCOVER_NEIGHBOR_ERROR_DROP] = "address overflow drops", - [IP6_DISCOVER_NEIGHBOR_ERROR_REQUEST_SENT] = "neighbor solicitations sent", - [IP6_DISCOVER_NEIGHBOR_ERROR_NO_SOURCE_ADDRESS] - = "no source address for ND solicitation", -}; - -/* *INDENT-OFF* */ -VLIB_REGISTER_NODE (ip6_glean_node) = -{ - .function = ip6_glean, - .name = "ip6-glean", - .vector_size = sizeof (u32), - .format_trace = format_ip6_forward_next_trace, - .n_errors = ARRAY_LEN (ip6_discover_neighbor_error_strings), - .error_strings = ip6_discover_neighbor_error_strings, - .n_next_nodes = IP6_DISCOVER_NEIGHBOR_N_NEXT, - .next_nodes = - { - [IP6_DISCOVER_NEIGHBOR_NEXT_DROP] = "ip6-drop", - [IP6_DISCOVER_NEIGHBOR_NEXT_REPLY_TX] = "ip6-rewrite-mcast", - }, -}; -VLIB_REGISTER_NODE (ip6_discover_neighbor_node) = -{ - .function = ip6_discover_neighbor, - .name = "ip6-discover-neighbor", - .vector_size = sizeof (u32), - .format_trace = format_ip6_forward_next_trace, - .n_errors = ARRAY_LEN (ip6_discover_neighbor_error_strings), - .error_strings = ip6_discover_neighbor_error_strings, - .n_next_nodes = IP6_DISCOVER_NEIGHBOR_N_NEXT, - .next_nodes = - { - [IP6_DISCOVER_NEIGHBOR_NEXT_DROP] = "ip6-drop", - [IP6_DISCOVER_NEIGHBOR_NEXT_REPLY_TX] = "ip6-rewrite-mcast", - }, -}; -/* *INDENT-ON* */ - -/* API support functions */ -int -ip6_neighbor_ra_config (vlib_main_t * vm, u32 sw_if_index, - u8 suppress, u8 managed, u8 other, - u8 ll_option, u8 send_unicast, u8 cease, - u8 use_lifetime, u32 lifetime, - u32 initial_count, u32 initial_interval, - u32 max_interval, u32 min_interval, u8 is_no) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - int error; - u32 ri; - - /* look up the radv_t information for this interface */ - vec_validate_init_empty (nm->if_radv_pool_index_by_sw_if_index, sw_if_index, - ~0); - ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - error = (ri != ~0) ? 0 : VNET_API_ERROR_INVALID_SW_IF_INDEX; - - if (!error) - { - - ip6_radv_t *radv_info; - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - - if ((max_interval != 0) && (min_interval == 0)) - min_interval = .75 * max_interval; - - max_interval = - (max_interval != - 0) ? ((is_no) ? DEF_MAX_RADV_INTERVAL : max_interval) : - radv_info->max_radv_interval; - min_interval = - (min_interval != - 0) ? ((is_no) ? DEF_MIN_RADV_INTERVAL : min_interval) : - radv_info->min_radv_interval; - lifetime = - (use_lifetime != - 0) ? ((is_no) ? DEF_DEF_RTR_LIFETIME : lifetime) : - radv_info->adv_router_lifetime_in_sec; - - if (lifetime) - { - if (lifetime > MAX_DEF_RTR_LIFETIME) - lifetime = MAX_DEF_RTR_LIFETIME; - - if (lifetime <= max_interval) - return VNET_API_ERROR_INVALID_VALUE; - } - - if (min_interval != 0) - { - if ((min_interval > .75 * max_interval) || (min_interval < 3)) - return VNET_API_ERROR_INVALID_VALUE; - } - - if ((initial_count > MAX_INITIAL_RTR_ADVERTISEMENTS) || - (initial_interval > MAX_INITIAL_RTR_ADVERT_INTERVAL)) - return VNET_API_ERROR_INVALID_VALUE; - - /* - if "flag" is set and is_no is true then restore default value else set value corresponding to "flag" - if "flag" is clear don't change corresponding value - */ - radv_info->send_radv = - (suppress != 0) ? ((is_no != 0) ? 1 : 0) : radv_info->send_radv; - radv_info->adv_managed_flag = - (managed != 0) ? ((is_no) ? 0 : 1) : radv_info->adv_managed_flag; - radv_info->adv_other_flag = - (other != 0) ? ((is_no) ? 0 : 1) : radv_info->adv_other_flag; - radv_info->adv_link_layer_address = - (ll_option != - 0) ? ((is_no) ? 1 : 0) : radv_info->adv_link_layer_address; - radv_info->send_unicast = - (send_unicast != 0) ? ((is_no) ? 0 : 1) : radv_info->send_unicast; - radv_info->cease_radv = - (cease != 0) ? ((is_no) ? 0 : 1) : radv_info->cease_radv; - - radv_info->min_radv_interval = min_interval; - radv_info->max_radv_interval = max_interval; - radv_info->adv_router_lifetime_in_sec = lifetime; - - radv_info->initial_adverts_count = - (initial_count != - 0) ? ((is_no) ? MAX_INITIAL_RTR_ADVERTISEMENTS : initial_count) : - radv_info->initial_adverts_count; - radv_info->initial_adverts_interval = - (initial_interval != - 0) ? ((is_no) ? MAX_INITIAL_RTR_ADVERT_INTERVAL : initial_interval) : - radv_info->initial_adverts_interval; - - /* restart */ - if ((cease != 0) && (is_no)) - radv_info->send_radv = 1; - - radv_info->initial_adverts_sent = radv_info->initial_adverts_count - 1; - radv_info->next_multicast_time = vlib_time_now (vm); - radv_info->last_multicast_time = vlib_time_now (vm); - radv_info->last_radv_time = 0; - } - return (error); -} - -int -ip6_neighbor_ra_prefix (vlib_main_t * vm, u32 sw_if_index, - ip6_address_t * prefix_addr, u8 prefix_len, - u8 use_default, u32 val_lifetime, u32 pref_lifetime, - u8 no_advertise, u8 off_link, u8 no_autoconfig, - u8 no_onlink, u8 is_no) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - int error; - - u32 ri; - - /* look up the radv_t information for this interface */ - vec_validate_init_empty (nm->if_radv_pool_index_by_sw_if_index, sw_if_index, - ~0); - - ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - - error = (ri != ~0) ? 0 : VNET_API_ERROR_INVALID_SW_IF_INDEX; - - if (!error) - { - f64 now = vlib_time_now (vm); - ip6_radv_t *radv_info; - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - - /* prefix info add, delete or update */ - ip6_radv_prefix_t *prefix; - - /* lookup prefix info for this address on this interface */ - uword *p = mhash_get (&radv_info->address_to_prefix_index, prefix_addr); - - prefix = p ? pool_elt_at_index (radv_info->adv_prefixes_pool, p[0]) : 0; - - if (is_no) - { - /* delete */ - if (!prefix) - return VNET_API_ERROR_INVALID_VALUE; /* invalid prefix */ - - if (prefix->prefix_len != prefix_len) - return VNET_API_ERROR_INVALID_VALUE_2; - - /* FIXME - Should the DP do this or the CP ? */ - /* do specific delete processing here before returning */ - /* try to remove from routing table */ - - mhash_unset (&radv_info->address_to_prefix_index, prefix_addr, - /* old_value */ 0); - pool_put (radv_info->adv_prefixes_pool, prefix); - - radv_info->initial_adverts_sent = - radv_info->initial_adverts_count - 1; - radv_info->next_multicast_time = vlib_time_now (vm); - radv_info->last_multicast_time = vlib_time_now (vm); - radv_info->last_radv_time = 0; - return (error); - } - - /* adding or changing */ - if (!prefix) - { - /* add */ - u32 pi; - pool_get (radv_info->adv_prefixes_pool, prefix); - memset (prefix, 0, sizeof (*prefix)); - pi = prefix - radv_info->adv_prefixes_pool; - mhash_set (&radv_info->address_to_prefix_index, prefix_addr, pi, - /* old_value */ 0); - - clib_memset (prefix, 0x0, sizeof (ip6_radv_prefix_t)); - - prefix->prefix_len = prefix_len; - clib_memcpy (&prefix->prefix, prefix_addr, sizeof (ip6_address_t)); - - /* initialize default values */ - prefix->adv_on_link_flag = 1; /* L bit set */ - prefix->adv_autonomous_flag = 1; /* A bit set */ - prefix->adv_valid_lifetime_in_secs = DEF_ADV_VALID_LIFETIME; - prefix->adv_pref_lifetime_in_secs = DEF_ADV_PREF_LIFETIME; - prefix->enabled = 1; - prefix->decrement_lifetime_flag = 1; - prefix->deprecated_prefix_flag = 1; - - if (off_link == 0) - { - /* FIXME - Should the DP do this or the CP ? */ - /* insert prefix into routing table as a connected prefix */ - } - - if (use_default) - goto restart; - } - else - { - - if (prefix->prefix_len != prefix_len) - return VNET_API_ERROR_INVALID_VALUE_2; - - if (off_link != 0) - { - /* FIXME - Should the DP do this or the CP ? */ - /* remove from routing table if already there */ - } - } - - if ((val_lifetime == ~0) || (pref_lifetime == ~0)) - { - prefix->adv_valid_lifetime_in_secs = ~0; - prefix->adv_pref_lifetime_in_secs = ~0; - prefix->decrement_lifetime_flag = 0; - } - else - { - prefix->adv_valid_lifetime_in_secs = val_lifetime;; - prefix->adv_pref_lifetime_in_secs = pref_lifetime; - } - - /* copy remaining */ - prefix->enabled = !(no_advertise != 0); - prefix->adv_on_link_flag = !((off_link != 0) || (no_onlink != 0)); - prefix->adv_autonomous_flag = !(no_autoconfig != 0); - - restart: - /* restart */ - /* fill in the expiration times */ - prefix->valid_lifetime_expires = - now + prefix->adv_valid_lifetime_in_secs; - prefix->pref_lifetime_expires = now + prefix->adv_pref_lifetime_in_secs; - - radv_info->initial_adverts_sent = radv_info->initial_adverts_count - 1; - radv_info->next_multicast_time = vlib_time_now (vm); - radv_info->last_multicast_time = vlib_time_now (vm); - radv_info->last_radv_time = 0; - } - return (error); -} - -clib_error_t * -ip6_neighbor_cmd (vlib_main_t * vm, unformat_input_t * main_input, - vlib_cli_command_t * cmd) -{ - vnet_main_t *vnm = vnet_get_main (); - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - clib_error_t *error = 0; - u8 is_no = 0; - u8 suppress = 0, managed = 0, other = 0; - u8 suppress_ll_option = 0, send_unicast = 0, cease = 0; - u8 use_lifetime = 0; - u32 sw_if_index, ra_lifetime = 0, ra_initial_count = - 0, ra_initial_interval = 0; - u32 ra_max_interval = 0, ra_min_interval = 0; - - unformat_input_t _line_input, *line_input = &_line_input; - vnet_sw_interface_t *sw_if0; - - int add_radv_info = 1; - __attribute__ ((unused)) ip6_radv_t *radv_info = 0; - ip6_address_t ip6_addr; - u32 addr_len; - - - /* Get a line of input. */ - if (!unformat_user (main_input, unformat_line_input, line_input)) - return 0; - - /* get basic radv info for this interface */ - if (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) - { - - if (unformat_user (line_input, - unformat_vnet_sw_interface, vnm, &sw_if_index)) - { - u32 ri; - ethernet_interface_t *eth_if0 = 0; - - sw_if0 = vnet_get_sup_sw_interface (vnm, sw_if_index); - if (sw_if0->type == VNET_SW_INTERFACE_TYPE_HARDWARE) - eth_if0 = - ethernet_get_interface (ðernet_main, sw_if0->hw_if_index); - - if (!eth_if0) - { - error = - clib_error_return (0, "Interface must be of ethernet type"); - goto done; - } - - /* look up the radv_t information for this interface */ - vec_validate_init_empty (nm->if_radv_pool_index_by_sw_if_index, - sw_if_index, ~0); - - ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - - if (ri != ~0) - { - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - } - else - { - error = clib_error_return (0, "unknown interface %U'", - format_unformat_error, line_input); - goto done; - } - } - else - { - error = clib_error_return (0, "invalid interface name %U'", - format_unformat_error, line_input); - goto done; - } - } - - /* get the rest of the command */ - while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) - { - if (unformat (line_input, "no")) - is_no = 1; - else if (unformat (line_input, "prefix %U/%d", - unformat_ip6_address, &ip6_addr, &addr_len)) - { - add_radv_info = 0; - break; - } - else if (unformat (line_input, "ra-managed-config-flag")) - { - managed = 1; - break; - } - else if (unformat (line_input, "ra-other-config-flag")) - { - other = 1; - break; - } - else if (unformat (line_input, "ra-suppress") || - unformat (line_input, "ra-surpress")) - { - suppress = 1; - break; - } - else if (unformat (line_input, "ra-suppress-link-layer") || - unformat (line_input, "ra-surpress-link-layer")) - { - suppress_ll_option = 1; - break; - } - else if (unformat (line_input, "ra-send-unicast")) - { - send_unicast = 1; - break; - } - else if (unformat (line_input, "ra-lifetime")) - { - if (!unformat (line_input, "%d", &ra_lifetime)) - { - error = unformat_parse_error (line_input); - goto done; - } - use_lifetime = 1; - break; - } - else if (unformat (line_input, "ra-initial")) - { - if (!unformat - (line_input, "%d %d", &ra_initial_count, &ra_initial_interval)) - { - error = unformat_parse_error (line_input); - goto done; - } - break; - } - else if (unformat (line_input, "ra-interval")) - { - if (!unformat (line_input, "%d", &ra_max_interval)) - { - error = unformat_parse_error (line_input); - goto done; - } - - if (!unformat (line_input, "%d", &ra_min_interval)) - ra_min_interval = 0; - break; - } - else if (unformat (line_input, "ra-cease")) - { - cease = 1; - break; - } - else - { - error = unformat_parse_error (line_input); - goto done; - } - } - - if (add_radv_info) - { - ip6_neighbor_ra_config (vm, sw_if_index, - suppress, managed, other, - suppress_ll_option, send_unicast, cease, - use_lifetime, ra_lifetime, - ra_initial_count, ra_initial_interval, - ra_max_interval, ra_min_interval, is_no); - } - else - { - u32 valid_lifetime_in_secs = 0; - u32 pref_lifetime_in_secs = 0; - u8 use_prefix_default_values = 0; - u8 no_advertise = 0; - u8 off_link = 0; - u8 no_autoconfig = 0; - u8 no_onlink = 0; - - /* get the rest of the command */ - while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) - { - if (unformat (line_input, "default")) - { - use_prefix_default_values = 1; - break; - } - else if (unformat (line_input, "infinite")) - { - valid_lifetime_in_secs = ~0; - pref_lifetime_in_secs = ~0; - break; - } - else if (unformat (line_input, "%d %d", &valid_lifetime_in_secs, - &pref_lifetime_in_secs)) - break; - else - break; - } - - - /* get the rest of the command */ - while (!use_prefix_default_values && - unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) - { - if (unformat (line_input, "no-advertise")) - no_advertise = 1; - else if (unformat (line_input, "off-link")) - off_link = 1; - else if (unformat (line_input, "no-autoconfig")) - no_autoconfig = 1; - else if (unformat (line_input, "no-onlink")) - no_onlink = 1; - else - { - error = unformat_parse_error (line_input); - goto done; - } - } - - ip6_neighbor_ra_prefix (vm, sw_if_index, - &ip6_addr, addr_len, - use_prefix_default_values, - valid_lifetime_in_secs, - pref_lifetime_in_secs, - no_advertise, - off_link, no_autoconfig, no_onlink, is_no); - } - -done: - unformat_free (line_input); - - return error; -} - -static void -ip6_print_addrs (vlib_main_t * vm, u32 * addrs) -{ - ip_lookup_main_t *lm = &ip6_main.lookup_main; - u32 i; - - for (i = 0; i < vec_len (addrs); i++) - { - ip_interface_address_t *a = - pool_elt_at_index (lm->if_address_pool, addrs[i]); - ip6_address_t *address = ip_interface_address_get_address (lm, a); - - vlib_cli_output (vm, "\t\t%U/%d", - format_ip6_address, address, a->address_length); - } -} - -static clib_error_t * -show_ip6_interface_cmd (vlib_main_t * vm, - unformat_input_t * input, vlib_cli_command_t * cmd) -{ - vnet_main_t *vnm = vnet_get_main (); - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - clib_error_t *error = 0; - u32 sw_if_index; - - sw_if_index = ~0; - - if (unformat_user (input, unformat_vnet_sw_interface, vnm, &sw_if_index)) - { - u32 ri; - - /* look up the radv_t information for this interface */ - vec_validate_init_empty (nm->if_radv_pool_index_by_sw_if_index, - sw_if_index, ~0); - - ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - - if (ri != ~0) - { - ip_lookup_main_t *lm = &ip6_main.lookup_main; - ip6_radv_t *radv_info; - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - - vlib_cli_output (vm, "%U is admin %s\n", - format_vnet_sw_interface_name, vnm, - vnet_get_sw_interface (vnm, sw_if_index), - (vnet_sw_interface_is_admin_up (vnm, sw_if_index) ? - "up" : "down")); - - u32 ai; - u32 *link_scope = 0, *global_scope = 0; - u32 *local_scope = 0, *unknown_scope = 0; - ip_interface_address_t *a; - - vec_validate_init_empty (lm->if_address_pool_index_by_sw_if_index, - sw_if_index, ~0); - ai = lm->if_address_pool_index_by_sw_if_index[sw_if_index]; - - while (ai != (u32) ~ 0) - { - a = pool_elt_at_index (lm->if_address_pool, ai); - ip6_address_t *address = - ip_interface_address_get_address (lm, a); - - if (ip6_address_is_link_local_unicast (address)) - vec_add1 (link_scope, ai); - else if (ip6_address_is_global_unicast (address)) - vec_add1 (global_scope, ai); - else if (ip6_address_is_local_unicast (address)) - vec_add1 (local_scope, ai); - else - vec_add1 (unknown_scope, ai); - - ai = a->next_this_sw_interface; - } - - if (vec_len (link_scope)) - { - vlib_cli_output (vm, "\tLink-local address(es):\n"); - ip6_print_addrs (vm, link_scope); - vec_free (link_scope); - } - - if (vec_len (local_scope)) - { - vlib_cli_output (vm, "\tLocal unicast address(es):\n"); - ip6_print_addrs (vm, local_scope); - vec_free (local_scope); - } - - if (vec_len (global_scope)) - { - vlib_cli_output (vm, "\tGlobal unicast address(es):\n"); - ip6_print_addrs (vm, global_scope); - vec_free (global_scope); - } - - if (vec_len (unknown_scope)) - { - vlib_cli_output (vm, "\tOther-scope address(es):\n"); - ip6_print_addrs (vm, unknown_scope); - vec_free (unknown_scope); - } - - vlib_cli_output (vm, "\tLink-local address(es):\n"); - vlib_cli_output (vm, "\t\t%U\n", format_ip6_address, - &radv_info->link_local_address); - - vlib_cli_output (vm, "\tJoined group address(es):\n"); - ip6_mldp_group_t *m; - /* *INDENT-OFF* */ - pool_foreach (m, radv_info->mldp_group_pool, - ({ - vlib_cli_output (vm, "\t\t%U\n", format_ip6_address, - &m->mcast_address); - })); - /* *INDENT-ON* */ - - vlib_cli_output (vm, "\tAdvertised Prefixes:\n"); - ip6_radv_prefix_t *p; - /* *INDENT-OFF* */ - pool_foreach (p, radv_info->adv_prefixes_pool, - ({ - vlib_cli_output (vm, "\t\tprefix %U, length %d\n", - format_ip6_address, &p->prefix, p->prefix_len); - })); - /* *INDENT-ON* */ - - vlib_cli_output (vm, "\tMTU is %d\n", radv_info->adv_link_mtu); - vlib_cli_output (vm, "\tICMP error messages are unlimited\n"); - vlib_cli_output (vm, "\tICMP redirects are disabled\n"); - vlib_cli_output (vm, "\tICMP unreachables are not sent\n"); - vlib_cli_output (vm, "\tND DAD is disabled\n"); - //vlib_cli_output (vm, "\tND reachable time is %d milliseconds\n",); - vlib_cli_output (vm, "\tND advertised reachable time is %d\n", - radv_info->adv_neighbor_reachable_time_in_msec); - vlib_cli_output (vm, - "\tND advertised retransmit interval is %d (msec)\n", - radv_info-> - adv_time_in_msec_between_retransmitted_neighbor_solicitations); - - u32 ra_interval = radv_info->max_radv_interval; - u32 ra_interval_min = radv_info->min_radv_interval; - vlib_cli_output (vm, - "\tND router advertisements are sent every %d seconds (min interval is %d)\n", - ra_interval, ra_interval_min); - vlib_cli_output (vm, - "\tND router advertisements live for %d seconds\n", - radv_info->adv_router_lifetime_in_sec); - vlib_cli_output (vm, - "\tHosts %s stateless autoconfig for addresses\n", - (radv_info->adv_managed_flag) ? "use" : - " don't use"); - vlib_cli_output (vm, "\tND router advertisements sent %d\n", - radv_info->n_advertisements_sent); - vlib_cli_output (vm, "\tND router solicitations received %d\n", - radv_info->n_solicitations_rcvd); - vlib_cli_output (vm, "\tND router solicitations dropped %d\n", - radv_info->n_solicitations_dropped); - } - else - { - error = clib_error_return (0, "IPv6 not enabled on interface", - format_unformat_error, input); - - } - } - return error; -} - -/*? - * This command is used to display various IPv6 attributes on a given - * interface. - * - * @cliexpar - * Example of how to display IPv6 settings: - * @cliexstart{show ip6 interface GigabitEthernet2/0/0} - * GigabitEthernet2/0/0 is admin up - * Link-local address(es): - * fe80::ab8/64 - * Joined group address(es): - * ff02::1 - * ff02::2 - * ff02::16 - * ff02::1:ff00:ab8 - * Advertised Prefixes: - * prefix fe80::fe:28ff:fe9c:75b3, length 64 - * MTU is 1500 - * ICMP error messages are unlimited - * ICMP redirects are disabled - * ICMP unreachables are not sent - * ND DAD is disabled - * ND advertised reachable time is 0 - * ND advertised retransmit interval is 0 (msec) - * ND router advertisements are sent every 200 seconds (min interval is 150) - * ND router advertisements live for 600 seconds - * Hosts use stateless autoconfig for addresses - * ND router advertisements sent 19336 - * ND router solicitations received 0 - * ND router solicitations dropped 0 - * @cliexend - * Example of output if IPv6 is not enabled on the interface: - * @cliexstart{show ip6 interface GigabitEthernet2/0/0} - * show ip6 interface: IPv6 not enabled on interface - * @cliexend -?*/ -/* *INDENT-OFF* */ -VLIB_CLI_COMMAND (show_ip6_interface_command, static) = -{ - .path = "show ip6 interface", - .function = show_ip6_interface_cmd, - .short_help = "show ip6 interface <interface>", -}; -/* *INDENT-ON* */ - -clib_error_t * -disable_ip6_interface (vlib_main_t * vm, u32 sw_if_index) -{ - clib_error_t *error = 0; - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - u32 ri; - - /* look up the radv_t information for this interface */ - vec_validate_init_empty (nm->if_radv_pool_index_by_sw_if_index, sw_if_index, - ~0); - ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - - /* if not created - do nothing */ - if (ri != ~0) - { - ip6_radv_t *radv_info; - - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - - /* check radv_info ref count for other ip6 addresses on this interface */ - /* This implicitly excludes the link local address */ - if (radv_info->ref_count == 0) - { - /* essentially "disables" ipv6 on this interface */ - ip6_ll_prefix_t ilp = { - .ilp_addr = radv_info->link_local_address, - .ilp_sw_if_index = sw_if_index, - }; - ip6_ll_table_entry_delete (&ilp); - ip6_sw_interface_enable_disable (sw_if_index, 0); - ip6_mfib_interface_enable_disable (sw_if_index, 0); - ip6_neighbor_sw_interface_add_del (vnet_get_main (), sw_if_index, - 0); - } - } - return error; -} - -int -ip6_interface_enabled (vlib_main_t * vm, u32 sw_if_index) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - u32 ri = ~0; - - /* look up the radv_t information for this interface */ - vec_validate_init_empty (nm->if_radv_pool_index_by_sw_if_index, sw_if_index, - ~0); - - ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - - return ri != ~0; -} - -clib_error_t * -enable_ip6_interface (vlib_main_t * vm, u32 sw_if_index) -{ - clib_error_t *error = 0; - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - u32 ri; - int is_add = 1; - - /* look up the radv_t information for this interface */ - vec_validate_init_empty (nm->if_radv_pool_index_by_sw_if_index, sw_if_index, - ~0); - - ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - - /* if not created yet */ - if (ri == ~0) - { - vnet_main_t *vnm = vnet_get_main (); - vnet_sw_interface_t *sw_if0; - - sw_if0 = vnet_get_sup_sw_interface (vnm, sw_if_index); - if (sw_if0->type == VNET_SW_INTERFACE_TYPE_HARDWARE) - { - ethernet_interface_t *eth_if0; - - eth_if0 = - ethernet_get_interface (ðernet_main, sw_if0->hw_if_index); - if (eth_if0) - { - /* create radv_info. for this interface. This holds all the info needed for router adverts */ - ri = - ip6_neighbor_sw_interface_add_del (vnm, sw_if_index, is_add); - - if (ri != ~0) - { - ip6_radv_t *radv_info; - ip6_address_t link_local_address; - - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - - ip6_link_local_address_from_ethernet_mac_address - (&link_local_address, eth_if0->address); - - sw_if0 = vnet_get_sw_interface (vnm, sw_if_index); - if (sw_if0->type == VNET_SW_INTERFACE_TYPE_SUB || - sw_if0->type == VNET_SW_INTERFACE_TYPE_PIPE || - sw_if0->type == VNET_SW_INTERFACE_TYPE_P2P) - { - /* make up an interface id */ - link_local_address.as_u64[1] = - random_u64 (&radv_info->randomizer); - - link_local_address.as_u64[0] = - clib_host_to_net_u64 (0xFE80000000000000ULL); - /* clear u bit */ - link_local_address.as_u8[8] &= 0xfd; - } - { - ip6_ll_prefix_t ilp = { - .ilp_addr = link_local_address, - .ilp_sw_if_index = sw_if_index, - }; - - ip6_ll_table_entry_update (&ilp, FIB_ROUTE_PATH_LOCAL); - } - - /* essentially "enables" ipv6 on this interface */ - ip6_mfib_interface_enable_disable (sw_if_index, 1); - ip6_sw_interface_enable_disable (sw_if_index, 1); - - if (!error) - { - radv_info->link_local_address = link_local_address; - } - } - } - } - } - return error; -} - -int -ip6_get_ll_address (u32 sw_if_index, ip6_address_t * addr) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_radv_t *radv_info; - u32 ri; - - if (vec_len (nm->if_radv_pool_index_by_sw_if_index) <= sw_if_index) - return 0; - - ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - - if (ri == ~0) - return 0; - - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - *addr = radv_info->link_local_address; - - return (!0); -} - -static clib_error_t * -enable_ip6_interface_cmd (vlib_main_t * vm, - unformat_input_t * input, vlib_cli_command_t * cmd) -{ - vnet_main_t *vnm = vnet_get_main (); - clib_error_t *error = 0; - u32 sw_if_index; - - sw_if_index = ~0; - - if (unformat_user (input, unformat_vnet_sw_interface, vnm, &sw_if_index)) - { - enable_ip6_interface (vm, sw_if_index); - } - else - { - error = clib_error_return (0, "unknown interface\n'", - format_unformat_error, input); - - } - return error; -} - -/*? - * This command is used to enable IPv6 on a given interface. - * - * @cliexpar - * Example of how enable IPv6 on a given interface: - * @cliexcmd{enable ip6 interface GigabitEthernet2/0/0} -?*/ -/* *INDENT-OFF* */ -VLIB_CLI_COMMAND (enable_ip6_interface_command, static) = -{ - .path = "enable ip6 interface", - .function = enable_ip6_interface_cmd, - .short_help = "enable ip6 interface <interface>", -}; -/* *INDENT-ON* */ - -static clib_error_t * -disable_ip6_interface_cmd (vlib_main_t * vm, - unformat_input_t * input, vlib_cli_command_t * cmd) -{ - vnet_main_t *vnm = vnet_get_main (); - clib_error_t *error = 0; - u32 sw_if_index; - - sw_if_index = ~0; - - if (unformat_user (input, unformat_vnet_sw_interface, vnm, &sw_if_index)) - { - error = disable_ip6_interface (vm, sw_if_index); - } - else - { - error = clib_error_return (0, "unknown interface\n'", - format_unformat_error, input); - - } - return error; -} - -/*? - * This command is used to disable IPv6 on a given interface. - * - * @cliexpar - * Example of how disable IPv6 on a given interface: - * @cliexcmd{disable ip6 interface GigabitEthernet2/0/0} -?*/ -/* *INDENT-OFF* */ -VLIB_CLI_COMMAND (disable_ip6_interface_command, static) = -{ - .path = "disable ip6 interface", - .function = disable_ip6_interface_cmd, - .short_help = "disable ip6 interface <interface>", -}; -/* *INDENT-ON* */ - -/*? - * This command is used to configure the neighbor discovery - * parameters on a given interface. Use the '<em>show ip6 interface</em>' - * command to display some of the current neighbor discovery parameters - * on a given interface. This command has three formats: - * - * - * <b>Format 1 - Router Advertisement Options:</b> (Only one can be entered in a single command) - * - * '<em><b>ip6 nd <interface> [no] [ra-managed-config-flag] | [ra-other-config-flag] | [ra-suppress] | [ra-suppress-link-layer] | [ra-send-unicast] | [ra-lifetime <lifetime>] | [ra-initial <cnt> <interval>] | [ra-interval <max-interval> [<min-interval>]] | [ra-cease]</b></em>' - * - * Where: - * - * <em>[no] ra-managed-config-flag</em> - Advertises in ICMPv6 - * router-advertisement messages to use stateful address - * auto-configuration to obtain address information (sets the M-bit). - * Default is the M-bit is not set and the '<em>no</em>' option - * returns it to this default state. - * - * <em>[no] ra-other-config-flag</em> - Indicates in ICMPv6 - * router-advertisement messages that hosts use stateful auto - * configuration to obtain nonaddress related information (sets - * the O-bit). Default is the O-bit is not set and the '<em>no</em>' - * option returns it to this default state. - * - * <em>[no] ra-suppress</em> - Disables sending ICMPv6 router-advertisement - * messages. The '<em>no</em>' option implies to enable sending ICMPv6 - * router-advertisement messages. - * - * <em>[no] ra-suppress-link-layer</em> - Indicates not to include the - * optional source link-layer address in the ICMPv6 router-advertisement - * messages. Default is to include the optional source link-layer address - * and the '<em>no</em>' option returns it to this default state. - * - * <em>[no] ra-send-unicast</em> - Use the source address of the - * router-solicitation message if availiable. The default is to use - * multicast address of all nodes, and the '<em>no</em>' option returns - * it to this default state. - * - * <em>[no] ra-lifetime <lifetime></em> - Advertises the lifetime of a - * default router in ICMPv6 router-advertisement messages. The range is - * from 0 to 9000 seconds. '<em><lifetime></em>' must be greater than - * '<em><max-interval></em>'. The default value is 600 seconds and the - * '<em>no</em>' option returns it to this default value. - * - * <em>[no] ra-initial <cnt> <interval></em> - Number of initial ICMPv6 - * router-advertisement messages sent and the interval between each - * message. Range for count is 1 - 3 and default is 3. Range for interval - * is 1 to 16 seconds, and default is 16 seconds. The '<em>no</em>' option - * returns both to their default value. - * - * <em>[no] ra-interval <max-interval> [<min-interval>]</em> - Configures the - * interval between sending ICMPv6 router-advertisement messages. The - * range for max-interval is from 4 to 200 seconds. min-interval can not - * be more than 75% of max-interval. If not set, min-interval will be - * set to 75% of max-interval. The range for min-interval is from 3 to - * 150 seconds. The '<em>no</em>' option returns both to their default - * value. - * - * <em>[no] ra-cease</em> - Cease sending ICMPv6 router-advertisement messages. - * The '<em>no</em>' options implies to start (or restart) sending - * ICMPv6 router-advertisement messages. - * - * - * <b>Format 2 - Prefix Options:</b> - * - * '<em><b>ip6 nd <interface> [no] prefix <ip6-address>/<width> [<valid-lifetime> <pref-lifetime> | infinite] [no-advertise] [off-link] [no-autoconfig] [no-onlink]</b></em>' - * - * Where: - * - * <em>no</em> - All additional flags are ignored and the prefix is deleted. - * - * <em><valid-lifetime> <pref-lifetime></em> - '<em><valid-lifetime></em>' is the - * length of time in seconds during what the prefix is valid for the purpose of - * on-link determination. Range is 7203 to 2592000 seconds and default is 2592000 - * seconds (30 days). '<em><pref-lifetime></em>' is the prefered-lifetime and is the - * length of time in seconds during what addresses generated from the prefix remain - * preferred. Range is 0 to 604800 seconds and default is 604800 seconds (7 days). - * - * <em>infinite</em> - Both '<em><valid-lifetime></em>' and '<em><<pref-lifetime></em>' - * are inifinte, no timeout. - * - * <em>no-advertise</em> - Do not send full router address in prefix - * advertisement. Default is to advertise (i.e. - This flag is off by default). - * - * <em>off-link</em> - Prefix is off-link, clear L-bit in packet. Default is on-link - * (i.e. - This flag is off and L-bit in packet is set by default and this prefix can - * be used for on-link determination). '<em>no-onlink</em>' also controls the L-bit. - * - * <em>no-autoconfig</em> - Do not use prefix for autoconfiguration, clear A-bit in packet. - * Default is autoconfig (i.e. - This flag is off and A-bit in packet is set by default. - * - * <em>no-onlink</em> - Do not use prefix for onlink determination, clear L-bit in packet. - * Default is on-link (i.e. - This flag is off and L-bit in packet is set by default and - * this prefix can be used for on-link determination). '<em>off-link</em>' also controls - * the L-bit. - * - * - * <b>Format 3: - Default of Prefix:</b> - * - * '<em><b>ip6 nd <interface> [no] prefix <ip6-address>/<width> default</b></em>' - * - * When a new prefix is added (or existing one is being overwritten) <em>default</em> - * uses default values for the prefix. If <em>no</em> is used, the <em>default</em> - * is ignored and the prefix is deleted. - * - * - * @cliexpar - * Example of how set a router advertisement option: - * @cliexcmd{ip6 nd GigabitEthernet2/0/0 ra-interval 100 20} - * Example of how to add a prefix: - * @cliexcmd{ip6 nd GigabitEthernet2/0/0 prefix fe80::fe:28ff:fe9c:75b3/64 infinite no-advertise} - * Example of how to delete a prefix: - * @cliexcmd{ip6 nd GigabitEthernet2/0/0 no prefix fe80::fe:28ff:fe9c:75b3/64} -?*/ -/* *INDENT-OFF* */ -VLIB_CLI_COMMAND (ip6_nd_command, static) = -{ - .path = "ip6 nd", - .short_help = "ip6 nd <interface> ...", - .function = ip6_neighbor_cmd, -}; -/* *INDENT-ON* */ - -clib_error_t * -ip6_neighbor_set_link_local_address (vlib_main_t * vm, u32 sw_if_index, - ip6_address_t * address) -{ - clib_error_t *error = 0; - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - u32 ri; - ip6_radv_t *radv_info; - vnet_main_t *vnm = vnet_get_main (); - ip6_ll_prefix_t pfx = { 0, }; - - if (!ip6_address_is_link_local_unicast (address)) - return (error = clib_error_return (0, "address not link-local", - format_unformat_error)); - - /* call enable ipv6 */ - enable_ip6_interface (vm, sw_if_index); - - ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - - if (ri != ~0) - { - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - - pfx.ilp_sw_if_index = sw_if_index; - - pfx.ilp_addr = radv_info->link_local_address; - ip6_ll_table_entry_delete (&pfx); - - pfx.ilp_addr = *address; - ip6_ll_table_entry_update (&pfx, FIB_ROUTE_PATH_LOCAL); - - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - radv_info->link_local_address = *address; - } - else - { - vnm->api_errno = VNET_API_ERROR_IP6_NOT_ENABLED; - error = clib_error_return (0, "ip6 not enabled for interface", - format_unformat_error); - } - return error; -} - -/** - * @brief callback when an interface address is added or deleted - */ -static void -ip6_neighbor_add_del_interface_address (ip6_main_t * im, - uword opaque, - u32 sw_if_index, - ip6_address_t * address, - u32 address_length, - u32 if_address_index, u32 is_delete) -{ - vnet_main_t *vnm = vnet_get_main (); - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - u32 ri; - vlib_main_t *vm = vnm->vlib_main; - ip6_radv_t *radv_info; - ip6_address_t a; - - /* create solicited node multicast address for this interface address */ - ip6_set_solicited_node_multicast_address (&a, 0); - - a.as_u8[0xd] = address->as_u8[0xd]; - a.as_u8[0xe] = address->as_u8[0xe]; - a.as_u8[0xf] = address->as_u8[0xf]; - - if (!is_delete) - { - /* try to create radv_info - does nothing if ipv6 already enabled */ - enable_ip6_interface (vm, sw_if_index); - - /* look up the radv_t information for this interface */ - vec_validate_init_empty (nm->if_radv_pool_index_by_sw_if_index, - sw_if_index, ~0); - ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - if (ri != ~0) - { - /* get radv_info */ - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - - /* add address */ - if (!ip6_address_is_link_local_unicast (address)) - radv_info->ref_count++; - - ip6_neighbor_add_mld_prefix (radv_info, &a); - } - } - else - { - - /* delete */ - /* look up the radv_t information for this interface */ - vec_validate_init_empty (nm->if_radv_pool_index_by_sw_if_index, - sw_if_index, ~0); - ri = nm->if_radv_pool_index_by_sw_if_index[sw_if_index]; - - if (ri != ~0) - { - /* get radv_info */ - radv_info = pool_elt_at_index (nm->if_radv_pool, ri); - - ip6_neighbor_del_mld_prefix (radv_info, &a); - - /* if interface up send MLDP "report" */ - radv_info->all_routers_mcast = 0; - - /* add address */ - if (!ip6_address_is_link_local_unicast (address)) - radv_info->ref_count--; - } - /* Ensure that IPv6 is disabled, and LL removed after ref_count reaches 0 */ - disable_ip6_interface (vm, sw_if_index); - } -} - -clib_error_t * -ip6_set_neighbor_limit (u32 neighbor_limit) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - - nm->limit_neighbor_cache_size = neighbor_limit; - return 0; -} - -static void -ip6_neighbor_table_bind (ip6_main_t * im, - uword opaque, - u32 sw_if_index, - u32 new_fib_index, u32 old_fib_index) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_neighbor_t *n = NULL; - u32 i, *to_re_add = 0; - - /* *INDENT-OFF* */ - pool_foreach (n, nm->neighbor_pool, - ({ - if (n->key.sw_if_index == sw_if_index) - vec_add1 (to_re_add, n - nm->neighbor_pool); - })); - /* *INDENT-ON* */ - - for (i = 0; i < vec_len (to_re_add); i++) - { - n = pool_elt_at_index (nm->neighbor_pool, to_re_add[i]); - ip6_neighbor_adj_fib_remove (n, old_fib_index); - ip6_neighbor_adj_fib_add (n, new_fib_index); - } - vec_free (to_re_add); -} - -static clib_error_t * -ip6_neighbor_init (vlib_main_t * vm) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_main_t *im = &ip6_main; - - mhash_init (&nm->neighbor_index_by_key, - /* value size */ sizeof (uword), - /* key size */ sizeof (ip6_neighbor_key_t)); - - icmp6_register_type (vm, ICMP6_neighbor_solicitation, - ip6_icmp_neighbor_solicitation_node.index); - icmp6_register_type (vm, ICMP6_neighbor_advertisement, - ip6_icmp_neighbor_advertisement_node.index); - icmp6_register_type (vm, ICMP6_router_solicitation, - ip6_icmp_router_solicitation_node.index); - icmp6_register_type (vm, ICMP6_router_advertisement, - ip6_icmp_router_advertisement_node.index); - - /* handler node for ip6 neighbor discovery events and timers */ - vlib_register_node (vm, &ip6_icmp_neighbor_discovery_event_node); - - /* add call backs */ - ip6_add_del_interface_address_callback_t cb; - clib_memset (&cb, 0x0, sizeof (ip6_add_del_interface_address_callback_t)); - - /* when an interface address changes... */ - cb.function = ip6_neighbor_add_del_interface_address; - cb.function_opaque = 0; - vec_add1 (im->add_del_interface_address_callbacks, cb); - - ip6_table_bind_callback_t cbt; - cbt.function = ip6_neighbor_table_bind; - cbt.function_opaque = 0; - vec_add1 (im->table_bind_callbacks, cbt); - - mhash_init (&nm->pending_resolutions_by_address, - /* value size */ sizeof (uword), - /* key size */ sizeof (ip6_address_t)); - - mhash_init (&nm->mac_changes_by_address, - /* value size */ sizeof (uword), - /* key size */ sizeof (ip6_address_t)); - - /* default, configurable */ - nm->limit_neighbor_cache_size = 50000; - - nm->wc_ip6_nd_publisher_node = (uword) ~ 0; - - nm->ip6_ra_publisher_node = (uword) ~ 0; - -#if 0 - /* $$$$ Hack fix for today */ - vec_validate_init_empty - (im->discover_neighbor_next_index_by_hw_if_index, 32, 0 /* drop */ ); -#endif - - return 0; -} - -VLIB_INIT_FUNCTION (ip6_neighbor_init); - - -void -vnet_register_ip6_neighbor_resolution_event (vnet_main_t * vnm, - void *address_arg, - uword node_index, - uword type_opaque, uword data) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_address_t *address = address_arg; - uword *p; - pending_resolution_t *pr; - - pool_get (nm->pending_resolutions, pr); - memset (pr, 0, sizeof (*pr)); - - pr->next_index = ~0; - pr->node_index = node_index; - pr->type_opaque = type_opaque; - pr->data = data; - - p = mhash_get (&nm->pending_resolutions_by_address, address); - if (p) - { - /* Insert new resolution at the head of the list */ - pr->next_index = p[0]; - mhash_unset (&nm->pending_resolutions_by_address, address, 0); - } - - mhash_set (&nm->pending_resolutions_by_address, address, - pr - nm->pending_resolutions, 0 /* old value */ ); -} - -int -vnet_add_del_ip6_nd_change_event (vnet_main_t * vnm, - ip6_nd_change_event_cb_t data_callback, - u32 pid, - void *address_arg, - uword node_index, - uword type_opaque, uword data, int is_add) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_address_t *address = address_arg; - - /* Try to find an existing entry */ - u32 *first = (u32 *) mhash_get (&nm->mac_changes_by_address, address); - u32 *p = first; - pending_resolution_t *mc; - while (p && *p != ~0) - { - mc = pool_elt_at_index (nm->mac_changes, *p); - if (mc->node_index == node_index && mc->type_opaque == type_opaque - && mc->pid == pid) - break; - p = &mc->next_index; - } - - int found = p && *p != ~0; - if (is_add) - { - if (found) - return VNET_API_ERROR_ENTRY_ALREADY_EXISTS; - - pool_get (nm->mac_changes, mc); - memset (mc, 0, sizeof (*mc)); - /* *INDENT-OFF* */ - *mc = (pending_resolution_t) - { - .next_index = ~0, - .node_index = node_index, - .type_opaque = type_opaque, - .data = data, - .data_callback = data_callback, - .pid = pid, - }; - /* *INDENT-ON* */ - - /* Insert new resolution at the end of the list */ - u32 new_idx = mc - nm->mac_changes; - if (p) - p[0] = new_idx; - else - mhash_set (&nm->mac_changes_by_address, address, new_idx, 0); - } - else - { - if (!found) - return VNET_API_ERROR_NO_SUCH_ENTRY; - - /* Clients may need to clean up pool entries, too */ - if (data_callback) - (data_callback) (mc->data, NULL /* no new mac addrs */ , 0, NULL); - - /* Remove the entry from the list and delete the entry */ - *p = mc->next_index; - pool_put (nm->mac_changes, mc); - - /* Remove from hash if we deleted the last entry */ - if (*p == ~0 && p == first) - mhash_unset (&nm->mac_changes_by_address, address, 0); - } - return 0; -} - -int -vnet_ip6_nd_term (vlib_main_t * vm, - vlib_node_runtime_t * node, - vlib_buffer_t * p0, - ethernet_header_t * eth, - ip6_header_t * ip, u32 sw_if_index, u16 bd_index) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - icmp6_neighbor_solicitation_or_advertisement_header_t *ndh; - mac_address_t mac; - - mac_address_from_bytes (&mac, eth->src_address); - ndh = ip6_next_header (ip); - if (ndh->icmp.type != ICMP6_neighbor_solicitation && - ndh->icmp.type != ICMP6_neighbor_advertisement) - return 0; - - if (PREDICT_FALSE ((node->flags & VLIB_NODE_FLAG_TRACE) && - (p0->flags & VLIB_BUFFER_IS_TRACED))) - { - u8 *t0 = vlib_add_trace (vm, node, p0, - sizeof (icmp6_input_trace_t)); - clib_memcpy (t0, ip, sizeof (icmp6_input_trace_t)); - } - - /* Check if anyone want ND events for L2 BDs */ - if (PREDICT_FALSE - (nm->wc_ip6_nd_publisher_node != (uword) ~ 0 - && !ip6_address_is_link_local_unicast (&ip->src_address))) - { - vnet_nd_wc_publish (sw_if_index, &mac, &ip->src_address); - } - - /* Check if MAC entry exsist for solicited target IP */ - if (ndh->icmp.type == ICMP6_neighbor_solicitation) - { - icmp6_neighbor_discovery_ethernet_link_layer_address_option_t *opt; - l2_bridge_domain_t *bd_config; - u8 *macp; - - opt = (void *) (ndh + 1); - if ((opt->header.type != - ICMP6_NEIGHBOR_DISCOVERY_OPTION_source_link_layer_address) || - (opt->header.n_data_u64s != 1)) - return 0; /* source link layer address option not present */ - - bd_config = vec_elt_at_index (l2input_main.bd_configs, bd_index); - macp = - (u8 *) hash_get_mem (bd_config->mac_by_ip6, &ndh->target_address); - if (macp) - { /* found ip-mac entry, generate eighbor advertisement response */ - int bogus_length; - vlib_node_runtime_t *error_node = - vlib_node_get_runtime (vm, ip6_icmp_input_node.index); - ip->dst_address = ip->src_address; - ip->src_address = ndh->target_address; - ip->hop_limit = 255; - opt->header.type = - ICMP6_NEIGHBOR_DISCOVERY_OPTION_target_link_layer_address; - clib_memcpy (opt->ethernet_address, macp, 6); - ndh->icmp.type = ICMP6_neighbor_advertisement; - ndh->advertisement_flags = clib_host_to_net_u32 - (ICMP6_NEIGHBOR_ADVERTISEMENT_FLAG_SOLICITED | - ICMP6_NEIGHBOR_ADVERTISEMENT_FLAG_OVERRIDE); - ndh->icmp.checksum = 0; - ndh->icmp.checksum = - ip6_tcp_udp_icmp_compute_checksum (vm, p0, ip, &bogus_length); - clib_memcpy (eth->dst_address, eth->src_address, 6); - clib_memcpy (eth->src_address, macp, 6); - vlib_error_count (vm, error_node->node_index, - ICMP6_ERROR_NEIGHBOR_ADVERTISEMENTS_TX, 1); - return 1; - } - } - - return 0; - -} - -int -ip6_neighbor_proxy_add_del (u32 sw_if_index, ip6_address_t * addr, u8 is_del) -{ - u32 fib_index; - - fib_prefix_t pfx = { - .fp_len = 128, - .fp_proto = FIB_PROTOCOL_IP6, - .fp_addr = { - .ip6 = *addr, - }, - }; - ip46_address_t nh = { - .ip6 = *addr, - }; - - fib_index = ip6_fib_table_get_index_for_sw_if_index (sw_if_index); - - if (~0 == fib_index) - return VNET_API_ERROR_NO_SUCH_FIB; - - if (is_del) - { - fib_table_entry_path_remove (fib_index, - &pfx, - FIB_SOURCE_IP6_ND_PROXY, - DPO_PROTO_IP6, - &nh, - sw_if_index, - ~0, 1, FIB_ROUTE_PATH_FLAG_NONE); - /* flush the ND cache of this address if it's there */ - vnet_unset_ip6_ethernet_neighbor (vlib_get_main (), sw_if_index, addr); - } - else - { - fib_table_entry_path_add (fib_index, - &pfx, - FIB_SOURCE_IP6_ND_PROXY, - FIB_ENTRY_FLAG_NONE, - DPO_PROTO_IP6, - &nh, - sw_if_index, - ~0, 1, NULL, FIB_ROUTE_PATH_FLAG_NONE); - } - return (0); -} - -static clib_error_t * -set_ip6_nd_proxy_cmd (vlib_main_t * vm, - unformat_input_t * input, vlib_cli_command_t * cmd) -{ - vnet_main_t *vnm = vnet_get_main (); - clib_error_t *error = 0; - ip6_address_t addr; - u32 sw_if_index; - u8 is_del = 0; - - if (unformat_user (input, unformat_vnet_sw_interface, vnm, &sw_if_index)) - { - /* get the rest of the command */ - while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT) - { - if (unformat (input, "%U", unformat_ip6_address, &addr)) - break; - else if (unformat (input, "delete") || unformat (input, "del")) - is_del = 1; - else - return (unformat_parse_error (input)); - } - } - - ip6_neighbor_proxy_add_del (sw_if_index, &addr, is_del); - - return error; -} - -/* *INDENT-OFF* */ -VLIB_CLI_COMMAND (set_ip6_nd_proxy_command, static) = -{ - .path = "set ip6 nd proxy", - .short_help = "set ip6 nd proxy <HOST> <INTERFACE>", - .function = set_ip6_nd_proxy_cmd, -}; -/* *INDENT-ON* */ - -void -ethernet_ndp_change_mac (u32 sw_if_index) -{ - ip6_neighbor_main_t *nm = &ip6_neighbor_main; - ip6_neighbor_t *n; - adj_index_t ai; - - /* *INDENT-OFF* */ - pool_foreach (n, nm->neighbor_pool, - ({ - if (n->key.sw_if_index == sw_if_index) - { - adj_nbr_walk_nh6 (sw_if_index, - &n->key.ip6_address, - ip6_nd_mk_complete_walk, n); - } - })); - /* *INDENT-ON* */ - - ai = adj_glean_get (FIB_PROTOCOL_IP6, sw_if_index); - - if (ADJ_INDEX_INVALID != ai) - adj_glean_update_rewrite (ai); -} - -void -send_ip6_na (vlib_main_t * vm, u32 sw_if_index) -{ - ip6_main_t *i6m = &ip6_main; - ip6_address_t *ip6_addr = ip6_interface_first_address (i6m, sw_if_index); - - send_ip6_na_w_addr (vm, ip6_addr, sw_if_index); -} - -void -send_ip6_na_w_addr (vlib_main_t * vm, - const ip6_address_t * ip6_addr, u32 sw_if_index) -{ - ip6_main_t *i6m = &ip6_main; - vnet_main_t *vnm = vnet_get_main (); - u8 *rewrite, rewrite_len; - vnet_hw_interface_t *hi = vnet_get_sup_hw_interface (vnm, sw_if_index); - u8 dst_address[6]; - - if (ip6_addr) - { - clib_warning - ("Sending unsolicitated NA IP6 address %U on sw_if_idex %d", - format_ip6_address, ip6_addr, sw_if_index); - - /* Form unsolicited neighbor advertisement packet from NS pkt template */ - int bogus_length; - u32 bi = 0; - icmp6_neighbor_solicitation_header_t *h = - vlib_packet_template_get_packet (vm, - &i6m->discover_neighbor_packet_template, - &bi); - if (!h) - return; - - ip6_set_reserved_multicast_address (&h->ip.dst_address, - IP6_MULTICAST_SCOPE_link_local, - IP6_MULTICAST_GROUP_ID_all_hosts); - h->ip.src_address = ip6_addr[0]; - h->neighbor.icmp.type = ICMP6_neighbor_advertisement; - h->neighbor.target_address = ip6_addr[0]; - h->neighbor.advertisement_flags = clib_host_to_net_u32 - (ICMP6_NEIGHBOR_ADVERTISEMENT_FLAG_OVERRIDE); - h->link_layer_option.header.type = - ICMP6_NEIGHBOR_DISCOVERY_OPTION_target_link_layer_address; - clib_memcpy (h->link_layer_option.ethernet_address, - hi->hw_address, vec_len (hi->hw_address)); - h->neighbor.icmp.checksum = - ip6_tcp_udp_icmp_compute_checksum (vm, 0, &h->ip, &bogus_length); - ASSERT (bogus_length == 0); - - /* Setup MAC header with IP6 Etype and mcast DMAC */ - vlib_buffer_t *b = vlib_get_buffer (vm, bi); - ip6_multicast_ethernet_address (dst_address, - IP6_MULTICAST_GROUP_ID_all_hosts); - rewrite = - ethernet_build_rewrite (vnm, sw_if_index, VNET_LINK_IP6, dst_address); - rewrite_len = vec_len (rewrite); - vlib_buffer_advance (b, -rewrite_len); - ethernet_header_t *e = vlib_buffer_get_current (b); - clib_memcpy (e->dst_address, rewrite, rewrite_len); - vec_free (rewrite); - - /* Send unsolicited ND advertisement packet out the specified interface */ - vnet_buffer (b)->sw_if_index[VLIB_RX] = - vnet_buffer (b)->sw_if_index[VLIB_TX] = sw_if_index; - 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); - } -} - -/* - * fd.io coding-style-patch-verification: ON - * - * Local Variables: - * eval: (c-set-style "gnu") - * End: - */ diff --git a/src/vnet/ip/ip6_neighbor.h b/src/vnet/ip/ip6_neighbor.h deleted file mode 100644 index 0707a2a8d21..00000000000 --- a/src/vnet/ip/ip6_neighbor.h +++ /dev/null @@ -1,205 +0,0 @@ -/* - * - * ip6_neighboor.h: ip6 neighbor structures - * - * Copyright (c) 2016 Cisco and/or its affiliates. - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at: - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -#ifndef included_ip6_neighbor_h -#define included_ip6_neighbor_h - -#include <vnet/fib/fib_types.h> -#include <vnet/ethernet/mac_address.h> -#include <vnet/ip/ip_neighbor.h> - -typedef struct -{ - ip6_address_t ip6_address; - u32 sw_if_index; - u32 pad; -} ip6_neighbor_key_t; - -typedef struct -{ - ip6_neighbor_key_t key; - mac_address_t mac; - ip_neighbor_flags_t flags; - f64 time_last_updated; - fib_node_index_t fib_entry_index; -} ip6_neighbor_t; - -extern ip6_address_t ip6_neighbor_get_link_local_address (u32 sw_if_index); - -extern clib_error_t *ip6_neighbor_set_link_local_address (vlib_main_t * vm, - u32 sw_if_index, - ip6_address_t * - address); - -extern ip6_neighbor_t *ip6_neighbors_pool (void); - -extern ip6_neighbor_t *ip6_neighbors_entries (u32 sw_if_index); - -extern int ip6_neighbor_ra_config (vlib_main_t * vm, u32 sw_if_index, - u8 suppress, u8 managed, u8 other, - u8 ll_option, u8 send_unicast, u8 cease, - u8 use_lifetime, u32 lifetime, - u32 initial_count, u32 initial_interval, - u32 max_interval, u32 min_interval, - u8 is_no); - -extern int ip6_neighbor_ra_prefix (vlib_main_t * vm, u32 sw_if_index, - ip6_address_t * prefix_addr, u8 prefix_len, - u8 use_default, u32 val_lifetime, - u32 pref_lifetime, u8 no_advertise, - u8 off_link, u8 no_autoconfig, - u8 no_onlink, u8 is_no); - -extern clib_error_t *ip6_set_neighbor_limit (u32 neighbor_limit); - -extern void vnet_register_ip6_neighbor_resolution_event (vnet_main_t * vnm, - void *address_arg, - uword node_index, - uword type_opaque, - uword data); - -extern int vnet_set_ip6_ethernet_neighbor (vlib_main_t * vm, - u32 sw_if_index, - const ip6_address_t * a, - const mac_address_t * mac, - ip_neighbor_flags_t flags); - -extern int vnet_unset_ip6_ethernet_neighbor (vlib_main_t * vm, - u32 sw_if_index, - const ip6_address_t * a); - -extern int ip6_neighbor_proxy_add_del (u32 sw_if_index, - ip6_address_t * addr, u8 is_add); - -u32 ip6_neighbor_sw_interface_add_del (vnet_main_t * vnm, u32 sw_if_index, - u32 is_add); -typedef struct -{ - u32 sw_if_index; - ip6_address_t ip6; - mac_address_t mac; -} wc_nd_report_t; - -void wc_nd_set_publisher_node (uword node_index, uword event_type); - -typedef struct -{ - u32 irt; - u32 mrt; - u32 mrc; - u32 mrd; -} icmp6_send_router_solicitation_params_t; - -void icmp6_send_router_solicitation (vlib_main_t * vm, u32 sw_if_index, - u8 stop, - icmp6_send_router_solicitation_params_t * - params); - -typedef struct -{ - fib_prefix_t prefix; - u8 flags; - u32 valid_time; - u32 preferred_time; -} ra_report_prefix_info_t; - -typedef struct -{ - u32 sw_if_index; - ip6_address_t router_address; - u8 current_hop_limit; - u8 flags; - u16 router_lifetime_in_sec; - u32 neighbor_reachable_time_in_msec; - u32 time_in_msec_between_retransmitted_neighbor_solicitations; - u8 slla[6]; - u32 mtu; - ra_report_prefix_info_t *prefixes; -} ra_report_t; - -void ra_set_publisher_node (uword node_index, uword event_type); - -typedef struct _vnet_ip6_neighbor_function_list_elt -{ - struct _vnet_ip6_neighbor_function_list_elt *next_ip6_neighbor_function; - clib_error_t *(*fp) (void *data); -} _vnet_ip6_neighbor_function_list_elt_t; - -typedef struct -{ - _vnet_ip6_neighbor_function_list_elt_t *ra_report_functions; -} ip6_neighbor_public_main_t; - -extern ip6_neighbor_public_main_t ip6_neighbor_public_main; - -#define _VNET_IP6_NEIGHBOR_FUNCTION_DECL(f,tag) \ - \ -static void __vnet_ip6_neighbor_function_init_##tag##_##f (void) \ - __attribute__((__constructor__)) ; \ - \ -static void __vnet_ip6_neighbor_function_init_##tag##_##f (void) \ -{ \ - ip6_neighbor_public_main_t * nm = &ip6_neighbor_public_main; \ - static _vnet_ip6_neighbor_function_list_elt_t init_function; \ - init_function.next_ip6_neighbor_function = nm->tag##_functions; \ - nm->tag##_functions = &init_function; \ - init_function.fp = (void *) &f; \ -} \ - \ -static void __vnet_ip6_neighbor_function_deinit_##tag##_##f (void) \ - __attribute__((__destructor__)) ; \ - \ -static void __vnet_ip6_neighbor_function_deinit_##tag##_##f (void) \ -{ \ - ip6_neighbor_public_main_t * nm = &ip6_neighbor_public_main; \ - _vnet_ip6_neighbor_function_list_elt_t *next; \ - if (nm->tag##_functions->fp == (void *) &f) \ - { \ - nm->tag##_functions = \ - nm->tag##_functions->next_ip6_neighbor_function; \ - return; \ - } \ - next = nm->tag##_functions; \ - while (next->next_ip6_neighbor_function) \ - { \ - if (next->next_ip6_neighbor_function->fp == (void *) &f) \ - { \ - next->next_ip6_neighbor_function = \ - next->next_ip6_neighbor_function->next_ip6_neighbor_function; \ - return; \ - } \ - next = next->next_ip6_neighbor_function; \ - } \ -} - -#define VNET_IP6_NEIGHBOR_RA_FUNCTION(f) \ - _VNET_IP6_NEIGHBOR_FUNCTION_DECL(f,ra_report) - -clib_error_t *call_ip6_neighbor_callbacks (void *data, - _vnet_ip6_neighbor_function_list_elt_t - * elt); - -#endif /* included_ip6_neighbor_h */ - -/* - * fd.io coding-style-patch-verification: ON - * - * Local Variables: - * eval: (c-set-style "gnu") - * End: - */ diff --git a/src/vnet/ip/ip6_packet.h b/src/vnet/ip/ip6_packet.h index 8c0698440e3..2ba55b75e09 100644 --- a/src/vnet/ip/ip6_packet.h +++ b/src/vnet/ip/ip6_packet.h @@ -67,75 +67,6 @@ typedef CLIB_PACKED (struct { }) ip6_address_fib_t; /* *INDENT-ON* */ -typedef enum -{ - IP46_TYPE_ANY, - IP46_TYPE_IP4, - IP46_TYPE_IP6 -} ip46_type_t; - -/* *INDENT-OFF* */ -typedef CLIB_PACKED (union ip46_address_t_ { - struct { - u32 pad[3]; - ip4_address_t ip4; - }; - ip6_address_t ip6; - u8 as_u8[16]; - u64 as_u64[2]; -}) ip46_address_t; -/* *INDENT-ON* */ -#define ip46_address_is_ip4(ip46) (((ip46)->pad[0] | (ip46)->pad[1] | (ip46)->pad[2]) == 0) -#define ip46_address_mask_ip4(ip46) ((ip46)->pad[0] = (ip46)->pad[1] = (ip46)->pad[2] = 0) -#define ip46_address_set_ip4(ip46, ip) (ip46_address_mask_ip4(ip46), (ip46)->ip4 = (ip)[0]) -#define ip46_address_reset(ip46) ((ip46)->as_u64[0] = (ip46)->as_u64[1] = 0) -#define ip46_address_cmp(ip46_1, ip46_2) (memcmp(ip46_1, ip46_2, sizeof(*ip46_1))) -#define ip46_address_is_zero(ip46) (((ip46)->as_u64[0] == 0) && ((ip46)->as_u64[1] == 0)) -#define ip46_address_is_equal(a1, a2) (((a1)->as_u64[0] == (a2)->as_u64[0]) \ - && ((a1)->as_u64[1] == (a2)->as_u64[1])) -#define ip46_address_initializer {{{ 0 }}} - -static_always_inline int -ip46_address_is_equal_v4 (const ip46_address_t * ip46, - const ip4_address_t * ip4) -{ - return (ip46->ip4.as_u32 == ip4->as_u32); -} - -static_always_inline int -ip46_address_is_equal_v6 (const ip46_address_t * ip46, - const ip6_address_t * ip6) -{ - return ((ip46->ip6.as_u64[0] == ip6->as_u64[0]) && - (ip46->ip6.as_u64[1] == ip6->as_u64[1])); -} - -static_always_inline void -ip46_address_copy (ip46_address_t * dst, const ip46_address_t * src) -{ - dst->as_u64[0] = src->as_u64[0]; - dst->as_u64[1] = src->as_u64[1]; -} - -static_always_inline void -ip46_address_set_ip6 (ip46_address_t * dst, const ip6_address_t * src) -{ - dst->as_u64[0] = src->as_u64[0]; - dst->as_u64[1] = src->as_u64[1]; -} - -always_inline ip46_address_t -to_ip46 (u32 is_ipv6, u8 * buf) -{ - ip46_address_t ip; - if (is_ipv6) - ip.ip6 = *((ip6_address_t *) buf); - else - ip46_address_set_ip4 (&ip, (ip4_address_t *) buf); - return ip; -} - - always_inline void ip6_addr_fib_init (ip6_address_fib_t * addr_fib, const ip6_address_t * address, u32 fib_index) @@ -189,13 +120,6 @@ ip6_address_is_multicast (const ip6_address_t * a) return a->as_u8[0] == 0xff; } -always_inline uword -ip46_address_is_multicast (const ip46_address_t * a) -{ - return ip46_address_is_ip4 (a) ? ip4_address_is_multicast (&a->ip4) : - ip6_address_is_multicast (&a->ip6); -} - always_inline void ip6_address_copy (ip6_address_t * dst, const ip6_address_t * src) { diff --git a/src/vnet/ip/ip_api.c b/src/vnet/ip/ip_api.c index 6da6ac34d48..4f5b6b9d0a6 100644 --- a/src/vnet/ip/ip_api.c +++ b/src/vnet/ip/ip_api.c @@ -25,9 +25,7 @@ #include <vnet/ethernet/ethernet.h> #include <vnet/ethernet/ethernet_types_api.h> #include <vnet/ip/ip.h> -#include <vnet/ip/ip_neighbor.h> #include <vnet/ip/ip_types_api.h> -#include <vnet/ip/ip6_neighbor.h> #include <vnet/ip/ip_punt_drop.h> #include <vnet/ip/ip_types_api.h> #include <vnet/fib/fib_table.h> @@ -43,12 +41,11 @@ #include <vnet/fib/ip6_fib.h> #include <vnet/fib/fib_path_list.h> #include <vnet/ip/ip6_hop_by_hop.h> +#include <vnet/ip/ip6_link.h> #include <vnet/ip/reass/ip4_sv_reass.h> #include <vnet/ip/reass/ip4_full_reass.h> #include <vnet/ip/reass/ip6_sv_reass.h> #include <vnet/ip/reass/ip6_full_reass.h> -#include <vnet/ethernet/arp.h> -#include <vnet/ip/ip_types_api.h> #include <vnet/vnet_msg_enum.h> @@ -71,27 +68,16 @@ #include <vnet/format_fns.h> #define foreach_ip_api_msg \ +_(SW_INTERFACE_IP6_ENABLE_DISABLE, sw_interface_ip6_enable_disable) \ _(IP_TABLE_DUMP, ip_table_dump) \ _(IP_ROUTE_DUMP, ip_route_dump) \ _(IP_MTABLE_DUMP, ip_mtable_dump) \ _(IP_MROUTE_DUMP, ip_mroute_dump) \ -_(IP_NEIGHBOR_DUMP, ip_neighbor_dump) \ _(IP_MROUTE_ADD_DEL, ip_mroute_add_del) \ _(MFIB_SIGNAL_DUMP, mfib_signal_dump) \ _(IP_ADDRESS_DUMP, ip_address_dump) \ _(IP_UNNUMBERED_DUMP, ip_unnumbered_dump) \ _(IP_DUMP, ip_dump) \ -_(IP_NEIGHBOR_ADD_DEL, ip_neighbor_add_del) \ -_(SET_ARP_NEIGHBOR_LIMIT, set_arp_neighbor_limit) \ -_(IP_PROBE_NEIGHBOR, ip_probe_neighbor) \ -_(IP_SCAN_NEIGHBOR_ENABLE_DISABLE, ip_scan_neighbor_enable_disable) \ -_(WANT_IP4_ARP_EVENTS, want_ip4_arp_events) \ -_(WANT_IP6_ND_EVENTS, want_ip6_nd_events) \ -_(WANT_IP6_RA_EVENTS, want_ip6_ra_events) \ -_(PROXY_ARP_ADD_DEL, proxy_arp_add_del) \ -_(PROXY_ARP_DUMP, proxy_arp_dump) \ -_(PROXY_ARP_INTFC_ENABLE_DISABLE, proxy_arp_intfc_enable_disable) \ - _(PROXY_ARP_INTFC_DUMP, proxy_arp_intfc_dump) \ _(IP_TABLE_REPLACE_BEGIN, ip_table_replace_begin) \ _(IP_TABLE_REPLACE_END, ip_table_replace_end) \ _(IP_TABLE_FLUSH, ip_table_flush) \ @@ -100,12 +86,6 @@ _(IP_TABLE_ADD_DEL, ip_table_add_del) \ _(IP_PUNT_POLICE, ip_punt_police) \ _(IP_PUNT_REDIRECT, ip_punt_redirect) \ _(SET_IP_FLOW_HASH,set_ip_flow_hash) \ -_(SW_INTERFACE_IP6ND_RA_CONFIG, sw_interface_ip6nd_ra_config) \ -_(SW_INTERFACE_IP6ND_RA_PREFIX, sw_interface_ip6nd_ra_prefix) \ -_(IP6ND_PROXY_ADD_DEL, ip6nd_proxy_add_del) \ -_(IP6ND_PROXY_DUMP, ip6nd_proxy_dump) \ -_(IP6ND_SEND_ROUTER_SOLICITATION, ip6nd_send_router_solicitation) \ -_(SW_INTERFACE_IP6_ENABLE_DISABLE, sw_interface_ip6_enable_disable ) \ _(IP_CONTAINER_PROXY_ADD_DEL, ip_container_proxy_add_del) \ _(IP_CONTAINER_PROXY_DUMP, ip_container_proxy_dump) \ _(IOAM_ENABLE, ioam_enable) \ @@ -116,100 +96,29 @@ _(IP_SOURCE_AND_PORT_RANGE_CHECK_INTERFACE_ADD_DEL, \ ip_source_and_port_range_check_interface_add_del) \ _(IP_SOURCE_CHECK_INTERFACE_ADD_DEL, \ ip_source_check_interface_add_del) \ + _(SW_INTERFACE_IP6_SET_LINK_LOCAL_ADDRESS, \ + sw_interface_ip6_set_link_local_address) \ _(IP_REASSEMBLY_SET, ip_reassembly_set) \ _(IP_REASSEMBLY_GET, ip_reassembly_get) \ _(IP_REASSEMBLY_ENABLE_DISABLE, ip_reassembly_enable_disable) \ _(IP_PUNT_REDIRECT_DUMP, ip_punt_redirect_dump) - -static vl_api_ip_neighbor_flags_t -ip_neighbor_flags_encode (ip_neighbor_flags_t f) -{ - vl_api_ip_neighbor_flags_t v = IP_API_NEIGHBOR_FLAG_NONE; - - if (f & IP_NEIGHBOR_FLAG_STATIC) - v |= IP_API_NEIGHBOR_FLAG_STATIC; - if (f & IP_NEIGHBOR_FLAG_NO_FIB_ENTRY) - v |= IP_API_NEIGHBOR_FLAG_NO_FIB_ENTRY; - - return (clib_host_to_net_u32 (v)); -} - static void -send_ip_neighbor_details (u32 sw_if_index, - const ip46_address_t * ip_address, - const mac_address_t * mac, - ip_neighbor_flags_t flags, - vl_api_registration_t * reg, u32 context) -{ - vl_api_ip_neighbor_details_t *mp; - - mp = vl_msg_api_alloc (sizeof (*mp)); - clib_memset (mp, 0, sizeof (*mp)); - mp->_vl_msg_id = ntohs (VL_API_IP_NEIGHBOR_DETAILS); - mp->context = context; - mp->neighbor.sw_if_index = htonl (sw_if_index); - mp->neighbor.flags = ip_neighbor_flags_encode (flags); - - ip_address_encode (ip_address, IP46_TYPE_ANY, &mp->neighbor.ip_address); - mac_address_encode (mac, mp->neighbor.mac_address); - - vl_api_send_msg (reg, (u8 *) mp); -} - -static void -vl_api_ip_neighbor_dump_t_handler (vl_api_ip_neighbor_dump_t * mp) + vl_api_sw_interface_ip6_enable_disable_t_handler + (vl_api_sw_interface_ip6_enable_disable_t * mp) { - vl_api_registration_t *reg; - - reg = vl_api_client_index_to_registration (mp->client_index); - if (!reg) - return; - - u32 sw_if_index = ntohl (mp->sw_if_index); + vl_api_sw_interface_ip6_enable_disable_reply_t *rmp; + int rv = 0; - if (mp->is_ipv6) - { - ip6_neighbor_t *n, *ns; + VALIDATE_SW_IF_INDEX (mp); - ns = ip6_neighbors_entries (sw_if_index); - /* *INDENT-OFF* */ - vec_foreach (n, ns) - { - ip46_address_t nh = { - .ip6 = { - .as_u64[0] = n->key.ip6_address.as_u64[0], - .as_u64[1] = n->key.ip6_address.as_u64[1], - }, - }; - send_ip_neighbor_details (n->key.sw_if_index, &nh, - &n->mac, n->flags, - reg, mp->context); - } - /* *INDENT-ON* */ - vec_free (ns); - } - else - { - ethernet_arp_ip4_entry_t *n, *ns; + rv = ((mp->enable == 1) ? + ip6_link_enable (ntohl (mp->sw_if_index)) : + ip6_link_disable (ntohl (mp->sw_if_index))); - ns = ip4_neighbor_entries (sw_if_index); - /* *INDENT-OFF* */ - vec_foreach (n, ns) - { - ip46_address_t nh = { - .ip4 = { - .as_u32 = n->ip4_address.as_u32, - }, - }; + BAD_SW_IF_INDEX_LABEL; - send_ip_neighbor_details (n->sw_if_index, &nh, - &n->mac, n->flags, - reg, mp->context); - } - /* *INDENT-ON* */ - vec_free (ns); - } + REPLY_MACRO (VL_API_SW_INTERFACE_IP6_ENABLE_DISABLE_REPLY); } static void @@ -544,61 +453,6 @@ vl_api_ip_punt_redirect_t_handler (vl_api_ip_punt_redirect_t * mp, REPLY_MACRO (VL_API_IP_PUNT_REDIRECT_REPLY); } -static ip_neighbor_flags_t -ip_neighbor_flags_decode (vl_api_ip_neighbor_flags_t v) -{ - ip_neighbor_flags_t f = IP_NEIGHBOR_FLAG_NONE; - - v = clib_net_to_host_u32 (v); - - if (v & IP_API_NEIGHBOR_FLAG_STATIC) - f |= IP_NEIGHBOR_FLAG_STATIC; - if (v & IP_API_NEIGHBOR_FLAG_NO_FIB_ENTRY) - f |= IP_NEIGHBOR_FLAG_NO_FIB_ENTRY; - - return (f); -} - -static void -vl_api_ip_neighbor_add_del_t_handler (vl_api_ip_neighbor_add_del_t * mp, - vlib_main_t * vm) -{ - vl_api_ip_neighbor_add_del_reply_t *rmp; - ip_neighbor_flags_t flags; - u32 stats_index = ~0; - ip46_address_t ip; - mac_address_t mac; - ip46_type_t type; - int rv; - - VALIDATE_SW_IF_INDEX ((&mp->neighbor)); - - flags = ip_neighbor_flags_decode (mp->neighbor.flags); - type = ip_address_decode (&mp->neighbor.ip_address, &ip); - mac_address_decode (mp->neighbor.mac_address, &mac); - - /* - * there's no validation here of the ND/ARP entry being added. - * The expectation is that the FIB will ensure that nothing bad - * will come of adding bogus entries. - */ - if (mp->is_add) - rv = ip_neighbor_add (&ip, type, &mac, - ntohl (mp->neighbor.sw_if_index), - flags, &stats_index); - else - rv = ip_neighbor_del (&ip, type, ntohl (mp->neighbor.sw_if_index)); - - BAD_SW_IF_INDEX_LABEL; - - /* *INDENT-OFF* */ - REPLY_MACRO2 (VL_API_IP_NEIGHBOR_ADD_DEL_REPLY, - ({ - rmp->stats_index = htonl (stats_index); - })); - /* *INDENT-ON* */ -} - void ip_table_delete (fib_protocol_t fproto, u32 table_id, u8 is_api) { @@ -1024,7 +878,7 @@ vl_api_ip_dump_t_handler (vl_api_ip_dump_t * mp) { vpe_api_main_t *am = &vpe_api_main; vnet_main_t *vnm = vnet_get_main (); - vlib_main_t *vm = vlib_get_main (); + //vlib_main_t *vm = vlib_get_main (); vnet_interface_main_t *im = &vnm->interface_main; vl_api_registration_t *reg; vnet_sw_interface_t *si, *sorted_sis; @@ -1048,10 +902,10 @@ vl_api_ip_dump_t_handler (vl_api_ip_dump_t * mp) { if (!(si->flags & VNET_SW_INTERFACE_FLAG_UNNUMBERED)) { - if (mp->is_ipv6 && !ip6_interface_enabled (vm, si->sw_if_index)) - { - continue; - } + /* if (mp->is_ipv6 && !ip6_interface_enabled (vm, si->sw_if_index)) */ + /* { */ + /* continue; */ + /* } */ sw_if_index = si->sw_if_index; send_ip_details (am, reg, sw_if_index, mp->is_ipv6, mp->context); } @@ -1108,225 +962,6 @@ vl_api_set_ip_flow_hash_t_handler (vl_api_set_ip_flow_hash_t * mp) set_ip6_flow_hash (mp); } -static void - vl_api_sw_interface_ip6nd_ra_config_t_handler - (vl_api_sw_interface_ip6nd_ra_config_t * mp) -{ - vl_api_sw_interface_ip6nd_ra_config_reply_t *rmp; - vlib_main_t *vm = vlib_get_main (); - int rv = 0; - u8 is_no, suppress, managed, other, ll_option, send_unicast, cease, - default_router; - - is_no = mp->is_no == 1; - suppress = mp->suppress == 1; - managed = mp->managed == 1; - other = mp->other == 1; - ll_option = mp->ll_option == 1; - send_unicast = mp->send_unicast == 1; - cease = mp->cease == 1; - default_router = mp->default_router == 1; - - VALIDATE_SW_IF_INDEX (mp); - - rv = ip6_neighbor_ra_config (vm, ntohl (mp->sw_if_index), - suppress, managed, other, - ll_option, send_unicast, cease, - default_router, ntohl (mp->lifetime), - ntohl (mp->initial_count), - ntohl (mp->initial_interval), - ntohl (mp->max_interval), - ntohl (mp->min_interval), is_no); - - BAD_SW_IF_INDEX_LABEL; - - REPLY_MACRO (VL_API_SW_INTERFACE_IP6ND_RA_CONFIG_REPLY); -} - -static void - vl_api_sw_interface_ip6nd_ra_prefix_t_handler - (vl_api_sw_interface_ip6nd_ra_prefix_t * mp) -{ - vlib_main_t *vm = vlib_get_main (); - vl_api_sw_interface_ip6nd_ra_prefix_reply_t *rmp; - fib_prefix_t pfx; - int rv = 0; - u8 is_no, use_default, no_advertise, off_link, no_autoconfig, no_onlink; - - VALIDATE_SW_IF_INDEX (mp); - - ip_prefix_decode (&mp->prefix, &pfx); - is_no = mp->is_no == 1; - use_default = mp->use_default == 1; - no_advertise = mp->no_advertise == 1; - off_link = mp->off_link == 1; - no_autoconfig = mp->no_autoconfig == 1; - no_onlink = mp->no_onlink == 1; - - rv = ip6_neighbor_ra_prefix (vm, ntohl (mp->sw_if_index), - &pfx.fp_addr.ip6, - pfx.fp_len, use_default, - ntohl (mp->val_lifetime), - ntohl (mp->pref_lifetime), no_advertise, - off_link, no_autoconfig, no_onlink, is_no); - - BAD_SW_IF_INDEX_LABEL; - REPLY_MACRO (VL_API_SW_INTERFACE_IP6ND_RA_PREFIX_REPLY); -} - -static void -send_ip6nd_proxy_details (vl_api_registration_t * reg, - u32 context, - const ip46_address_t * addr, u32 sw_if_index) -{ - vl_api_ip6nd_proxy_details_t *mp; - - mp = vl_msg_api_alloc (sizeof (*mp)); - clib_memset (mp, 0, sizeof (*mp)); - mp->_vl_msg_id = ntohs (VL_API_IP6ND_PROXY_DETAILS); - mp->context = context; - mp->sw_if_index = htonl (sw_if_index); - - ip6_address_encode (&addr->ip6, mp->ip); - - vl_api_send_msg (reg, (u8 *) mp); -} - -typedef struct api_ip6nd_proxy_fib_table_walk_ctx_t_ -{ - u32 *indices; -} api_ip6nd_proxy_fib_table_walk_ctx_t; - -static fib_table_walk_rc_t -api_ip6nd_proxy_fib_table_walk (fib_node_index_t fei, void *arg) -{ - api_ip6nd_proxy_fib_table_walk_ctx_t *ctx = arg; - - if (fib_entry_is_sourced (fei, FIB_SOURCE_IP6_ND_PROXY)) - { - vec_add1 (ctx->indices, fei); - } - - return (FIB_TABLE_WALK_CONTINUE); -} - -static void -vl_api_ip6nd_proxy_dump_t_handler (vl_api_ip6nd_proxy_dump_t * mp) -{ - ip6_main_t *im6 = &ip6_main; - fib_table_t *fib_table; - api_ip6nd_proxy_fib_table_walk_ctx_t ctx = { - .indices = NULL, - }; - fib_node_index_t *feip; - const fib_prefix_t *pfx; - vl_api_registration_t *reg; - - reg = vl_api_client_index_to_registration (mp->client_index); - if (!reg) - return; - - /* *INDENT-OFF* */ - pool_foreach (fib_table, im6->fibs, - ({ - fib_table_walk(fib_table->ft_index, - FIB_PROTOCOL_IP6, - api_ip6nd_proxy_fib_table_walk, - &ctx); - })); - /* *INDENT-ON* */ - - vec_sort_with_function (ctx.indices, fib_entry_cmp_for_sort); - - vec_foreach (feip, ctx.indices) - { - pfx = fib_entry_get_prefix (*feip); - - send_ip6nd_proxy_details (reg, - mp->context, - &pfx->fp_addr, - fib_entry_get_resolving_interface (*feip)); - } - - vec_free (ctx.indices); -} - -static void -vl_api_ip6nd_proxy_add_del_t_handler (vl_api_ip6nd_proxy_add_del_t * mp) -{ - vl_api_ip6nd_proxy_add_del_reply_t *rmp; - ip6_address_t ip6; - int rv = 0; - - VALIDATE_SW_IF_INDEX (mp); - - ip6_address_decode (mp->ip, &ip6); - rv = ip6_neighbor_proxy_add_del (ntohl (mp->sw_if_index), &ip6, mp->is_del); - - BAD_SW_IF_INDEX_LABEL; - REPLY_MACRO (VL_API_IP6ND_PROXY_ADD_DEL_REPLY); -} - -static void - vl_api_ip6nd_send_router_solicitation_t_handler - (vl_api_ip6nd_send_router_solicitation_t * mp) -{ - vl_api_ip6nd_send_router_solicitation_reply_t *rmp; - icmp6_send_router_solicitation_params_t params; - vlib_main_t *vm = vlib_get_main (); - int rv = 0; - - VALIDATE_SW_IF_INDEX (mp); - - BAD_SW_IF_INDEX_LABEL; - REPLY_MACRO (VL_API_IP6ND_SEND_ROUTER_SOLICITATION_REPLY); - - if (rv != 0) - return; - - params.irt = ntohl (mp->irt); - params.mrt = ntohl (mp->mrt); - params.mrc = ntohl (mp->mrc); - params.mrd = ntohl (mp->mrd); - - icmp6_send_router_solicitation (vm, ntohl (mp->sw_if_index), mp->stop, - ¶ms); -} - -static void - vl_api_sw_interface_ip6_enable_disable_t_handler - (vl_api_sw_interface_ip6_enable_disable_t * mp) -{ - vlib_main_t *vm = vlib_get_main (); - vl_api_sw_interface_ip6_enable_disable_reply_t *rmp; - vnet_main_t *vnm = vnet_get_main (); - int rv = 0; - clib_error_t *error; - - vnm->api_errno = 0; - - VALIDATE_SW_IF_INDEX (mp); - - error = - (mp->enable == 1) ? enable_ip6_interface (vm, - ntohl (mp->sw_if_index)) : - disable_ip6_interface (vm, ntohl (mp->sw_if_index)); - - if (error) - { - clib_error_report (error); - rv = VNET_API_ERROR_UNSPECIFIED; - } - else - { - rv = vnm->api_errno; - } - - BAD_SW_IF_INDEX_LABEL; - - REPLY_MACRO (VL_API_SW_INTERFACE_IP6_ENABLE_DISABLE_REPLY); -} - void vl_mfib_signal_send_one (vl_api_registration_t * reg, u32 context, const mfib_signal_t * mfs) @@ -1617,6 +1252,24 @@ reply: REPLY_MACRO (VL_API_IP_SOURCE_AND_PORT_RANGE_CHECK_INTERFACE_ADD_DEL_REPLY); } +static void + vl_api_sw_interface_ip6_set_link_local_address_t_handler + (vl_api_sw_interface_ip6_set_link_local_address_t * mp) +{ + vl_api_sw_interface_ip6_set_link_local_address_reply_t *rmp; + ip6_address_t ip; + int rv; + + VALIDATE_SW_IF_INDEX (mp); + + ip6_address_decode (mp->ip, &ip); + + rv = ip6_set_link_local_address (ntohl (mp->sw_if_index), &ip); + + BAD_SW_IF_INDEX_LABEL; + REPLY_MACRO (VL_API_SW_INTERFACE_IP6_SET_LINK_LOCAL_ADDRESS_REPLY); +} + typedef union { u32 fib_index; @@ -1647,933 +1300,6 @@ static void REPLY_MACRO (VL_API_IP_SOURCE_CHECK_INTERFACE_ADD_DEL_REPLY); } -#define IP4_ARP_EVENT 3 -#define IP6_ND_EVENT 4 - -static vlib_node_registration_t ip_resolver_process_node; - -static int -arp_change_delete_callback (u32 pool_index, - const mac_address_t * mac, - u32 sw_if_index, const ip4_address_t * address) -{ - vpe_api_main_t *am = &vpe_api_main; - - if (pool_is_free_index (am->arp_events, pool_index)) - return 1; - - pool_put_index (am->arp_events, pool_index); - return 0; -} - -static void -handle_ip4_arp_event (u32 pool_index) -{ - vpe_api_main_t *vam = &vpe_api_main; - vnet_main_t *vnm = vam->vnet_main; - vlib_main_t *vm = vam->vlib_main; - vl_api_ip4_arp_event_t *event; - vl_api_ip4_arp_event_t *mp; - vl_api_registration_t *reg; - - /* Client can cancel, die, etc. */ - if (pool_is_free_index (vam->arp_events, pool_index)) - return; - - event = pool_elt_at_index (vam->arp_events, pool_index); - - reg = vl_api_client_index_to_registration (event->client_index); - if (!reg) - { - (void) vnet_add_del_ip4_arp_change_event - (vnm, arp_change_delete_callback, - event->pid, event->ip, - ip_resolver_process_node.index, IP4_ARP_EVENT, - ~0 /* pool index, notused */ , 0 /* is_add */ ); - return; - } - - if (vl_api_can_send_msg (reg)) - { - mp = vl_msg_api_alloc (sizeof (*mp)); - clib_memcpy (mp, event, sizeof (*mp)); - vl_api_send_msg (reg, (u8 *) mp); - } - else - { - static f64 last_time; - /* - * Throttle syslog msgs. - * It's pretty tempting to just revoke the registration... - */ - if (vlib_time_now (vm) > last_time + 10.0) - { - clib_warning ("arp event for %U to pid %d: queue stuffed!", - format_ip4_address, event->ip, event->pid); - last_time = vlib_time_now (vm); - } - } -} - -static int -nd_change_delete_callback (u32 pool_index, - const mac_address_t * mac, - u32 sw_if_index, const ip6_address_t * addr) -{ - vpe_api_main_t *am = &vpe_api_main; - - if (pool_is_free_index (am->nd_events, pool_index)) - return 1; - - pool_put_index (am->nd_events, pool_index); - return 0; -} - -static void -handle_ip6_nd_event (u32 pool_index) -{ - vpe_api_main_t *vam = &vpe_api_main; - vnet_main_t *vnm = vam->vnet_main; - vlib_main_t *vm = vam->vlib_main; - vl_api_ip6_nd_event_t *event; - vl_api_ip6_nd_event_t *mp; - vl_api_registration_t *reg; - - /* Client can cancel, die, etc. */ - if (pool_is_free_index (vam->nd_events, pool_index)) - return; - - event = pool_elt_at_index (vam->nd_events, pool_index); - - reg = vl_api_client_index_to_registration (event->client_index); - if (!reg) - { - (void) vnet_add_del_ip6_nd_change_event - (vnm, nd_change_delete_callback, - event->pid, event->ip, - ip_resolver_process_node.index, IP6_ND_EVENT, - ~0 /* pool index, notused */ , 0 /* is_add */ ); - return; - } - - if (vl_api_can_send_msg (reg)) - { - mp = vl_msg_api_alloc (sizeof (*mp)); - clib_memcpy (mp, event, sizeof (*mp)); - vl_api_send_msg (reg, (u8 *) mp); - } - else - { - static f64 last_time; - /* - * Throttle syslog msgs. - * It's pretty tempting to just revoke the registration... - */ - if (vlib_time_now (vm) > last_time + 10.0) - { - clib_warning ("ip6 nd event for %U to pid %d: queue stuffed!", - format_ip6_address, event->ip, event->pid); - last_time = vlib_time_now (vm); - } - } -} - -static uword -resolver_process (vlib_main_t * vm, - vlib_node_runtime_t * rt, vlib_frame_t * f) -{ - volatile f64 timeout = 100.0; - volatile uword *event_data = 0; - - while (1) - { - vlib_process_wait_for_event_or_clock (vm, timeout); - - uword event_type = - vlib_process_get_events (vm, (uword **) & event_data); - - int i; - switch (event_type) - { - case IP4_ARP_EVENT: - for (i = 0; i < vec_len (event_data); i++) - handle_ip4_arp_event (event_data[i]); - break; - - case IP6_ND_EVENT: - for (i = 0; i < vec_len (event_data); i++) - handle_ip6_nd_event (event_data[i]); - break; - - case ~0: /* timeout */ - break; - } - - vec_reset_length (event_data); - } - return 0; /* or not */ -} - -/* *INDENT-OFF* */ -VLIB_REGISTER_NODE (ip_resolver_process_node,static) = { - .function = resolver_process, - .type = VLIB_NODE_TYPE_PROCESS, - .name = "ip-route-resolver-process", -}; -/* *INDENT-ON* */ - -static int -nd_change_data_callback (u32 pool_index, const mac_address_t * new_mac, - u32 sw_if_index, const ip6_address_t * address) -{ - vpe_api_main_t *am = &vpe_api_main; - vl_api_ip6_nd_event_t *event; - - if (pool_is_free_index (am->nd_events, pool_index)) - return 1; - - event = pool_elt_at_index (am->nd_events, pool_index); - if (ethernet_mac_address_equal (event->mac, new_mac->bytes) && - sw_if_index == ntohl (event->sw_if_index)) - { - return 1; - } - - mac_address_encode (new_mac, event->mac); - event->sw_if_index = htonl (sw_if_index); - return 0; -} - -static vlib_node_registration_t wc_arp_process_node; - -enum -{ WC_ARP_REPORT, WC_ND_REPORT, RA_REPORT, REPORT_MAX }; - -static uword -wc_arp_process (vlib_main_t * vm, vlib_node_runtime_t * rt, vlib_frame_t * f) -{ - /* These cross the longjmp boundary (vlib_process_wait_for_event) - * and need to be volatile - to prevent them from being optimized into - * a register - which could change during suspension */ - - volatile wc_arp_report_t arp_prev = { 0 }; - volatile wc_nd_report_t nd_prev = { 0 }; - volatile f64 last_arp = vlib_time_now (vm); - volatile f64 last_nd = vlib_time_now (vm); - - while (1) - { - vlib_process_wait_for_event (vm); - uword event_type = WC_ARP_REPORT; - void *event_data = vlib_process_get_event_data (vm, &event_type); - - f64 now = vlib_time_now (vm); - int i; - if (event_type == WC_ARP_REPORT) - { - wc_arp_report_t *arp_events = event_data; - for (i = 0; i < vec_len (arp_events); i++) - { - /* discard dup event - cast away volatile */ - if (arp_prev.ip.as_u32 == arp_events[i].ip.as_u32 && - mac_address_equal ((const mac_address_t *) &arp_prev.mac, - &arp_events[i].mac) && - arp_prev.sw_if_index == arp_events[i].sw_if_index && - (now - last_arp) < 10.0) - { - continue; - } - arp_prev = arp_events[i]; - last_arp = now; - vpe_client_registration_t *reg; - /* *INDENT-OFF* */ - pool_foreach(reg, vpe_api_main.wc_ip4_arp_events_registrations, - ({ - vl_api_registration_t *vl_reg; - vl_reg = vl_api_client_index_to_registration (reg->client_index); - ASSERT (vl_reg != NULL); - if (reg && vl_api_can_send_msg (vl_reg)) - { - vl_api_ip4_arp_event_t * event = vl_msg_api_alloc (sizeof *event); - clib_memset (event, 0, sizeof *event); - event->_vl_msg_id = htons (VL_API_IP4_ARP_EVENT); - event->client_index = reg->client_index; - event->pid = reg->client_pid; - event->mac_ip = 1; - ip4_address_encode(&arp_events[i].ip, event->ip); - event->sw_if_index = htonl(arp_events[i].sw_if_index); - mac_address_encode(&arp_events[i].mac, event->mac); - vl_api_send_msg (vl_reg, (u8 *) event); - } - })); - /* *INDENT-ON* */ - } - } - else if (event_type == WC_ND_REPORT) - { - wc_nd_report_t *nd_events = event_data; - for (i = 0; i < vec_len (nd_events); i++) - { - /* discard dup event - cast away volatile */ - if (ip6_address_is_equal ((const ip6_address_t *) &nd_prev.ip6, - &nd_events[i].ip6) - && mac_address_equal ((const mac_address_t *) &nd_prev.mac, - &nd_events[i].mac) - && nd_prev.sw_if_index == nd_events[i].sw_if_index - && (now - last_nd) < 10.0) - { - continue; - } - nd_prev = nd_events[i]; - last_nd = now; - vpe_client_registration_t *reg; - /* *INDENT-OFF* */ - pool_foreach(reg, vpe_api_main.wc_ip6_nd_events_registrations, - ({ - vl_api_registration_t *vl_reg; - vl_reg = vl_api_client_index_to_registration (reg->client_index); - if (vl_reg && vl_api_can_send_msg (vl_reg)) - { - vl_api_ip6_nd_event_t * event = vl_msg_api_alloc (sizeof *event); - clib_memset (event, 0, sizeof *event); - event->_vl_msg_id = htons (VL_API_IP6_ND_EVENT); - event->client_index = reg->client_index; - event->pid = reg->client_pid; - event->mac_ip = 1; - ip6_address_encode(&nd_events[i].ip6, event->ip); - event->sw_if_index = htonl(nd_events[i].sw_if_index); - mac_address_encode(&nd_events[i].mac, event->mac); - vl_api_send_msg (vl_reg, (u8 *) event); - } - })); - /* *INDENT-ON* */ - } - } - else if (event_type == RA_REPORT) - { - ra_report_t *ra_events = event_data; - for (i = 0; i < vec_len (ra_events); i++) - { - ip6_neighbor_public_main_t *npm = &ip6_neighbor_public_main; - call_ip6_neighbor_callbacks (&ra_events[i], - npm->ra_report_functions); - - vpe_client_registration_t *reg; - /* *INDENT-OFF* */ - pool_foreach(reg, vpe_api_main.ip6_ra_events_registrations, - ({ - vl_api_registration_t *vl_reg; - vl_reg = - vl_api_client_index_to_registration (reg->client_index); - if (vl_reg && vl_api_can_send_msg (vl_reg)) - { - u32 event_size = - sizeof (vl_api_ip6_ra_event_t) + - vec_len (ra_events[i].prefixes) * - sizeof (vl_api_ip6_ra_prefix_info_t); - vl_api_ip6_ra_event_t *event = - vl_msg_api_alloc (event_size); - clib_memset (event, 0, event_size); - event->_vl_msg_id = htons (VL_API_IP6_RA_EVENT); - event->client_index = reg->client_index; - event->pid = reg->client_pid; - - event->sw_if_index = clib_host_to_net_u32 (ra_events[i].sw_if_index); - - ip6_address_encode (&ra_events[i].router_address, - event->router_addr); - - event->current_hop_limit = ra_events[i].current_hop_limit; - event->flags = ra_events[i].flags; - event->router_lifetime_in_sec = - clib_host_to_net_u16 (ra_events - [i].router_lifetime_in_sec); - event->neighbor_reachable_time_in_msec = - clib_host_to_net_u32 (ra_events - [i].neighbor_reachable_time_in_msec); - event->time_in_msec_between_retransmitted_neighbor_solicitations - = - clib_host_to_net_u32 (ra_events - [i].time_in_msec_between_retransmitted_neighbor_solicitations); - - event->n_prefixes = - clib_host_to_net_u32 (vec_len (ra_events[i].prefixes)); - vl_api_ip6_ra_prefix_info_t *prefix = - (typeof (prefix)) event->prefixes; - u32 j; - for (j = 0; j < vec_len (ra_events[i].prefixes); j++) - { - ra_report_prefix_info_t *info = - &ra_events[i].prefixes[j]; - ip_prefix_encode(&info->prefix, &prefix->prefix); - prefix->flags = info->flags; - prefix->valid_time = - clib_host_to_net_u32 (info->valid_time); - prefix->preferred_time = - clib_host_to_net_u32 (info->preferred_time); - prefix++; - } - - vl_api_send_msg (vl_reg, (u8 *) event); - } - })); - /* *INDENT-ON* */ - vec_free (ra_events[i].prefixes); - } - } - vlib_process_put_event_data (vm, event_data); - } - - return 0; -} - -/* *INDENT-OFF* */ -VLIB_REGISTER_NODE (wc_arp_process_node,static) = { - .function = wc_arp_process, - .type = VLIB_NODE_TYPE_PROCESS, - .name = "wildcard-ip4-arp-publisher-process", -}; -/* *INDENT-ON* */ - -static int -arp_change_data_callback (u32 pool_index, - const mac_address_t * mac, - u32 sw_if_index, const ip4_address_t * address) -{ - vpe_api_main_t *am = &vpe_api_main; - vl_api_ip4_arp_event_t *event; - - if (pool_is_free_index (am->arp_events, pool_index)) - return 1; - - event = pool_elt_at_index (am->arp_events, pool_index); - if (ethernet_mac_address_equal (event->mac, mac->bytes) && - sw_if_index == ntohl (event->sw_if_index)) - { - return 1; - } - - mac_address_encode (mac, event->mac); - event->sw_if_index = htonl (sw_if_index); - return 0; -} - -static void -vl_api_want_ip4_arp_events_t_handler (vl_api_want_ip4_arp_events_t * mp) -{ - vpe_api_main_t *am = &vpe_api_main; - vnet_main_t *vnm = vnet_get_main (); - vl_api_want_ip4_arp_events_reply_t *rmp; - ip4_address_t ip; - int rv = 0; - - ip4_address_decode (mp->ip, &ip); - - if (ip.as_u32 == 0) - { - uword *p = - hash_get (am->wc_ip4_arp_events_registration_hash, mp->client_index); - vpe_client_registration_t *rp; - if (p) - { - if (mp->enable_disable) - { - clib_warning ("pid %d: already enabled...", mp->pid); - rv = VNET_API_ERROR_INVALID_REGISTRATION; - goto reply; - } - else - { - rp = - pool_elt_at_index (am->wc_ip4_arp_events_registrations, p[0]); - pool_put (am->wc_ip4_arp_events_registrations, rp); - hash_unset (am->wc_ip4_arp_events_registration_hash, - mp->client_index); - if (pool_elts (am->wc_ip4_arp_events_registrations) == 0) - wc_arp_set_publisher_node (~0, REPORT_MAX); - goto reply; - } - } - if (mp->enable_disable == 0) - { - clib_warning ("pid %d: already disabled...", mp->pid); - rv = VNET_API_ERROR_INVALID_REGISTRATION; - goto reply; - } - pool_get (am->wc_ip4_arp_events_registrations, rp); - rp->client_index = mp->client_index; - rp->client_pid = mp->pid; - hash_set (am->wc_ip4_arp_events_registration_hash, rp->client_index, - rp - am->wc_ip4_arp_events_registrations); - wc_arp_set_publisher_node (wc_arp_process_node.index, WC_ARP_REPORT); - goto reply; - } - - if (mp->enable_disable) - { - vl_api_ip4_arp_event_t *event; - pool_get (am->arp_events, event); - rv = vnet_add_del_ip4_arp_change_event - (vnm, arp_change_data_callback, - mp->pid, mp->ip /* addr, in net byte order */ , - ip_resolver_process_node.index, - IP4_ARP_EVENT, event - am->arp_events, 1 /* is_add */ ); - - if (rv) - { - pool_put (am->arp_events, event); - goto reply; - } - clib_memset (event, 0, sizeof (*event)); - - /* Python API expects events to have no context */ - event->_vl_msg_id = htons (VL_API_IP4_ARP_EVENT); - event->client_index = mp->client_index; - memcpy (event->ip, mp->ip, 4); - event->pid = mp->pid; - if (ip.as_u32 == 0) - event->mac_ip = 1; - } - else - { - rv = vnet_add_del_ip4_arp_change_event - (vnm, arp_change_delete_callback, - mp->pid, mp->ip /* addr, in net byte order */ , - ip_resolver_process_node.index, - IP4_ARP_EVENT, ~0 /* pool index */ , 0 /* is_add */ ); - } -reply: - REPLY_MACRO (VL_API_WANT_IP4_ARP_EVENTS_REPLY); -} - -static clib_error_t * -want_ip4_arp_events_reaper (u32 client_index) -{ - vpe_client_registration_t *rp; - vl_api_ip4_arp_event_t *event; - u32 *to_delete, *event_id; - vpe_api_main_t *am; - vnet_main_t *vnm; - uword *p; - - am = &vpe_api_main; - vnm = vnet_get_main (); - to_delete = NULL; - - /* clear out all of its pending resolutions */ - /* *INDENT-OFF* */ - pool_foreach(event, am->arp_events, - ({ - if (event->client_index == client_index) - { - vec_add1(to_delete, event - am->arp_events); - } - })); - /* *INDENT-ON* */ - - vec_foreach (event_id, to_delete) - { - event = pool_elt_at_index (am->arp_events, *event_id); - vnet_add_del_ip4_arp_change_event - (vnm, arp_change_delete_callback, - event->pid, event->ip, - ip_resolver_process_node.index, IP4_ARP_EVENT, - ~0 /* pool index, notused */ , 0 /* is_add */ ); - } - vec_free (to_delete); - - /* remove from the registration hash */ - p = hash_get (am->wc_ip4_arp_events_registration_hash, client_index); - - if (p) - { - rp = pool_elt_at_index (am->wc_ip4_arp_events_registrations, p[0]); - pool_put (am->wc_ip4_arp_events_registrations, rp); - hash_unset (am->wc_ip4_arp_events_registration_hash, client_index); - if (pool_elts (am->wc_ip4_arp_events_registrations) == 0) - wc_arp_set_publisher_node (~0, REPORT_MAX); - } - return (NULL); -} - -VL_MSG_API_REAPER_FUNCTION (want_ip4_arp_events_reaper); - -static void -vl_api_want_ip6_nd_events_t_handler (vl_api_want_ip6_nd_events_t * mp) -{ - vpe_api_main_t *am = &vpe_api_main; - vnet_main_t *vnm = vnet_get_main (); - vl_api_want_ip6_nd_events_reply_t *rmp; - ip6_address_t ip6; - int rv = 0; - - ip6_address_decode (mp->ip, &ip6); - - if (ip6_address_is_zero (&ip6)) - { - uword *p = - hash_get (am->wc_ip6_nd_events_registration_hash, mp->client_index); - vpe_client_registration_t *rp; - if (p) - { - if (mp->enable_disable) - { - clib_warning ("pid %d: already enabled...", mp->pid); - rv = VNET_API_ERROR_INVALID_REGISTRATION; - goto reply; - } - else - { - rp = - pool_elt_at_index (am->wc_ip6_nd_events_registrations, p[0]); - pool_put (am->wc_ip6_nd_events_registrations, rp); - hash_unset (am->wc_ip6_nd_events_registration_hash, - mp->client_index); - if (pool_elts (am->wc_ip6_nd_events_registrations) == 0) - wc_nd_set_publisher_node (~0, REPORT_MAX); - goto reply; - } - } - if (mp->enable_disable == 0) - { - clib_warning ("pid %d: already disabled...", mp->pid); - rv = VNET_API_ERROR_INVALID_REGISTRATION; - goto reply; - } - pool_get (am->wc_ip6_nd_events_registrations, rp); - rp->client_index = mp->client_index; - rp->client_pid = mp->pid; - hash_set (am->wc_ip6_nd_events_registration_hash, rp->client_index, - rp - am->wc_ip6_nd_events_registrations); - wc_nd_set_publisher_node (wc_arp_process_node.index, WC_ND_REPORT); - goto reply; - } - - if (mp->enable_disable) - { - vl_api_ip6_nd_event_t *event; - pool_get (am->nd_events, event); - - rv = vnet_add_del_ip6_nd_change_event - (vnm, nd_change_data_callback, - mp->pid, &ip6, - ip_resolver_process_node.index, - IP6_ND_EVENT, event - am->nd_events, 1 /* is_add */ ); - - if (rv) - { - pool_put (am->nd_events, event); - goto reply; - } - clib_memset (event, 0, sizeof (*event)); - - event->_vl_msg_id = ntohs (VL_API_IP6_ND_EVENT); - event->client_index = mp->client_index; - ip6_address_encode (&ip6, event->ip); - event->pid = mp->pid; - } - else - { - rv = vnet_add_del_ip6_nd_change_event - (vnm, nd_change_delete_callback, - mp->pid, &ip6 /* addr, in net byte order */ , - ip_resolver_process_node.index, - IP6_ND_EVENT, ~0 /* pool index */ , 0 /* is_add */ ); - } -reply: - REPLY_MACRO (VL_API_WANT_IP6_ND_EVENTS_REPLY); -} - -static clib_error_t * -want_ip6_nd_events_reaper (u32 client_index) -{ - - vpe_client_registration_t *rp; - vl_api_ip6_nd_event_t *event; - u32 *to_delete, *event_id; - vpe_api_main_t *am; - vnet_main_t *vnm; - uword *p; - - am = &vpe_api_main; - vnm = vnet_get_main (); - to_delete = NULL; - - /* clear out all of its pending resolutions */ - /* *INDENT-OFF* */ - pool_foreach(event, am->nd_events, - ({ - if (event->client_index == client_index) - { - vec_add1(to_delete, event - am->nd_events); - } - })); - /* *INDENT-ON* */ - - vec_foreach (event_id, to_delete) - { - event = pool_elt_at_index (am->nd_events, *event_id); - vnet_add_del_ip6_nd_change_event - (vnm, nd_change_delete_callback, - event->pid, event->ip, - ip_resolver_process_node.index, IP6_ND_EVENT, - ~0 /* pool index, notused */ , 0 /* is_add */ ); - } - vec_free (to_delete); - - /* remove from the registration hash */ - p = hash_get (am->wc_ip6_nd_events_registration_hash, client_index); - - if (p) - { - rp = pool_elt_at_index (am->wc_ip6_nd_events_registrations, p[0]); - pool_put (am->wc_ip6_nd_events_registrations, rp); - hash_unset (am->wc_ip6_nd_events_registration_hash, client_index); - if (pool_elts (am->wc_ip6_nd_events_registrations) == 0) - wc_nd_set_publisher_node (~0, REPORT_MAX); - } - return (NULL); -} - -VL_MSG_API_REAPER_FUNCTION (want_ip6_nd_events_reaper); - -static void -vl_api_want_ip6_ra_events_t_handler (vl_api_want_ip6_ra_events_t * mp) -{ - vpe_api_main_t *am = &vpe_api_main; - vl_api_want_ip6_ra_events_reply_t *rmp; - int rv = 0; - - uword *p = hash_get (am->ip6_ra_events_registration_hash, mp->client_index); - vpe_client_registration_t *rp; - if (p) - { - if (mp->enable_disable) - { - clib_warning ("pid %d: already enabled...", ntohl (mp->pid)); - rv = VNET_API_ERROR_INVALID_REGISTRATION; - goto reply; - } - else - { - rp = pool_elt_at_index (am->ip6_ra_events_registrations, p[0]); - pool_put (am->ip6_ra_events_registrations, rp); - hash_unset (am->ip6_ra_events_registration_hash, mp->client_index); - goto reply; - } - } - if (mp->enable_disable == 0) - { - clib_warning ("pid %d: already disabled...", ntohl (mp->pid)); - rv = VNET_API_ERROR_INVALID_REGISTRATION; - goto reply; - } - pool_get (am->ip6_ra_events_registrations, rp); - rp->client_index = mp->client_index; - rp->client_pid = ntohl (mp->pid); - hash_set (am->ip6_ra_events_registration_hash, rp->client_index, - rp - am->ip6_ra_events_registrations); - -reply: - REPLY_MACRO (VL_API_WANT_IP6_RA_EVENTS_REPLY); -} - -static clib_error_t * -want_ip6_ra_events_reaper (u32 client_index) -{ - vpe_api_main_t *am = &vpe_api_main; - vpe_client_registration_t *rp; - uword *p; - - p = hash_get (am->ip6_ra_events_registration_hash, client_index); - - if (p) - { - rp = pool_elt_at_index (am->ip6_ra_events_registrations, p[0]); - pool_put (am->ip6_ra_events_registrations, rp); - hash_unset (am->ip6_ra_events_registration_hash, client_index); - } - return (NULL); -} - -VL_MSG_API_REAPER_FUNCTION (want_ip6_ra_events_reaper); - -static void -vl_api_proxy_arp_add_del_t_handler (vl_api_proxy_arp_add_del_t * mp) -{ - vl_api_proxy_arp_add_del_reply_t *rmp; - ip4_address_t lo, hi; - u32 fib_index; - int rv; - - fib_index = fib_table_find (FIB_PROTOCOL_IP4, ntohl (mp->proxy.table_id)); - - if (~0 == fib_index) - { - rv = VNET_API_ERROR_NO_SUCH_FIB; - goto out; - } - - ip4_address_decode (mp->proxy.low, &lo); - ip4_address_decode (mp->proxy.hi, &hi); - - rv = vnet_proxy_arp_add_del (&lo, &hi, fib_index, mp->is_add == 0); - -out: - REPLY_MACRO (VL_API_PROXY_ARP_ADD_DEL_REPLY); -} - -typedef struct proxy_arp_walk_ctx_t_ -{ - vl_api_registration_t *reg; - u32 context; -} proxy_arp_walk_ctx_t; - -static walk_rc_t -send_proxy_arp_details (const ip4_address_t * lo_addr, - const ip4_address_t * hi_addr, - u32 fib_index, void *data) -{ - vl_api_proxy_arp_details_t *mp; - proxy_arp_walk_ctx_t *ctx; - - ctx = data; - - mp = vl_msg_api_alloc (sizeof (*mp)); - clib_memset (mp, 0, sizeof (*mp)); - mp->_vl_msg_id = ntohs (VL_API_PROXY_ARP_DETAILS); - mp->context = ctx->context; - mp->proxy.table_id = htonl (fib_index); - - ip4_address_encode (lo_addr, mp->proxy.low); - ip4_address_encode (hi_addr, mp->proxy.hi); - - vl_api_send_msg (ctx->reg, (u8 *) mp); - - return (WALK_CONTINUE); -} - -static void -vl_api_proxy_arp_dump_t_handler (vl_api_proxy_arp_dump_t * mp) -{ - vl_api_registration_t *reg; - - reg = vl_api_client_index_to_registration (mp->client_index); - if (!reg) - return; - - proxy_arp_walk_ctx_t wctx = { - .reg = reg, - .context = mp->context, - }; - - proxy_arp_walk (send_proxy_arp_details, &wctx); -} - -static walk_rc_t -send_proxy_arp_intfc_details (u32 sw_if_index, void *data) -{ - vl_api_proxy_arp_intfc_details_t *mp; - proxy_arp_walk_ctx_t *ctx; - - ctx = data; - - mp = vl_msg_api_alloc (sizeof (*mp)); - clib_memset (mp, 0, sizeof (*mp)); - mp->_vl_msg_id = ntohs (VL_API_PROXY_ARP_INTFC_DETAILS); - mp->context = ctx->context; - mp->sw_if_index = htonl (sw_if_index); - - vl_api_send_msg (ctx->reg, (u8 *) mp); - - return (WALK_CONTINUE); -} - -static void -vl_api_proxy_arp_intfc_dump_t_handler (vl_api_proxy_arp_intfc_dump_t * mp) -{ - vl_api_registration_t *reg; - - reg = vl_api_client_index_to_registration (mp->client_index); - if (!reg) - return; - - proxy_arp_walk_ctx_t wctx = { - .reg = reg, - .context = mp->context, - }; - - proxy_arp_intfc_walk (send_proxy_arp_intfc_details, &wctx); -} - -static void - vl_api_proxy_arp_intfc_enable_disable_t_handler - (vl_api_proxy_arp_intfc_enable_disable_t * mp) -{ - int rv = 0; - vnet_main_t *vnm = vnet_get_main (); - vl_api_proxy_arp_intfc_enable_disable_reply_t *rmp; - - VALIDATE_SW_IF_INDEX (mp); - - rv = vnet_proxy_arp_enable_disable (vnm, - ntohl (mp->sw_if_index), - mp->enable_disable); - - BAD_SW_IF_INDEX_LABEL; - - REPLY_MACRO (VL_API_PROXY_ARP_INTFC_ENABLE_DISABLE_REPLY); -} - -static void -vl_api_ip_probe_neighbor_t_handler (vl_api_ip_probe_neighbor_t * mp) -{ - int rv = 0; - vlib_main_t *vm = vlib_get_main (); - vl_api_ip_probe_neighbor_reply_t *rmp; - clib_error_t *error; - ip46_address_t dst; - ip46_type_t itype; - - VALIDATE_SW_IF_INDEX (mp); - - u32 sw_if_index = ntohl (mp->sw_if_index); - itype = ip_address_decode (&mp->dst, &dst); - - if (IP46_TYPE_IP6 == itype) - error = ip6_probe_neighbor (vm, &dst.ip6, sw_if_index, 0); - else - error = ip4_probe_neighbor (vm, &dst.ip4, sw_if_index, 0); - - if (error) - { - clib_error_report (error); - rv = clib_error_get_code (error); - } - - BAD_SW_IF_INDEX_LABEL; - - REPLY_MACRO (VL_API_IP_PROBE_NEIGHBOR_REPLY); -} - -static void - vl_api_ip_scan_neighbor_enable_disable_t_handler - (vl_api_ip_scan_neighbor_enable_disable_t * mp) -{ - int rv = 0; - vl_api_ip_scan_neighbor_enable_disable_reply_t *rmp; - ip_neighbor_scan_arg_t arg; - - arg.mode = mp->mode; - arg.scan_interval = mp->scan_interval; - arg.max_proc_time = mp->max_proc_time; - arg.max_update = mp->max_update; - arg.scan_int_delay = mp->scan_int_delay; - arg.stale_threshold = mp->stale_threshold; - ip_neighbor_scan_enable_disable (&arg); - - REPLY_MACRO (VL_API_IP_SCAN_NEIGHBOR_ENABLE_DISABLE_REPLY); -} static void vl_api_ip_table_replace_begin_t_handler (vl_api_ip_table_replace_begin_t * mp) @@ -2661,34 +1387,6 @@ vl_api_ip_table_flush_t_handler (vl_api_ip_table_flush_t * mp) REPLY_MACRO (VL_API_IP_TABLE_FLUSH_REPLY); } -static void -vl_api_set_arp_neighbor_limit_t_handler (vl_api_set_arp_neighbor_limit_t * mp) -{ - int rv; - vl_api_set_arp_neighbor_limit_reply_t *rmp; - vnet_main_t *vnm = vnet_get_main (); - clib_error_t *error; - - vnm->api_errno = 0; - - if (mp->is_ipv6) - error = ip6_set_neighbor_limit (ntohl (mp->arp_neighbor_limit)); - else - error = ip4_set_arp_limit (ntohl (mp->arp_neighbor_limit)); - - if (error) - { - clib_error_report (error); - rv = VNET_API_ERROR_UNSPECIFIED; - } - else - { - rv = vnm->api_errno; - } - - REPLY_MACRO (VL_API_SET_ARP_NEIGHBOR_LIMIT_REPLY); -} - void vl_api_ip_reassembly_set_t_handler (vl_api_ip_reassembly_set_t * mp) { @@ -2948,8 +1646,6 @@ ip_api_hookup (vlib_main_t * vm) */ setup_message_id_table (am); - ra_set_publisher_node (wc_arp_process_node.index, RA_REPORT); - return 0; } diff --git a/src/vnet/ip/ip_neighbor.c b/src/vnet/ip/ip_neighbor.c deleted file mode 100644 index ef61dde0e05..00000000000 --- a/src/vnet/ip/ip_neighbor.c +++ /dev/null @@ -1,492 +0,0 @@ -/* - * src/vnet/ip/ip_neighbor.c: ip neighbor generic handling - * - * Copyright (c) 2018 Cisco and/or its affiliates. - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at: - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ - -#include <vnet/vnet.h> -#include <vnet/ip/ip.h> -#include <vnet/ip/ip6_neighbor.h> -#include <vnet/ip/ip_neighbor.h> -#include <vnet/ethernet/arp.h> - -/* - * IP neighbor scan parameter defaults are as follows: - * - Scan interval : 60 sec - * - Max processing allowed per run : 20 usec - * - Max probe/delete operations per run : 10 - * - Scan interrupt delay to resume scan : 1 msec - * - Neighbor stale threshold : 4 x scan-interval - */ -#define IP_NEIGHBOR_DEF_SCAN_INTERVAL (60.0) -#define IP_NEIGHBOR_DEF_MAX_PROC_TIME (20e-6) -#define IP_NEIGHBOR_DEF_SCAN_INT_DELAY (1e-3) -#define IP_NEIGHBOR_DEF_STALE (4*IP_NEIGHBOR_DEF_SCAN_INTERVAL) -#define IP_NEIGHBOR_DEF_MAX_UPDATE 10 - -typedef struct -{ - f64 scan_interval; /* Periodic scan interval */ - f64 max_proc_time; /* Max processing time allowed per run */ - f64 scan_int_delay; /* Scan interrupt delay to resume scan */ - f64 stale_threshold; /* IP neighbor stale threshod */ - u8 max_update; /* Max probe/delete actions allowed per run */ - u8 mode; /* IP neighbor scan mode */ -} ip_neighbor_scan_config_t; - -static ip_neighbor_scan_config_t ip_neighbor_scan_conf; - -u8 * -format_ip_neighbor_flags (u8 * s, va_list * args) -{ - const ip_neighbor_flags_t flags = va_arg (*args, int); - - if (flags & IP_NEIGHBOR_FLAG_STATIC) - s = format (s, "S"); - - if (flags & IP_NEIGHBOR_FLAG_DYNAMIC) - s = format (s, "D"); - - if (flags & IP_NEIGHBOR_FLAG_NO_FIB_ENTRY) - s = format (s, "N"); - - return s; -} - -int -ip_neighbor_add (const ip46_address_t * ip, - ip46_type_t type, - const mac_address_t * mac, - u32 sw_if_index, - ip_neighbor_flags_t flags, u32 * stats_index) -{ - fib_protocol_t fproto; - vnet_link_t linkt; - int rv; - - /* - * there's no validation here of the ND/ARP entry being added. - * The expectation is that the FIB will ensure that nothing bad - * will come of adding bogus entries. - */ - if (IP46_TYPE_IP6 == type) - { - rv = vnet_set_ip6_ethernet_neighbor (vlib_get_main (), - sw_if_index, &ip->ip6, mac, flags); - fproto = FIB_PROTOCOL_IP6; - linkt = VNET_LINK_IP6; - } - else - { - ethernet_arp_ip4_over_ethernet_address_t a = { - .ip4 = ip->ip4, - .mac = *mac, - }; - - rv = - vnet_arp_set_ip4_over_ethernet (vnet_get_main (), sw_if_index, &a, - flags); - fproto = FIB_PROTOCOL_IP4; - linkt = VNET_LINK_IP4; - } - - if (0 == rv && stats_index) - *stats_index = adj_nbr_find (fproto, linkt, ip, sw_if_index); - - return (rv); -} - -int -ip_neighbor_del (const ip46_address_t * ip, ip46_type_t type, u32 sw_if_index) -{ - int rv; - - if (IP46_TYPE_IP6 == type) - { - rv = vnet_unset_ip6_ethernet_neighbor (vlib_get_main (), - sw_if_index, &ip->ip6); - } - else - { - ethernet_arp_ip4_over_ethernet_address_t a = { - .ip4 = ip->ip4, - }; - - rv = - vnet_arp_unset_ip4_over_ethernet (vnet_get_main (), sw_if_index, &a); - } - - return (rv); -} - -void -ip_neighbor_scan_enable_disable (ip_neighbor_scan_arg_t * arg) -{ - ip_neighbor_scan_config_t *cfg = &ip_neighbor_scan_conf; - - cfg->mode = arg->mode; - - if (arg->mode) - { - cfg->scan_interval = arg->scan_interval ? - arg->scan_interval * 60.0 : IP_NEIGHBOR_DEF_SCAN_INTERVAL; - cfg->max_proc_time = arg->max_proc_time ? - arg->max_proc_time * 1e-6 : IP_NEIGHBOR_DEF_MAX_PROC_TIME; - cfg->scan_int_delay = arg->scan_int_delay ? - arg->scan_int_delay * 1e-3 : IP_NEIGHBOR_DEF_SCAN_INT_DELAY; - cfg->stale_threshold = arg->stale_threshold ? - arg->stale_threshold * 60.0 : cfg->scan_interval * 4; - cfg->max_update = arg->max_update ? - cfg->max_update : IP_NEIGHBOR_DEF_MAX_UPDATE; - } - else - cfg->scan_interval = IP_NEIGHBOR_DEF_SCAN_INTERVAL; -} - -static_always_inline u32 -ip_neighbor_scan (vlib_main_t * vm, f64 start_time, u32 start_idx, - u8 is_ip6, u8 delete_stale, u8 * update_count) -{ - vnet_main_t *vnm = vnet_get_main (); - ip_neighbor_scan_config_t *cfg = &ip_neighbor_scan_conf; - ethernet_arp_ip4_entry_t *np4 = ip4_neighbors_pool (); - ip6_neighbor_t *np6 = ip6_neighbors_pool (); - ethernet_arp_ip4_entry_t *n4; - ip6_neighbor_t *n6; - u32 curr_idx = start_idx; - u32 loop_count = 0; - f64 delta, update_time; - - if (!is_ip6) - { - if (pool_is_free_index (np4, start_idx)) - curr_idx = pool_next_index (np4, start_idx); - } - else - { - if (pool_is_free_index (np6, start_idx)) - curr_idx = pool_next_index (np6, start_idx); - } - - while (curr_idx != ~0) - { - /* allow no more than 10 neighbor updates or 20 usec of scan */ - if ((update_count[0] >= cfg->max_update) || - (((loop_count % 100) == 0) && - ((vlib_time_now (vm) - start_time) > cfg->max_proc_time))) - break; - - if (!is_ip6) - { - n4 = pool_elt_at_index (np4, curr_idx); - if (n4->flags & IP_NEIGHBOR_FLAG_STATIC) - goto next_neighbor; - update_time = n4->time_last_updated; - } - else - { - n6 = pool_elt_at_index (np6, curr_idx); - if (n6->flags & IP_NEIGHBOR_FLAG_STATIC) - goto next_neighbor; - update_time = n6->time_last_updated; - } - - delta = start_time - update_time; - if (delete_stale && (delta >= cfg->stale_threshold)) - { - update_count[0]++; - /* delete stale neighbor */ - if (!is_ip6) - { - ethernet_arp_ip4_over_ethernet_address_t delme = { - .ip4.as_u32 = n4->ip4_address.as_u32, - .mac = n4->mac, - }; - - vnet_arp_unset_ip4_over_ethernet (vnm, n4->sw_if_index, &delme); - } - else - { - vnet_unset_ip6_ethernet_neighbor - (vm, n6->key.sw_if_index, &n6->key.ip6_address); - } - } - else if (delta >= cfg->scan_interval) - { - update_count[0]++; - /* probe neighbor */ - if (!is_ip6) - ip4_probe_neighbor (vm, &n4->ip4_address, n4->sw_if_index, 1); - else - ip6_probe_neighbor (vm, &n6->key.ip6_address, - n6->key.sw_if_index, 1); - } - - next_neighbor: - loop_count++; - - if (!is_ip6) - curr_idx = pool_next_index (np4, curr_idx); - else - curr_idx = pool_next_index (np6, curr_idx); - } - - return curr_idx; -} - -static uword -neighbor_scan_process (vlib_main_t * vm, - vlib_node_runtime_t * rt, vlib_frame_t * f) -{ - ip_neighbor_scan_config_t *cfg = &ip_neighbor_scan_conf; - f64 timeout = IP_NEIGHBOR_DEF_SCAN_INTERVAL; - f64 start, next_scan = CLIB_TIME_MAX; - u32 ip4_nidx = 0; /* ip4 neighbor pool index */ - u32 ip6_nidx = 0; /* ip6 neighbor pool index */ - uword *event_data = 0; - u8 purge4 = 0, purge6 = 0; /* flags to purge stale entry during scan */ - u8 update; - - cfg->mode = IP_SCAN_DISABLED; - cfg->scan_interval = IP_NEIGHBOR_DEF_SCAN_INTERVAL; - cfg->scan_int_delay = IP_NEIGHBOR_DEF_SCAN_INTERVAL; - - while (1) - { - vlib_process_wait_for_event_or_clock (vm, timeout); - vlib_process_get_events (vm, &event_data); - vec_reset_length (event_data); - - start = vlib_time_now (vm); - update = 0; - - if ((ip4_nidx == 0) && (ip6_nidx == 0)) /* starting a fresh scan */ - next_scan = start + cfg->scan_interval; - - if ((cfg->mode & IP_SCAN_V4_NEIGHBORS) == 0) - ip4_nidx = ~0; /* disable ip4 neighbor scan */ - - if ((cfg->mode & IP_SCAN_V6_NEIGHBORS) == 0) - ip6_nidx = ~0; /* disable ip6 neighbor scan */ - - if (ip4_nidx != ~0) /* scan ip4 neighbors */ - ip4_nidx = ip_neighbor_scan (vm, start, ip4_nidx, /* ip4 */ 0, - purge4, &update); - - if (ip6_nidx != ~0) /* scan ip6 neighbors */ - ip6_nidx = ip_neighbor_scan (vm, start, ip6_nidx, /* ip6 */ 1, - purge6, &update); - - if ((ip4_nidx == ~0) && (ip6_nidx == ~0)) - { /* scan complete */ - timeout = next_scan - vlib_time_now (vm); - ip4_nidx = ip6_nidx = 0; - purge4 = cfg->mode & IP_SCAN_V4_NEIGHBORS; - purge6 = cfg->mode & IP_SCAN_V6_NEIGHBORS; - } - else /* scan incomplete */ - timeout = cfg->scan_int_delay; - - if (timeout > cfg->scan_interval) - timeout = cfg->scan_interval; - else if (timeout < cfg->scan_int_delay) - timeout = cfg->scan_int_delay; - - } - return 0; -} - -/* *INDENT-OFF* */ -VLIB_REGISTER_NODE (neighbor_scan_process_node,static) = { - .function = neighbor_scan_process, - .type = VLIB_NODE_TYPE_PROCESS, - .name = "ip-neighbor-scan-process", -}; -/* *INDENT-ON* */ - -static clib_error_t * -ip_neighbor_scan_cli (vlib_main_t * vm, unformat_input_t * input, - vlib_cli_command_t * cmd) -{ - unformat_input_t _line_input, *line_input = &_line_input; - clib_error_t *error = 0; - u32 interval = 0, time = 0, update = 0, delay = 0, stale = 0; - ip_neighbor_scan_arg_t arg; - - clib_memset (&arg, 0, sizeof (arg)); - arg.mode = IP_SCAN_V46_NEIGHBORS; - - /* Get a line of input. */ - if (!unformat_user (input, unformat_line_input, line_input)) - { - ip_neighbor_scan_enable_disable (&arg); - return error; - } - - while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) - { - if (unformat (line_input, "ip4")) - arg.mode = IP_SCAN_V4_NEIGHBORS; - - else if (unformat (line_input, "ip6")) - arg.mode = IP_SCAN_V6_NEIGHBORS; - - else if (unformat (line_input, "both")) - arg.mode = IP_SCAN_V46_NEIGHBORS; - - else if (unformat (line_input, "disable")) - arg.mode = IP_SCAN_DISABLED; - - else if (unformat (line_input, "interval %d", &interval)) - arg.scan_interval = interval; - - else if (unformat (line_input, "max-time %d", &time)) - arg.max_proc_time = time; - - else if (unformat (line_input, "max-update %d", &update)) - arg.max_update = update; - - else if (unformat (line_input, "delay %d", &delay)) - arg.scan_int_delay = delay; - - else if (unformat (line_input, "stale %d", &stale)) - arg.stale_threshold = stale; - - else - { - error = clib_error_return (0, "unknown input '%U'", - format_unformat_error, line_input); - goto done; - } - } - - if (interval > 255) - { - error = clib_error_return (0, "interval cannot exceed 255 minutes."); - goto done; - } - if (time > 255) - { - error = clib_error_return (0, "max-time cannot exceed 255 usec."); - goto done; - } - if (update > 255) - { - error = clib_error_return (0, "max-update cannot exceed 255."); - goto done; - } - if (delay > 255) - { - error = clib_error_return (0, "delay cannot exceed 255 msec."); - goto done; - } - if (stale > 255) - { - error = clib_error_return (0, "stale cannot exceed 255 minutes."); - goto done; - } - - ip_neighbor_scan_enable_disable (&arg); - -done: - unformat_free (line_input); - - return error; -} - -/*? - * The '<em>ip scan-neighbor</em>' command can be used to enable and disable - * periodic IP neighbor scan and change various scan parameters. - * - * @note The default parameters used for IP neighbor scan should work fine - * under normal conditions. They should not be changed from the default unless - * properly tested to work as desired. - * - * @cliexpar - * Example of enabling IP neighbor scan: - * @cliexcmd{ip neighbor-scan enable} -?*/ -/* *INDENT-OFF* */ -VLIB_CLI_COMMAND (ip_scan_neighbor_command, static) = { - .path = "ip scan-neighbor", - .function = ip_neighbor_scan_cli, - .short_help = "ip scan-neighbor [ip4|ip6|both|disable] [interval <n-min>] [max-time <n-usec>] [max-update <n>] [delay <n-msec>] [stale <n-min>]", - .is_mp_safe = 1, -}; -/* *INDENT-ON* */ - -static u8 * -format_ip_scan_mode (u8 * s, va_list * args) -{ - u8 mode = va_arg (*args, u32); - switch (mode) - { - case IP_SCAN_V4_NEIGHBORS: - return format (s, "IPv4"); - case IP_SCAN_V6_NEIGHBORS: - return format (s, "IPv6"); - case IP_SCAN_V46_NEIGHBORS: - return format (s, "IPv4 and IPv6"); - } - return format (s, "unknown"); -} - -static clib_error_t * -show_ip_neighbor_scan (vlib_main_t * vm, unformat_input_t * input, - vlib_cli_command_t * cmd) -{ - ip_neighbor_scan_config_t *cfg = &ip_neighbor_scan_conf; - - if (cfg->mode == 0) - vlib_cli_output (vm, - "IP neighbor scan disabled - current time is %.4f sec", - vlib_time_now (vm)); - else - vlib_cli_output (vm, "IP neighbor scan enabled for %U neighbors - " - "current time is %.4f sec\n " - "Full_scan_interval: %f min " - "Stale_purge_threshod: %f min\n " - "Max_process_time: %f usec Max_updates %d " - "Delay_to_resume_after_max_limit: %f msec", - format_ip_scan_mode, cfg->mode, - vlib_time_now (vm), cfg->scan_interval / 60.0, - cfg->stale_threshold / 60.0, cfg->max_proc_time / 1e-6, - cfg->max_update, cfg->scan_int_delay / 1e-3); - return 0; -} - -/*? - * The '<em>show ip scan-neighbor</em>' command can be used to show the current - * periodic IP neighbor scan parameters - * - * @cliexpar - * Example of showing IP neighbor scan current parameters: - * @cliexcmd{show ip neighbor-scan} -?*/ -/* *INDENT-OFF* */ -VLIB_CLI_COMMAND (show_ip_scan_neighbor_command, static) = { - .path = "show ip scan-neighbor", - .function = show_ip_neighbor_scan, - .short_help = "show ip scan-neighbor", - .is_mp_safe = 1, -}; -/* *INDENT-ON* */ - -/* - * fd.io coding-style-patch-verification: ON - * - * Local Variables: - * eval: (c-set-style "gnu") - * End: - */ diff --git a/src/vnet/ip/lookup.c b/src/vnet/ip/lookup.c index c1fbc429b97..3b30fb9ce5e 100644 --- a/src/vnet/ip/lookup.c +++ b/src/vnet/ip/lookup.c @@ -50,8 +50,6 @@ #include <vnet/dpo/receive_dpo.h> #include <vnet/dpo/ip_null_dpo.h> #include <vnet/dpo/l3_proxy_dpo.h> -#include <vnet/ip/ip6_neighbor.h> -#include <vnet/ethernet/arp.h> /** * @file @@ -1029,210 +1027,6 @@ VLIB_CLI_COMMAND (ip_mroute_command, static) = }; /* *INDENT-ON* */ -/* - * The next two routines address a longstanding script hemorrhoid. - * Probing a v4 or v6 neighbor needs to appear to be synchronous, - * or dependent route-adds will simply fail. - */ -static clib_error_t * -ip6_probe_neighbor_wait (vlib_main_t * vm, ip6_address_t * a, u32 sw_if_index, - int retry_count) -{ - vnet_main_t *vnm = vnet_get_main (); - clib_error_t *e; - int i; - int resolved = 0; - uword event_type; - uword *event_data = 0; - - ASSERT (vlib_in_process_context (vm)); - - if (retry_count > 0) - vnet_register_ip6_neighbor_resolution_event - (vnm, a, vlib_get_current_process (vm)->node_runtime.node_index, - 1 /* event */ , 0 /* data */ ); - - for (i = 0; i < retry_count; i++) - { - /* The interface may be down, etc. */ - e = ip6_probe_neighbor (vm, a, sw_if_index, 0); - - if (e) - return e; - - vlib_process_wait_for_event_or_clock (vm, 1.0); - event_type = vlib_process_get_events (vm, &event_data); - switch (event_type) - { - case 1: /* resolved... */ - vlib_cli_output (vm, "Resolved %U", format_ip6_address, a); - resolved = 1; - goto done; - - case ~0: /* timeout */ - break; - - default: - clib_warning ("unknown event_type %d", event_type); - } - vec_reset_length (event_data); - } - -done: - - if (!resolved) - return clib_error_return (0, "Resolution failed for %U", - format_ip6_address, a); - return 0; -} - -static clib_error_t * -ip4_probe_neighbor_wait (vlib_main_t * vm, ip4_address_t * a, u32 sw_if_index, - int retry_count) -{ - vnet_main_t *vnm = vnet_get_main (); - clib_error_t *e; - int i; - int resolved = 0; - uword event_type; - uword *event_data = 0; - - ASSERT (vlib_in_process_context (vm)); - - if (retry_count > 0) - vnet_register_ip4_arp_resolution_event - (vnm, a, vlib_get_current_process (vm)->node_runtime.node_index, - 1 /* event */ , 0 /* data */ ); - - for (i = 0; i < retry_count; i++) - { - /* The interface may be down, etc. */ - e = ip4_probe_neighbor (vm, a, sw_if_index, 0); - - if (e) - return e; - - vlib_process_wait_for_event_or_clock (vm, 1.0); - event_type = vlib_process_get_events (vm, &event_data); - switch (event_type) - { - case 1: /* resolved... */ - vlib_cli_output (vm, "Resolved %U", format_ip4_address, a); - resolved = 1; - goto done; - - case ~0: /* timeout */ - break; - - default: - clib_warning ("unknown event_type %d", event_type); - } - vec_reset_length (event_data); - } - -done: - - vec_reset_length (event_data); - - if (!resolved) - return clib_error_return (0, "Resolution failed for %U", - format_ip4_address, a); - return 0; -} - -static clib_error_t * -probe_neighbor_address (vlib_main_t * vm, - unformat_input_t * input, vlib_cli_command_t * cmd) -{ - vnet_main_t *vnm = vnet_get_main (); - unformat_input_t _line_input, *line_input = &_line_input; - ip4_address_t a4; - ip6_address_t a6; - clib_error_t *error = 0; - u32 sw_if_index = ~0; - int retry_count = 3; - int is_ip4 = 1; - int address_set = 0; - - /* Get a line of input. */ - if (!unformat_user (input, unformat_line_input, line_input)) - return 0; - - while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) - { - if (unformat_user (line_input, unformat_vnet_sw_interface, vnm, - &sw_if_index)) - ; - else if (unformat (line_input, "retry %d", &retry_count)) - ; - - else if (unformat (line_input, "%U", unformat_ip4_address, &a4)) - address_set++; - else if (unformat (line_input, "%U", unformat_ip6_address, &a6)) - { - address_set++; - is_ip4 = 0; - } - else - { - error = clib_error_return (0, "unknown input '%U'", - format_unformat_error, line_input); - goto done; - } - } - - if (sw_if_index == ~0) - { - error = clib_error_return (0, "Interface required, not set."); - goto done; - } - if (address_set == 0) - { - error = clib_error_return (0, "ip address required, not set."); - goto done; - } - if (address_set > 1) - { - error = clib_error_return (0, "Multiple ip addresses not supported."); - goto done; - } - - if (is_ip4) - error = ip4_probe_neighbor_wait (vm, &a4, sw_if_index, retry_count); - else - error = ip6_probe_neighbor_wait (vm, &a6, sw_if_index, retry_count); - -done: - unformat_free (line_input); - - return error; -} - -/*? - * The '<em>ip probe-neighbor</em>' command ARPs for IPv4 addresses or - * attempts IPv6 neighbor discovery depending on the supplied IP address - * format. - * - * @note This command will not immediately affect the indicated FIB; it - * is not suitable for use in establishing a FIB entry prior to adding - * recursive FIB entries. As in: don't use it in a script to probe a - * gateway prior to adding a default route. It won't work. Instead, - * configure a static ARP cache entry [see '<em>set ip arp</em>'], or - * a static IPv6 neighbor [see '<em>set ip6 neighbor</em>']. - * - * @cliexpar - * Example of probe for an IPv4 address: - * @cliexcmd{ip probe-neighbor GigabitEthernet2/0/0 172.16.1.2} -?*/ -/* *INDENT-OFF* */ -VLIB_CLI_COMMAND (ip_probe_neighbor_command, static) = { - .path = "ip probe-neighbor", - .function = probe_neighbor_address, - .short_help = "ip probe-neighbor <interface> <ip4-addr> | <ip6-addr> [retry nn]", - .is_mp_safe = 1, -}; -/* *INDENT-ON* */ - clib_error_t * vnet_ip_container_proxy_add_del (vnet_ip_container_proxy_args_t * args) { diff --git a/src/vnet/ip6-nd/ip6_mld.c b/src/vnet/ip6-nd/ip6_mld.c new file mode 100644 index 00000000000..bd5f8489b7b --- /dev/null +++ b/src/vnet/ip6-nd/ip6_mld.c @@ -0,0 +1,541 @@ +/* + * ip/ip6_neighbor.c: IP6 neighbor handling + * + * Copyright (c) 2010 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/ip6-nd/ip6_nd.h> + +#include <vnet/ip/ip.h> +#include <vnet/ip-neighbor/ip_neighbor_dp.h> + +#include <vnet/fib/ip6_fib.h> +#include <vnet/ip/ip6_link.h> +#include <vnet/ip/ip6_ll_table.h> + +/** + * @file + * @brief IPv6 Neighbor Adjacency and Neighbor Discovery. + * + * The files contains the API and CLI code for managing IPv6 neighbor + * adjacency tables and neighbor discovery logic. + */ + +/* *INDENT-OFF*/ +/* multicast listener report packet format for ethernet. */ +typedef CLIB_PACKED (struct +{ + ip6_hop_by_hop_ext_t ext_hdr; + ip6_router_alert_option_t alert; + ip6_padN_option_t pad; + icmp46_header_t icmp; + u16 rsvd; + u16 num_addr_records; + icmp6_multicast_address_record_t records[0]; +}) icmp6_multicast_listener_report_header_t; + +typedef CLIB_PACKED (struct +{ + ip6_header_t ip; + icmp6_multicast_listener_report_header_t report_hdr; +}) icmp6_multicast_listener_report_packet_t; +/* *INDENT-ON*/ + +typedef struct +{ + /* group information */ + u16 num_sources; + u8 type; + ip6_address_t mcast_address; + ip6_address_t *mcast_source_address_pool; +} ip6_mldp_group_t; + +typedef struct ip6_nd_t_ +{ + /* local information */ + u32 sw_if_index; + int all_routers_mcast; + + /* MLDP group information */ + ip6_mldp_group_t *mldp_group_pool; + + /* Hash table mapping address to index in mldp address pool. */ + mhash_t address_to_mldp_index; + +} ip6_mld_t; + + +static ip6_link_delegate_id_t ip6_mld_delegate_id; +static ip6_mld_t *ip6_mld_pool; + +///// + +static inline ip6_mld_t * +ip6_mld_get_itf (u32 sw_if_index) +{ + index_t imi; + + imi = ip6_link_delegate_get (sw_if_index, ip6_mld_delegate_id); + + if (INDEX_INVALID != imi) + return (pool_elt_at_index (ip6_mld_pool, imi)); + + return (NULL); +} + +/** + * @brief Add a multicast Address to the advertised MLD set + */ +static void +ip6_neighbor_add_mld_prefix (ip6_mld_t * imd, ip6_address_t * addr) +{ + ip6_mldp_group_t *mcast_group_info; + uword *p; + + /* lookup mldp info for this interface */ + p = mhash_get (&imd->address_to_mldp_index, addr); + mcast_group_info = p ? pool_elt_at_index (imd->mldp_group_pool, p[0]) : 0; + + /* add address */ + if (!mcast_group_info) + { + /* add */ + u32 mi; + pool_get_zero (imd->mldp_group_pool, mcast_group_info); + + mi = mcast_group_info - imd->mldp_group_pool; + mhash_set (&imd->address_to_mldp_index, addr, mi, /* old_value */ + 0); + + mcast_group_info->type = 4; + mcast_group_info->mcast_source_address_pool = 0; + mcast_group_info->num_sources = 0; + clib_memcpy (&mcast_group_info->mcast_address, addr, + sizeof (ip6_address_t)); + } +} + +/** + * @brief Delete a multicast Address from the advertised MLD set + */ +static void +ip6_neighbor_del_mld_prefix (ip6_mld_t * imd, ip6_address_t * addr) +{ + ip6_mldp_group_t *mcast_group_info; + uword *p; + + p = mhash_get (&imd->address_to_mldp_index, &addr); + mcast_group_info = p ? pool_elt_at_index (imd->mldp_group_pool, p[0]) : 0; + + if (mcast_group_info) + { + mhash_unset (&imd->address_to_mldp_index, &addr, + /* old_value */ 0); + pool_put (imd->mldp_group_pool, mcast_group_info); + } +} + +/** + * @brief Add a multicast Address to the advertised MLD set + */ +static void +ip6_neighbor_add_mld_grp (ip6_mld_t * a, + ip6_multicast_address_scope_t scope, + ip6_multicast_link_local_group_id_t group) +{ + ip6_address_t addr; + + ip6_set_reserved_multicast_address (&addr, scope, group); + + ip6_neighbor_add_mld_prefix (a, &addr); +} + +static const ethernet_interface_t * +ip6_mld_get_eth_itf (u32 sw_if_index) +{ + const vnet_sw_interface_t *sw; + + /* lookup radv container - ethernet interfaces only */ + sw = vnet_get_sup_sw_interface (vnet_get_main (), sw_if_index); + if (sw->type == VNET_SW_INTERFACE_TYPE_HARDWARE) + return (ethernet_get_interface (ðernet_main, sw->hw_if_index)); + + return (NULL); +} + +/** + * @brief create and initialize router advertisement parameters with default + * values for this intfc + */ +static void +ip6_mld_link_enable (u32 sw_if_index) +{ + const ethernet_interface_t *eth; + ip6_mld_t *imd; + + eth = ip6_mld_get_eth_itf (sw_if_index); + + if (NULL == eth) + return; + + ASSERT (INDEX_INVALID == ip6_link_delegate_get (sw_if_index, + ip6_mld_delegate_id)); + + pool_get_zero (ip6_mld_pool, imd); + + imd->sw_if_index = sw_if_index; + + mhash_init (&imd->address_to_mldp_index, sizeof (uword), + sizeof (ip6_address_t)); + + /* add multicast groups we will always be reporting */ + ip6_neighbor_add_mld_grp (imd, + IP6_MULTICAST_SCOPE_link_local, + IP6_MULTICAST_GROUP_ID_all_hosts); + ip6_neighbor_add_mld_grp (imd, + IP6_MULTICAST_SCOPE_link_local, + IP6_MULTICAST_GROUP_ID_all_routers); + ip6_neighbor_add_mld_grp (imd, + IP6_MULTICAST_SCOPE_link_local, + IP6_MULTICAST_GROUP_ID_mldv2_routers); + + ip6_link_delegate_update (sw_if_index, ip6_mld_delegate_id, + imd - ip6_mld_pool); +} + +static void +ip6_mld_delegate_disable (index_t imdi) +{ + ip6_mldp_group_t *m; + ip6_mld_t *imd; + + imd = pool_elt_at_index (ip6_mld_pool, imdi); + + /* clean MLD pools */ + /* *INDENT-OFF* */ + pool_flush (m, imd->mldp_group_pool, + ({ + mhash_unset (&imd->address_to_mldp_index, &m->mcast_address, 0); + })); + /* *INDENT-ON* */ + + pool_free (imd->mldp_group_pool); + + mhash_free (&imd->address_to_mldp_index); + + pool_put (ip6_mld_pool, imd); +} + +/* send an mldpv2 report */ +static void +ip6_neighbor_send_mldpv2_report (u32 sw_if_index) +{ + vnet_main_t *vnm = vnet_get_main (); + vlib_main_t *vm = vnm->vlib_main; + int bogus_length; + + ip6_mld_t *imd; + u16 payload_length; + vlib_buffer_t *b0; + ip6_header_t *ip0; + u32 *to_next; + vlib_frame_t *f; + u32 bo0; + u32 n_to_alloc = 1; + + icmp6_multicast_listener_report_header_t *rh0; + icmp6_multicast_listener_report_packet_t *rp0; + + if (! !vnet_sw_interface_is_admin_up (vnm, sw_if_index)) + return; + + imd = ip6_mld_get_itf (sw_if_index); + + if (NULL == imd) + return; + + /* send report now - build a mldpv2 report packet */ + if (0 == vlib_buffer_alloc (vm, &bo0, n_to_alloc)) + { + alloc_fail: + clib_warning ("buffer allocation failure"); + return; + } + + b0 = vlib_get_buffer (vm, bo0); + + /* adjust the sizeof the buffer to just include the ipv6 header */ + b0->current_length = sizeof (icmp6_multicast_listener_report_packet_t); + + payload_length = sizeof (icmp6_multicast_listener_report_header_t); + + b0->error = ICMP6_ERROR_NONE; + + rp0 = vlib_buffer_get_current (b0); + ip0 = (ip6_header_t *) & rp0->ip; + rh0 = (icmp6_multicast_listener_report_header_t *) & rp0->report_hdr; + + clib_memset (rp0, 0x0, sizeof (icmp6_multicast_listener_report_packet_t)); + + ip0->ip_version_traffic_class_and_flow_label = + clib_host_to_net_u32 (0x6 << 28); + + ip0->protocol = IP_PROTOCOL_IP6_HOP_BY_HOP_OPTIONS; + /* for DEBUG - vnet driver won't seem to emit router alerts */ + /* ip0->protocol = IP_PROTOCOL_ICMP6; */ + ip0->hop_limit = 1; + + rh0->icmp.type = ICMP6_multicast_listener_report_v2; + + /* source address MUST be the link-local address */ + ip6_address_copy (&ip0->src_address, + ip6_get_link_local_address (sw_if_index)); + + /* destination is all mldpv2 routers */ + ip6_set_reserved_multicast_address (&ip0->dst_address, + IP6_MULTICAST_SCOPE_link_local, + IP6_MULTICAST_GROUP_ID_mldv2_routers); + + /* add reports here */ + ip6_mldp_group_t *m; + int num_addr_records = 0; + icmp6_multicast_address_record_t rr; + + /* fill in the hop-by-hop extension header (router alert) info */ + rh0->ext_hdr.next_hdr = IP_PROTOCOL_ICMP6; + rh0->ext_hdr.n_data_u64s = 0; + + rh0->alert.type = IP6_MLDP_ALERT_TYPE; + rh0->alert.len = 2; + rh0->alert.value = 0; + + rh0->pad.type = 1; + rh0->pad.len = 0; + + rh0->icmp.checksum = 0; + + /* *INDENT-OFF* */ + pool_foreach (m, imd->mldp_group_pool, + ({ + rr.type = m->type; + rr.aux_data_len_u32s = 0; + rr.num_sources = clib_host_to_net_u16 (m->num_sources); + clib_memcpy(&rr.mcast_addr, &m->mcast_address, sizeof(ip6_address_t)); + + num_addr_records++; + + if(vlib_buffer_add_data (vm, &bo0, (void *)&rr, + sizeof(icmp6_multicast_address_record_t))) + { + vlib_buffer_free (vm, &bo0, 1); + goto alloc_fail; + } + + payload_length += sizeof( icmp6_multicast_address_record_t); + })); + /* *INDENT-ON* */ + + rh0->rsvd = 0; + rh0->num_addr_records = clib_host_to_net_u16 (num_addr_records); + + /* update lengths */ + ip0->payload_length = clib_host_to_net_u16 (payload_length); + + rh0->icmp.checksum = ip6_tcp_udp_icmp_compute_checksum (vm, b0, ip0, + &bogus_length); + ASSERT (bogus_length == 0); + + /* + * OK to override w/ no regard for actual FIB, because + * ip6-rewrite only looks at the adjacency. + */ + vnet_buffer (b0)->sw_if_index[VLIB_RX] = + vnet_main.local_interface_sw_if_index; + + vnet_buffer (b0)->ip.adj_index[VLIB_TX] = + ip6_link_get_mcast_adj (sw_if_index); + b0->flags |= VNET_BUFFER_F_LOCALLY_ORIGINATED; + + vlib_node_t *node = vlib_get_node_by_name (vm, (u8 *) "ip6-rewrite-mcast"); + + f = vlib_get_frame_to_node (vm, node->index); + to_next = vlib_frame_vector_args (f); + to_next[0] = bo0; + f->n_vectors = 1; + + vlib_put_frame_to_node (vm, node->index, f); + return; +} + +/* send a RA or update the timer info etc.. */ +static uword +ip6_mld_timer_event (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + vnet_main_t *vnm = vnet_get_main (); + ip6_mld_t *imd; + + /* Interface ip6 radv info list */ + /* *INDENT-OFF* */ + pool_foreach (imd, ip6_mld_pool, + ({ + if(!vnet_sw_interface_is_admin_up (vnm, imd->sw_if_index)) + { + imd->all_routers_mcast = 0; + continue; + } + + /* Make sure that we've joined the all-routers multicast group */ + if(!imd->all_routers_mcast) + { + /* send MDLP_REPORT_EVENT message */ + ip6_neighbor_send_mldpv2_report(imd->sw_if_index); + imd->all_routers_mcast = 1; + } + })); + /* *INDENT-ON* */ + + return 0; +} + +static uword +ip6_mld_event_process (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + uword event_type; + + /* init code here */ + + while (1) + { + vlib_process_wait_for_event_or_clock (vm, 1. /* seconds */ ); + + if (!vlib_process_get_event_data (vm, &event_type)) + { + /* No events found: timer expired. */ + /* process interface list and send RAs as appropriate, update timer info */ + ip6_mld_timer_event (vm, node, frame); + } + /* else; no events */ + } + return frame->n_vectors; +} + +/* *INDENT-OFF* */ +VLIB_REGISTER_NODE (ip6_mld_event_process_node) = { + .function = ip6_mld_event_process, + .name = "ip6-mld-process", + .type = VLIB_NODE_TYPE_PROCESS, +}; +/* *INDENT-ON* */ + +static u8 * +format_ip6_mld (u8 * s, va_list * args) +{ + index_t imi = va_arg (*args, index_t); + u32 indent = va_arg (*args, u32); + ip6_mldp_group_t *m; + ip6_mld_t *imd; + + imd = pool_elt_at_index (ip6_mld_pool, imi); + + s = format (s, "%UJoined group address(es):\n", format_white_space, indent); + + /* *INDENT-OFF* */ + pool_foreach (m, imd->mldp_group_pool, + ({ + s = format (s, "%U%U\n", + format_white_space, indent+2, + format_ip6_address, + &m->mcast_address); + })); + /* *INDENT-ON* */ + + return (s); +} + +/** + * @brief callback when an interface address is added or deleted + */ +static void +ip6_mld_address_add (u32 imi, + const ip6_address_t * address, u8 address_oength) +{ + ip6_mld_t *imd; + ip6_address_t a; + + imd = pool_elt_at_index (ip6_mld_pool, imi); + + /* create solicited node multicast address for this interface address */ + ip6_set_solicited_node_multicast_address (&a, 0); + + a.as_u8[0xd] = address->as_u8[0xd]; + a.as_u8[0xe] = address->as_u8[0xe]; + a.as_u8[0xf] = address->as_u8[0xf]; + + ip6_neighbor_add_mld_prefix (imd, &a); +} + +static void +ip6_mld_address_del (u32 imi, + const ip6_address_t * address, u8 address_oength) +{ + ip6_mld_t *imd; + ip6_address_t a; + + imd = pool_elt_at_index (ip6_mld_pool, imi); + + /* create solicited node multicast address for this interface address */ + ip6_set_solicited_node_multicast_address (&a, 0); + + a.as_u8[0xd] = address->as_u8[0xd]; + a.as_u8[0xe] = address->as_u8[0xe]; + a.as_u8[0xf] = address->as_u8[0xf]; + + ip6_neighbor_del_mld_prefix (imd, &a); +} + +/** + * VFT for registering as a delegate to an IP6 link + */ +const static ip6_link_delegate_vft_t ip6_mld_delegate_vft = { + .ildv_disable = ip6_mld_delegate_disable, + .ildv_enable = ip6_mld_link_enable, + .ildv_format = format_ip6_mld, + .ildv_addr_add = ip6_mld_address_add, + .ildv_addr_del = ip6_mld_address_del, +}; + +static clib_error_t * +ip6_mld_init (vlib_main_t * vm) +{ + ip6_mld_delegate_id = ip6_link_delegate_register (&ip6_mld_delegate_vft); + + return (NULL); +} + +/* *INDENT-OFF* */ +VLIB_INIT_FUNCTION (ip6_mld_init) = +{ + .runs_after = VLIB_INITS("icmp6_init"), +}; +/* *INDENT-ON* */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip6-nd/ip6_nd.api b/src/vnet/ip6-nd/ip6_nd.api new file mode 100644 index 00000000000..91b5faf9bdf --- /dev/null +++ b/src/vnet/ip6-nd/ip6_nd.api @@ -0,0 +1,236 @@ +/* Hey Emacs use -*- mode: C -*- */ +/* + * Copyright (c) 2018 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. + */ + +/** \file + + This file defines vpp IP control-plane API messages which are generally + called through a shared memory interface. +*/ + +option version = "1.0.0"; + +import "vnet/ip/ip_types.api"; +import "vnet/interface_types.api"; + +/** \brief IPv6 router advertisement config request + @param client_index - opaque cookie to identify the sender + @param context - sender context, to match reply w/ request + @param suppress - + @param managed - + @param other - + @param ll_option - + @param send_unicast - + @param cease - + @param is_no - + @param default_router - + @param max_interval - + @param min_interval - + @param lifetime - + @param initial_count - + @param initial_interval - +*/ +autoreply define sw_interface_ip6nd_ra_config +{ + u32 client_index; + u32 context; + vl_api_interface_index_t sw_if_index; + u8 suppress; + u8 managed; + u8 other; + u8 ll_option; + u8 send_unicast; + u8 cease; + bool is_no; + u8 default_router; + u32 max_interval; + u32 min_interval; + u32 lifetime; + u32 initial_count; + u32 initial_interval; +}; + +/** \brief IPv6 router advertisement prefix config request + @param client_index - opaque cookie to identify the sender + @param context - sender context, to match reply w/ request + @param sw_if_index - The interface the RA prefix information is for + @param prefix - The prefix to advertise + @param use_default - Revert to default settings + @param no_advertise - Do not advertise this prefix + @param off_link - The prefix is off link (it is not configured on the interface) + Configures the L-flag, When set, indicates that this + prefix can be used for on-link determination. + @param no_autoconfig - Setting for the A-flag. When + set indicates that this prefix can be used for + stateless address configuration. + @param no_onlink - The prefix is not on link. Make sure this is consistent + with the off_link parameter else YMMV + @param is_no - add/delete + @param val_lifetime - The length of time in + seconds (relative to the time the packet is sent) + that the prefix is valid for the purpose of on-link + determination. A value of all one bits + (0xffffffff) represents infinity + @param pref_lifetime - The length of time in + seconds (relative to the time the packet is sent) + that addresses generated from the prefix via + stateless address autoconfiguration remain + preferred [ADDRCONF]. A value of all one bits + (0xffffffff) represents infinity. +*/ +autoreply define sw_interface_ip6nd_ra_prefix +{ + u32 client_index; + u32 context; + vl_api_interface_index_t sw_if_index; + vl_api_prefix_t prefix; + bool use_default; + bool no_advertise; + bool off_link; + bool no_autoconfig; + bool no_onlink; + bool is_no; + u32 val_lifetime; + u32 pref_lifetime; +}; + +/** \brief IPv6 ND proxy config + @param client_index - opaque cookie to identify the sender + @param context - sender context, to match reply w/ request + @param sw_if_index - The interface the host is on + @param ip - The address of the host for which to proxy for + @param is_add - Adding or deleting +*/ +autoreply define ip6nd_proxy_add_del +{ + u32 client_index; + u32 context; + vl_api_interface_index_t sw_if_index; + bool is_add; + vl_api_ip6_address_t ip; +}; + +/** \brief IPv6 ND proxy details returned after request + @param context - sender context, to match reply w/ request + @param sw_if_index - The interface the host is on + @param ip - The address of the host for which to proxy for +*/ +define ip6nd_proxy_details +{ + u32 context; + vl_api_interface_index_t sw_if_index; + vl_api_ip6_address_t ip; +}; + +/** \brief IPv6 ND proxy dump request + @param context - sender context, to match reply w/ request +*/ +define ip6nd_proxy_dump +{ + u32 client_index; + u32 context; +}; + +/** \brief Start / stop sending router solicitation + @param client_index - opaque cookie to identify the sender + @param context - sender context, to match reply w/ request + @param irt - initial retransmission time + @param mrt - maximum retransmission time + @param mrc - maximum retransmission count + @param mrd - maximum retransmission duration + @param sw_if_index - software interface index of interface + for sending router solicitation + @param stop - if non-zero then stop sending router solicitation, + otherwise start sending router solicitation +*/ +autoreply define ip6nd_send_router_solicitation +{ + u32 client_index; + u32 context; + u32 irt; + u32 mrt; + u32 mrc; + u32 mrd; + vl_api_interface_index_t sw_if_index; + bool stop; +}; + +service { + rpc want_ip6_ra_events returns want_ip6_ra_events_reply + events ip6_ra_event; +}; + +/** \brief Register for ip6 router advertisement events + @param client_index - opaque cookie to identify the sender + @param context - sender context, to match reply w/ request + @param enable - 1 => register for events, 0 => cancel registration + @param pid - sender's pid +*/ +autoreply define want_ip6_ra_events +{ + u32 client_index; + u32 context; + bool enable; + u32 pid; +}; + +/** \brief Struct representing RA prefix info + @param prefix - RA prefix info destination address + @param flags - RA prefix info flags + @param valid_time - RA prefix info valid time + @param preferred_time - RA prefix info preferred time +*/ +typedef ip6_ra_prefix_info +{ + vl_api_prefix_t prefix; + u8 flags; + u32 valid_time; + u32 preferred_time; +}; + +/** \brief Tell client about a router advertisement event + @param client_index - opaque cookie to identify the sender + @param pid - client pid registered to receive notification + @param current_hop_limit - RA current hop limit + @param flags - RA flags + @param router_lifetime_in_sec - RA lifetime in seconds + @param router_addr - The router's address + @param neighbor_reachable_time_in_msec - RA neighbor reachable time in msec + @param time_in_msec_between_retransmitted_neighbor_solicitations - + time in msec between retransmitted neighbor solicitations + @param n_prefixes - + @param prefixes - +*/ +define ip6_ra_event +{ + u32 client_index; + u32 pid; + vl_api_interface_index_t sw_if_index; + vl_api_ip6_address_t router_addr; + u8 current_hop_limit; + u8 flags; + u16 router_lifetime_in_sec; + u32 neighbor_reachable_time_in_msec; + u32 time_in_msec_between_retransmitted_neighbor_solicitations; + u32 n_prefixes; + vl_api_ip6_ra_prefix_info_t prefixes[n_prefixes]; +}; + + +/* + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip6-nd/ip6_nd.c b/src/vnet/ip6-nd/ip6_nd.c new file mode 100644 index 00000000000..eb89eb817f6 --- /dev/null +++ b/src/vnet/ip6-nd/ip6_nd.c @@ -0,0 +1,482 @@ +/* + * ip/ip6_neighbor.c: IP6 neighbor handling + * + * Copyright (c) 2010 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/ip6-nd/ip6_nd.h> + +#include <vnet/ip-neighbor/ip_neighbor.h> +#include <vnet/ip-neighbor/ip_neighbor_dp.h> + +#include <vnet/fib/ip6_fib.h> +#include <vnet/ip/ip6_link.h> +#include <vnet/ip/ip6_ll_table.h> + +/** + * @file + * @brief IPv6 Neighbor Adjacency and Neighbor Discovery. + * + * The files contains the API and CLI code for managing IPv6 neighbor + * adjacency tables and neighbor discovery logic. + */ + +#define DEF_MAX_RADV_INTERVAL 200 +#define DEF_MIN_RADV_INTERVAL .75 * DEF_MAX_RADV_INTERVAL + +typedef struct ip6_nd_t_ +{ + /* local information */ + u32 sw_if_index; + + /* stats */ + u32 n_solicitations_rcvd; + u32 n_solicitations_dropped; +} ip6_nd_t; + +static ip6_link_delegate_id_t ip6_nd_delegate_id; +static ip6_nd_t *ip6_nd_pool; + + +typedef enum +{ + ICMP6_NEIGHBOR_SOLICITATION_NEXT_DROP, + ICMP6_NEIGHBOR_SOLICITATION_NEXT_REPLY, + ICMP6_NEIGHBOR_SOLICITATION_N_NEXT, +} icmp6_neighbor_solicitation_or_advertisement_next_t; + +static_always_inline uword +icmp6_neighbor_solicitation_or_advertisement (vlib_main_t * vm, + vlib_node_runtime_t * node, + vlib_frame_t * frame, + uword is_solicitation) +{ + vnet_main_t *vnm = vnet_get_main (); + ip6_main_t *im = &ip6_main; + uword n_packets = frame->n_vectors; + u32 *from, *to_next; + u32 n_left_from, n_left_to_next, next_index, n_advertisements_sent; + icmp6_neighbor_discovery_option_type_t option_type; + vlib_node_runtime_t *error_node = + vlib_node_get_runtime (vm, ip6_icmp_input_node.index); + int bogus_length; + + from = vlib_frame_vector_args (frame); + n_left_from = n_packets; + next_index = node->cached_next_index; + + if (node->flags & VLIB_NODE_FLAG_TRACE) + vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors, + /* stride */ 1, + sizeof (icmp6_input_trace_t)); + + option_type = + (is_solicitation + ? ICMP6_NEIGHBOR_DISCOVERY_OPTION_source_link_layer_address + : ICMP6_NEIGHBOR_DISCOVERY_OPTION_target_link_layer_address); + n_advertisements_sent = 0; + + 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) + { + vlib_buffer_t *p0; + ip6_header_t *ip0; + icmp6_neighbor_solicitation_or_advertisement_header_t *h0; + icmp6_neighbor_discovery_ethernet_link_layer_address_option_t *o0; + u32 bi0, options_len0, sw_if_index0, next0, error0; + u32 ip6_sadd_link_local, ip6_sadd_unspecified; + int is_rewrite0; + u32 ni0; + + bi0 = to_next[0] = from[0]; + + from += 1; + to_next += 1; + n_left_from -= 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, bi0); + ip0 = vlib_buffer_get_current (p0); + h0 = ip6_next_header (ip0); + options_len0 = + clib_net_to_host_u16 (ip0->payload_length) - sizeof (h0[0]); + + error0 = ICMP6_ERROR_NONE; + sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX]; + ip6_sadd_link_local = + ip6_address_is_link_local_unicast (&ip0->src_address); + ip6_sadd_unspecified = + ip6_address_is_unspecified (&ip0->src_address); + + /* Check that source address is unspecified, link-local or else on-link. */ + if (!ip6_sadd_unspecified && !ip6_sadd_link_local) + { + u32 src_adj_index0 = ip6_src_lookup_for_packet (im, p0, ip0); + + if (ADJ_INDEX_INVALID != src_adj_index0) + { + ip_adjacency_t *adj0 = adj_get (src_adj_index0); + + /* Allow all realistic-looking rewrite adjacencies to pass */ + ni0 = adj0->lookup_next_index; + is_rewrite0 = (ni0 >= IP_LOOKUP_NEXT_ARP) && + (ni0 < IP6_LOOKUP_N_NEXT); + + error0 = ((adj0->rewrite_header.sw_if_index != sw_if_index0 + || !is_rewrite0) + ? + ICMP6_ERROR_NEIGHBOR_SOLICITATION_SOURCE_NOT_ON_LINK + : error0); + } + else + { + error0 = + ICMP6_ERROR_NEIGHBOR_SOLICITATION_SOURCE_NOT_ON_LINK; + } + } + + o0 = (void *) (h0 + 1); + o0 = ((options_len0 == 8 && o0->header.type == option_type + && o0->header.n_data_u64s == 1) ? o0 : 0); + + /* If src address unspecified or link local, donot learn neighbor MAC */ + if (PREDICT_TRUE (error0 == ICMP6_ERROR_NONE && o0 != 0 && + !ip6_sadd_unspecified)) + { + ip_neighbor_learn_t learn = { + .sw_if_index = sw_if_index0, + .type = IP46_TYPE_IP6, + .ip.ip6 = (is_solicitation ? + ip0->src_address : h0->target_address), + }; + memcpy (&learn.mac, o0->ethernet_address, sizeof (learn.mac)); + ip_neighbor_learn_dp (&learn); + } + + if (is_solicitation && error0 == ICMP6_ERROR_NONE) + { + /* Check that target address is local to this router. */ + fib_node_index_t fei; + u32 fib_index; + + fib_index = + ip6_fib_table_get_index_for_sw_if_index (sw_if_index0); + + if (~0 == fib_index) + { + error0 = ICMP6_ERROR_NEIGHBOR_SOLICITATION_SOURCE_UNKNOWN; + } + else + { + if (ip6_address_is_link_local_unicast (&h0->target_address)) + { + fei = ip6_fib_table_lookup_exact_match + (ip6_ll_fib_get (sw_if_index0), + &h0->target_address, 128); + } + else + { + fei = ip6_fib_table_lookup_exact_match (fib_index, + &h0->target_address, + 128); + } + + if (FIB_NODE_INDEX_INVALID == fei) + { + /* The target address is not in the FIB */ + error0 = + ICMP6_ERROR_NEIGHBOR_SOLICITATION_SOURCE_UNKNOWN; + } + else + { + if (FIB_ENTRY_FLAG_LOCAL & + fib_entry_get_flags_for_source (fei, + FIB_SOURCE_INTERFACE)) + { + /* It's an address that belongs to one of our interfaces + * that's good. */ + } + else + if (fib_entry_is_sourced + (fei, FIB_SOURCE_IP6_ND_PROXY) || + fib_entry_is_sourced (fei, FIB_SOURCE_IP6_ND)) + { + /* The address was added by IPv6 Proxy ND config. + * We should only respond to these if the NS arrived on + * the link that has a matching covering prefix */ + } + else + { + error0 = + ICMP6_ERROR_NEIGHBOR_SOLICITATION_SOURCE_UNKNOWN; + } + } + } + } + + if (is_solicitation) + next0 = (error0 != ICMP6_ERROR_NONE + ? ICMP6_NEIGHBOR_SOLICITATION_NEXT_DROP + : ICMP6_NEIGHBOR_SOLICITATION_NEXT_REPLY); + else + { + next0 = 0; + error0 = error0 == ICMP6_ERROR_NONE ? + ICMP6_ERROR_NEIGHBOR_ADVERTISEMENTS_RX : error0; + } + + if (is_solicitation && error0 == ICMP6_ERROR_NONE) + { + vnet_sw_interface_t *sw_if0; + ethernet_interface_t *eth_if0; + ethernet_header_t *eth0; + + /* dst address is either source address or the all-nodes mcast addr */ + if (!ip6_sadd_unspecified) + ip0->dst_address = ip0->src_address; + else + ip6_set_reserved_multicast_address (&ip0->dst_address, + IP6_MULTICAST_SCOPE_link_local, + IP6_MULTICAST_GROUP_ID_all_hosts); + + ip0->src_address = h0->target_address; + ip0->hop_limit = 255; + h0->icmp.type = ICMP6_neighbor_advertisement; + + sw_if0 = vnet_get_sup_sw_interface (vnm, sw_if_index0); + ASSERT (sw_if0->type == VNET_SW_INTERFACE_TYPE_HARDWARE); + eth_if0 = + ethernet_get_interface (ðernet_main, sw_if0->hw_if_index); + if (eth_if0 && o0) + { + clib_memcpy (o0->ethernet_address, eth_if0->address, 6); + o0->header.type = + ICMP6_NEIGHBOR_DISCOVERY_OPTION_target_link_layer_address; + } + + h0->advertisement_flags = clib_host_to_net_u32 + (ICMP6_NEIGHBOR_ADVERTISEMENT_FLAG_SOLICITED + | ICMP6_NEIGHBOR_ADVERTISEMENT_FLAG_OVERRIDE); + + h0->icmp.checksum = 0; + h0->icmp.checksum = + ip6_tcp_udp_icmp_compute_checksum (vm, p0, ip0, + &bogus_length); + ASSERT (bogus_length == 0); + + /* Reuse current MAC header, copy SMAC to DMAC and + * interface MAC to SMAC */ + vlib_buffer_advance (p0, -ethernet_buffer_header_size (p0)); + eth0 = vlib_buffer_get_current (p0); + clib_memcpy (eth0->dst_address, eth0->src_address, 6); + if (eth_if0) + clib_memcpy (eth0->src_address, eth_if0->address, 6); + + /* Setup input and output sw_if_index for packet */ + ASSERT (vnet_buffer (p0)->sw_if_index[VLIB_RX] == sw_if_index0); + vnet_buffer (p0)->sw_if_index[VLIB_TX] = sw_if_index0; + vnet_buffer (p0)->sw_if_index[VLIB_RX] = + vnet_main.local_interface_sw_if_index; + + n_advertisements_sent++; + } + + p0->error = error_node->errors[error0]; + + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, + bi0, next0); + } + + vlib_put_next_frame (vm, node, next_index, n_left_to_next); + } + + /* Account for advertisements sent. */ + vlib_error_count (vm, error_node->node_index, + ICMP6_ERROR_NEIGHBOR_ADVERTISEMENTS_TX, + n_advertisements_sent); + + return frame->n_vectors; +} + +static const ethernet_interface_t * +ip6_nd_get_eth_itf (u32 sw_if_index) +{ + const vnet_sw_interface_t *sw; + + /* lookup radv container - ethernet interfaces only */ + sw = vnet_get_sup_sw_interface (vnet_get_main (), sw_if_index); + if (sw->type == VNET_SW_INTERFACE_TYPE_HARDWARE) + return (ethernet_get_interface (ðernet_main, sw->hw_if_index)); + + return (NULL); +} + +/** + * @brief called when IP6 is enabled on a link. + * create and initialize router advertisement parameters with default + * values for this intfc + */ +static void +ip6_nd_link_enable (u32 sw_if_index) +{ + const ethernet_interface_t *eth; + ip6_nd_t *ind; + + eth = ip6_nd_get_eth_itf (sw_if_index); + + if (NULL == eth) + return; + + ASSERT (INDEX_INVALID == ip6_link_delegate_get (sw_if_index, + ip6_nd_delegate_id)); + + pool_get_zero (ip6_nd_pool, ind); + + ind->sw_if_index = sw_if_index; + + ip6_link_delegate_update (sw_if_index, ip6_nd_delegate_id, + ind - ip6_nd_pool); +} + +static void +ip6_nd_delegate_disable (index_t indi) +{ + ip6_nd_t *ind; + + ind = pool_elt_at_index (ip6_nd_pool, indi); + + pool_put (ip6_nd_pool, ind); +} + +static uword +icmp6_neighbor_solicitation (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + return icmp6_neighbor_solicitation_or_advertisement (vm, node, frame, + /* is_solicitation */ + 1); +} + +static uword +icmp6_neighbor_advertisement (vlib_main_t * vm, + vlib_node_runtime_t * node, + vlib_frame_t * frame) +{ + return icmp6_neighbor_solicitation_or_advertisement (vm, node, frame, + /* is_solicitation */ + 0); +} + +/* *INDENT-OFF* */ +VLIB_REGISTER_NODE (ip6_icmp_neighbor_solicitation_node,static) = +{ + .function = icmp6_neighbor_solicitation, + .name = "icmp6-neighbor-solicitation", + + .vector_size = sizeof (u32), + + .format_trace = format_icmp6_input_trace, + + .n_next_nodes = ICMP6_NEIGHBOR_SOLICITATION_N_NEXT, + .next_nodes = { + [ICMP6_NEIGHBOR_SOLICITATION_NEXT_DROP] = "ip6-drop", + [ICMP6_NEIGHBOR_SOLICITATION_NEXT_REPLY] = "interface-output", + }, +}; + +VLIB_REGISTER_NODE (ip6_icmp_neighbor_advertisement_node,static) = +{ + .function = icmp6_neighbor_advertisement, + .name = "icmp6-neighbor-advertisement", + + .vector_size = sizeof (u32), + + .format_trace = format_icmp6_input_trace, + + .n_next_nodes = 1, + .next_nodes = { + [0] = "ip6-drop", + }, +}; +/* *INDENT-ON* */ + +static u8 * +format_ip6_nd (u8 * s, va_list * args) +{ + CLIB_UNUSED (index_t indi) = va_arg (*args, index_t); + u32 indent = va_arg (*args, u32); + + s = format (s, "%UNeighbor Discovery: enabled\n", + format_white_space, indent); + + s = format (s, "%UICMP redirects are disabled\n", + format_white_space, indent + 2); + s = format (s, "%UICMP unreachables are not sent\n", + format_white_space, indent + 2); + s = format (s, "%UND DAD is disabled\n", format_white_space, indent + 2); + //s = format (s, "%UND reachable time is %d milliseconds\n",); + + return (s); +} + +/** + * VFT to act as an implementation of a neighbour protocol + */ +const static ip_neighbor_vft_t ip6_nd_impl_vft = { + .inv_proxy6_add = ip6_nd_proxy_add, + .inv_proxy6_del = ip6_nd_proxy_del, +}; + +/** + * VFT for registering as a delegate to an IP6 link + */ +const static ip6_link_delegate_vft_t ip6_nd_delegate_vft = { + .ildv_disable = ip6_nd_delegate_disable, + .ildv_enable = ip6_nd_link_enable, + .ildv_format = format_ip6_nd, +}; + +static clib_error_t * +ip6_nd_init (vlib_main_t * vm) +{ + icmp6_register_type (vm, ICMP6_neighbor_solicitation, + ip6_icmp_neighbor_solicitation_node.index); + icmp6_register_type (vm, ICMP6_neighbor_advertisement, + ip6_icmp_neighbor_advertisement_node.index); + + ip_neighbor_register (IP46_TYPE_IP6, &ip6_nd_impl_vft); + + ip6_nd_delegate_id = ip6_link_delegate_register (&ip6_nd_delegate_vft); + + return 0; +} + +/* *INDENT-OFF* */ +VLIB_INIT_FUNCTION (ip6_nd_init) = +{ + .runs_after = VLIB_INITS("icmp6_init"), +}; +/* *INDENT-ON* */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip6-nd/ip6_nd.h b/src/vnet/ip6-nd/ip6_nd.h new file mode 100644 index 00000000000..4dab7440b4a --- /dev/null +++ b/src/vnet/ip6-nd/ip6_nd.h @@ -0,0 +1,35 @@ +/* + * + * ip6_neighboor.h: ip6 neighbor structures + * + * Copyright (c) 2016 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef __IP6_ND_H__ +#define __IP6_ND_H__ + +#include <vnet/ip/ip6_packet.h> + +extern int ip6_nd_proxy_add (u32 sw_if_index, const ip6_address_t * addr); +extern int ip6_nd_proxy_del (u32 sw_if_index, const ip6_address_t * addr); + +#endif /* included_ip6_neighbor_h */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip6-nd/ip6_nd_api.c b/src/vnet/ip6-nd/ip6_nd_api.c new file mode 100644 index 00000000000..65b3ee381b3 --- /dev/null +++ b/src/vnet/ip6-nd/ip6_nd_api.c @@ -0,0 +1,382 @@ +/* + *------------------------------------------------------------------ + * ip_api.c - vnet ip api + * + * Copyright (c) 2016 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 <stddef.h> + +#include <vnet/ip6-nd/ip6_nd.h> +#include <vnet/ip6-nd/ip6_ra.h> + +#include <vnet/fib/fib_table.h> +#include <vnet/ip/ip_types_api.h> + +#include <vpp/app/version.h> + +#include <vlibapi/api.h> +#include <vlibmemory/api.h> + +/* define message IDs */ +#include <vnet/format_fns.h> +#include <vnet/ip6-nd/ip6_nd.api_enum.h> +#include <vnet/ip6-nd/ip6_nd.api_types.h> + +/** + * Base message ID fot the plugin + */ +static u32 ip6_nd_base_msg_id; +#define REPLY_MSG_ID_BASE ip6_nd_base_msg_id + +#include <vlibapi/api_helper_macros.h> + +static void +send_ip6nd_proxy_details (vl_api_registration_t * reg, + u32 context, + const ip46_address_t * addr, u32 sw_if_index) +{ + vl_api_ip6nd_proxy_details_t *mp; + + mp = vl_msg_api_alloc (sizeof (*mp)); + clib_memset (mp, 0, sizeof (*mp)); + mp->_vl_msg_id = ntohs (VL_API_IP6ND_PROXY_DETAILS); + mp->context = context; + mp->sw_if_index = htonl (sw_if_index); + + ip6_address_encode (&addr->ip6, mp->ip); + + vl_api_send_msg (reg, (u8 *) mp); +} + +typedef struct api_ip6nd_proxy_fib_table_walk_ctx_t_ +{ + u32 *indices; +} api_ip6nd_proxy_fib_table_walk_ctx_t; + +static fib_table_walk_rc_t +api_ip6nd_proxy_fib_table_walk (fib_node_index_t fei, void *arg) +{ + api_ip6nd_proxy_fib_table_walk_ctx_t *ctx = arg; + + if (fib_entry_is_sourced (fei, FIB_SOURCE_IP6_ND_PROXY)) + { + vec_add1 (ctx->indices, fei); + } + + return (FIB_TABLE_WALK_CONTINUE); +} + +static void +vl_api_ip6nd_proxy_dump_t_handler (vl_api_ip6nd_proxy_dump_t * mp) +{ + ip6_main_t *im6 = &ip6_main; + fib_table_t *fib_table; + api_ip6nd_proxy_fib_table_walk_ctx_t ctx = { + .indices = NULL, + }; + fib_node_index_t *feip; + const fib_prefix_t *pfx; + vl_api_registration_t *reg; + + reg = vl_api_client_index_to_registration (mp->client_index); + if (!reg) + return; + + /* *INDENT-OFF* */ + pool_foreach (fib_table, im6->fibs, + ({ + fib_table_walk(fib_table->ft_index, + FIB_PROTOCOL_IP6, + api_ip6nd_proxy_fib_table_walk, + &ctx); + })); + /* *INDENT-ON* */ + + vec_sort_with_function (ctx.indices, fib_entry_cmp_for_sort); + + vec_foreach (feip, ctx.indices) + { + pfx = fib_entry_get_prefix (*feip); + + send_ip6nd_proxy_details (reg, + mp->context, + &pfx->fp_addr, + fib_entry_get_resolving_interface (*feip)); + } + + vec_free (ctx.indices); +} + +static void +vl_api_ip6nd_proxy_add_del_t_handler (vl_api_ip6nd_proxy_add_del_t * mp) +{ + vl_api_ip6nd_proxy_add_del_reply_t *rmp; + ip6_address_t ip6; + int rv = 0; + + VALIDATE_SW_IF_INDEX (mp); + + ip6_address_decode (mp->ip, &ip6); + if (mp->is_add) + rv = ip6_nd_proxy_add (ntohl (mp->sw_if_index), &ip6); + else + rv = ip6_nd_proxy_del (ntohl (mp->sw_if_index), &ip6); + + BAD_SW_IF_INDEX_LABEL; + REPLY_MACRO (VL_API_IP6ND_PROXY_ADD_DEL_REPLY); +} + +static void + vl_api_sw_interface_ip6nd_ra_config_t_handler + (vl_api_sw_interface_ip6nd_ra_config_t * mp) +{ + vl_api_sw_interface_ip6nd_ra_config_reply_t *rmp; + vlib_main_t *vm = vlib_get_main (); + int rv = 0; + u8 is_no, suppress, managed, other, ll_option, send_unicast, cease, + default_router; + + is_no = mp->is_no == 1; + suppress = mp->suppress == 1; + managed = mp->managed == 1; + other = mp->other == 1; + ll_option = mp->ll_option == 1; + send_unicast = mp->send_unicast == 1; + cease = mp->cease == 1; + default_router = mp->default_router == 1; + + VALIDATE_SW_IF_INDEX (mp); + + rv = ip6_ra_config (vm, ntohl (mp->sw_if_index), + suppress, managed, other, + ll_option, send_unicast, cease, + default_router, ntohl (mp->lifetime), + ntohl (mp->initial_count), + ntohl (mp->initial_interval), + ntohl (mp->max_interval), + ntohl (mp->min_interval), is_no); + + BAD_SW_IF_INDEX_LABEL; + + REPLY_MACRO (VL_API_SW_INTERFACE_IP6ND_RA_CONFIG_REPLY); +} + +static void + vl_api_sw_interface_ip6nd_ra_prefix_t_handler + (vl_api_sw_interface_ip6nd_ra_prefix_t * mp) +{ + vlib_main_t *vm = vlib_get_main (); + vl_api_sw_interface_ip6nd_ra_prefix_reply_t *rmp; + fib_prefix_t pfx; + int rv = 0; + u8 is_no, use_default, no_advertise, off_link, no_autoconfig, no_onlink; + + VALIDATE_SW_IF_INDEX (mp); + + ip_prefix_decode (&mp->prefix, &pfx); + is_no = mp->is_no == 1; + use_default = mp->use_default == 1; + no_advertise = mp->no_advertise == 1; + off_link = mp->off_link == 1; + no_autoconfig = mp->no_autoconfig == 1; + no_onlink = mp->no_onlink == 1; + + rv = ip6_ra_prefix (vm, ntohl (mp->sw_if_index), + &pfx.fp_addr.ip6, + pfx.fp_len, use_default, + ntohl (mp->val_lifetime), + ntohl (mp->pref_lifetime), no_advertise, + off_link, no_autoconfig, no_onlink, is_no); + + BAD_SW_IF_INDEX_LABEL; + REPLY_MACRO (VL_API_SW_INTERFACE_IP6ND_RA_PREFIX_REPLY); +} + +static void + vl_api_ip6nd_send_router_solicitation_t_handler + (vl_api_ip6nd_send_router_solicitation_t * mp) +{ + vl_api_ip6nd_send_router_solicitation_reply_t *rmp; + icmp6_send_router_solicitation_params_t params; + vlib_main_t *vm = vlib_get_main (); + int rv = 0; + + VALIDATE_SW_IF_INDEX (mp); + + BAD_SW_IF_INDEX_LABEL; + REPLY_MACRO (VL_API_IP6ND_SEND_ROUTER_SOLICITATION_REPLY); + + if (rv != 0) + return; + + params.irt = ntohl (mp->irt); + params.mrt = ntohl (mp->mrt); + params.mrc = ntohl (mp->mrc); + params.mrd = ntohl (mp->mrd); + + icmp6_send_router_solicitation (vm, ntohl (mp->sw_if_index), mp->stop, + ¶ms); +} + +static void +ip6_ra_handle_report (const ip6_ra_report_t * rap) +{ + /* *INDENT-OFF* */ + vpe_client_registration_t *rp; + + pool_foreach(rp, vpe_api_main.ip6_ra_events_registrations, + ({ + vl_api_registration_t *vl_reg; + + vl_reg = vl_api_client_index_to_registration (rp->client_index); + + if (vl_reg && vl_api_can_send_msg (vl_reg)) + { + vl_api_ip6_ra_prefix_info_t *prefix; + vl_api_ip6_ra_event_t *event; + + u32 event_size = (sizeof (vl_api_ip6_ra_event_t) + + vec_len (rap->prefixes) * + sizeof (vl_api_ip6_ra_prefix_info_t)); + event = vl_msg_api_alloc_zero (event_size); + + event->_vl_msg_id = htons (VL_API_IP6_RA_EVENT + REPLY_MSG_ID_BASE); + event->client_index = rp->client_index; + event->pid = rp->client_pid; + event->sw_if_index = clib_host_to_net_u32 (rap->sw_if_index); + + ip6_address_encode (&rap->router_address, + event->router_addr); + + event->current_hop_limit = rap->current_hop_limit; + event->flags = rap->flags; + event->router_lifetime_in_sec = + clib_host_to_net_u16 (rap->router_lifetime_in_sec); + event->neighbor_reachable_time_in_msec = + clib_host_to_net_u32 (rap->neighbor_reachable_time_in_msec); + event->time_in_msec_between_retransmitted_neighbor_solicitations = + clib_host_to_net_u32 (rap->time_in_msec_between_retransmitted_neighbor_solicitations); + event->n_prefixes = clib_host_to_net_u32 (vec_len (rap->prefixes)); + + prefix = event->prefixes; + // (typeof (prefix)) event->prefixes; + u32 j; + for (j = 0; j < vec_len (rap->prefixes); j++) + { + ra_report_prefix_info_t *info = &rap->prefixes[j]; + ip_prefix_encode(&info->prefix, &prefix->prefix); + prefix->flags = info->flags; + prefix->valid_time = clib_host_to_net_u32 (info->valid_time); + prefix->preferred_time = + clib_host_to_net_u32 (info->preferred_time); + prefix++; + } + + vl_api_send_msg (vl_reg, (u8 *) event); + } + })); + /* *INDENT-ON* */ +} + +static void +vl_api_want_ip6_ra_events_t_handler (vl_api_want_ip6_ra_events_t * mp) +{ + vpe_api_main_t *am = &vpe_api_main; + vl_api_want_ip6_ra_events_reply_t *rmp; + int rv = 0, had_reg, have_reg; + + had_reg = hash_elts (am->ip6_ra_events_registration_hash); + uword *p = hash_get (am->ip6_ra_events_registration_hash, mp->client_index); + vpe_client_registration_t *rp; + if (p) + { + if (mp->enable) + { + clib_warning ("pid %d: already enabled...", ntohl (mp->pid)); + rv = VNET_API_ERROR_INVALID_REGISTRATION; + goto reply; + } + else + { + rp = pool_elt_at_index (am->ip6_ra_events_registrations, p[0]); + pool_put (am->ip6_ra_events_registrations, rp); + hash_unset (am->ip6_ra_events_registration_hash, mp->client_index); + goto reply; + } + } + if (mp->enable == 0) + { + clib_warning ("pid %d: already disabled...", ntohl (mp->pid)); + rv = VNET_API_ERROR_INVALID_REGISTRATION; + goto reply; + } + pool_get (am->ip6_ra_events_registrations, rp); + rp->client_index = mp->client_index; + rp->client_pid = ntohl (mp->pid); + hash_set (am->ip6_ra_events_registration_hash, rp->client_index, + rp - am->ip6_ra_events_registrations); + +reply: + have_reg = hash_elts (am->ip6_ra_events_registration_hash); + + if (!had_reg && have_reg) + ip6_ra_report_register (ip6_ra_handle_report); + else if (had_reg && !have_reg) + ip6_ra_report_unregister (ip6_ra_handle_report); + + REPLY_MACRO (VL_API_WANT_IP6_RA_EVENTS_REPLY); +} + +static clib_error_t * +want_ip6_ra_events_reaper (u32 client_index) +{ + vpe_api_main_t *am = &vpe_api_main; + vpe_client_registration_t *rp; + uword *p; + + p = hash_get (am->ip6_ra_events_registration_hash, client_index); + + if (p) + { + rp = pool_elt_at_index (am->ip6_ra_events_registrations, p[0]); + pool_put (am->ip6_ra_events_registrations, rp); + hash_unset (am->ip6_ra_events_registration_hash, client_index); + } + return (NULL); +} + +VL_MSG_API_REAPER_FUNCTION (want_ip6_ra_events_reaper); + +#include <vnet/ip6-nd/ip6_nd.api.c> + +static clib_error_t * +ip6_nd_api_init (vlib_main_t * vm) +{ + /* Ask for a correctly-sized block of API message decode slots */ + ip6_nd_base_msg_id = setup_message_id_table (); + + return 0; +} + +VLIB_INIT_FUNCTION (ip6_nd_api_init); + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip6-nd/ip6_nd_proxy.c b/src/vnet/ip6-nd/ip6_nd_proxy.c new file mode 100644 index 00000000000..110847fff84 --- /dev/null +++ b/src/vnet/ip6-nd/ip6_nd_proxy.c @@ -0,0 +1,127 @@ +/* + * ip/ip6_neighbor.c: IP6 neighbor handling + * + * Copyright (c) 2010 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/ip6-nd/ip6_nd.h> +#include <vnet/ip-neighbor/ip_neighbor.h> + +#include <vnet/fib/ip6_fib.h> + +static int +ip6_nd_proxy_add_del (u32 sw_if_index, const ip6_address_t * addr, u8 is_del) +{ + /* *INDENT-OFF* */ + u32 fib_index; + fib_prefix_t pfx = { + .fp_len = 128, + .fp_proto = FIB_PROTOCOL_IP6, + .fp_addr = { + .ip6 = *addr, + }, + }; + ip46_address_t nh = { + .ip6 = *addr, + }; + /* *INDENT-ON* */ + + fib_index = ip6_fib_table_get_index_for_sw_if_index (sw_if_index); + + if (~0 == fib_index) + return VNET_API_ERROR_NO_SUCH_FIB; + + if (is_del) + { + fib_table_entry_path_remove (fib_index, + &pfx, + FIB_SOURCE_IP6_ND_PROXY, + DPO_PROTO_IP6, + &nh, + sw_if_index, + ~0, 1, FIB_ROUTE_PATH_FLAG_NONE); + /* flush the ND cache of this address if it's there */ + ip_neighbor_del (&nh, IP46_TYPE_IP6, sw_if_index); + } + else + { + fib_table_entry_path_add (fib_index, + &pfx, + FIB_SOURCE_IP6_ND_PROXY, + FIB_ENTRY_FLAG_NONE, + DPO_PROTO_IP6, + &nh, + sw_if_index, + ~0, 1, NULL, FIB_ROUTE_PATH_FLAG_NONE); + } + return (0); +} + +int +ip6_nd_proxy_add (u32 sw_if_index, const ip6_address_t * addr) +{ + return (ip6_nd_proxy_add_del (sw_if_index, addr, 0)); +} + +int +ip6_nd_proxy_del (u32 sw_if_index, const ip6_address_t * addr) +{ + return (ip6_nd_proxy_add_del (sw_if_index, addr, 1)); +} + +static clib_error_t * +set_ip6_nd_proxy_cmd (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) +{ + vnet_main_t *vnm = vnet_get_main (); + clib_error_t *error = 0; + ip6_address_t addr; + u32 sw_if_index; + u8 is_del = 0; + + if (unformat_user (input, unformat_vnet_sw_interface, vnm, &sw_if_index)) + { + /* get the rest of the command */ + while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (input, "%U", unformat_ip6_address, &addr)) + break; + else if (unformat (input, "delete") || unformat (input, "del")) + is_del = 1; + else + return (unformat_parse_error (input)); + } + } + + ip6_nd_proxy_add_del (sw_if_index, &addr, is_del); + + return error; +} + +/* *INDENT-OFF* */ +VLIB_CLI_COMMAND (set_ip6_nd_proxy_command, static) = +{ + .path = "set ip6 nd proxy", + .short_help = "set ip6 nd proxy <HOST> <INTERFACE>", + .function = set_ip6_nd_proxy_cmd, +}; +/* *INDENT-ON* */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip6-nd/ip6_nd_test.c b/src/vnet/ip6-nd/ip6_nd_test.c new file mode 100644 index 00000000000..5ca37029a76 --- /dev/null +++ b/src/vnet/ip6-nd/ip6_nd_test.c @@ -0,0 +1,330 @@ +/* + * 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 <vat/vat.h> +#include <vlibapi/api.h> +#include <vlibmemory/api.h> +#include <vppinfra/error.h> +#include <vpp/api/types.h> + +#include <vnet/ip/ip_format_fns.h> +#include <vnet/ethernet/ethernet_format_fns.h> + +/* define message IDs */ +#include <ip6-nd/ip6_nd.api_enum.h> +#include <ip6-nd/ip6_nd.api_types.h> +#include <vpp/api/vpe.api_types.h> + +typedef struct +{ + /* API message ID base */ + u16 msg_id_base; + u32 ping_id; + vat_main_t *vat_main; +} ip6_nd_test_main_t; + +ip6_nd_test_main_t ip6_nd_test_main; + +#define __plugin_msg_base ip6_nd_test_main.msg_id_base +#include <vlibapi/vat_helper_macros.h> + +static int +api_want_ip6_ra_events (vat_main_t * vam) +{ + return -1; +} + +static int +api_ip6nd_send_router_solicitation (vat_main_t * vam) +{ + return -1; +} + +static int +api_ip6nd_proxy_add_del (vat_main_t * vam) +{ + unformat_input_t *i = vam->input; + vl_api_ip6nd_proxy_add_del_t *mp; + u32 sw_if_index = ~0; + u8 v6_address_set = 0; + vl_api_ip6_address_t v6address; + u8 is_add = 1; + int ret; + + /* Parse args required to build the message */ + while (unformat_check_input (i) != UNFORMAT_END_OF_INPUT) + { + if (unformat (i, "%U", unformat_sw_if_index, vam, &sw_if_index)) + ; + else if (unformat (i, "sw_if_index %d", &sw_if_index)) + ; + else if (unformat (i, "%U", unformat_vl_api_ip6_address, &v6address)) + v6_address_set = 1; + if (unformat (i, "del")) + is_add = 0; + else + { + clib_warning ("parse error '%U'", format_unformat_error, i); + return -99; + } + } + + if (sw_if_index == ~0) + { + errmsg ("missing interface name or sw_if_index"); + return -99; + } + if (!v6_address_set) + { + errmsg ("no address set"); + return -99; + } + + /* Construct the API message */ + M (IP6ND_PROXY_ADD_DEL, mp); + + mp->is_add = is_add; + mp->sw_if_index = ntohl (sw_if_index); + clib_memcpy (mp->ip, v6address, sizeof (v6address)); + + /* send it... */ + S (mp); + + /* Wait for a reply, return good/bad news */ + W (ret); + return ret; +} + +static int +api_ip6nd_proxy_dump (vat_main_t * vam) +{ + vl_api_ip6nd_proxy_dump_t *mp; + vl_api_control_ping_t *mp_ping; + int ret; + + M (IP6ND_PROXY_DUMP, mp); + + S (mp); + + /* Use a control ping for synchronization */ + /* Use a control ping for synchronization */ + mp_ping = vl_msg_api_alloc_as_if_client (sizeof (*mp_ping)); + mp_ping->_vl_msg_id = htons (ip6_nd_test_main.ping_id); + mp_ping->client_index = vam->my_client_index; + vam->result_ready = 0; + + S (mp_ping); + + W (ret); + return ret; +} + +static void vl_api_ip6nd_proxy_details_t_handler + (vl_api_ip6nd_proxy_details_t * mp) +{ + vat_main_t *vam = &vat_main; + + print (vam->ofp, "host %U sw_if_index %d", + format_vl_api_ip6_address, mp->ip, ntohl (mp->sw_if_index)); +} + +static int +api_sw_interface_ip6nd_ra_prefix (vat_main_t * vam) +{ + unformat_input_t *i = vam->input; + vl_api_sw_interface_ip6nd_ra_prefix_t *mp; + u32 sw_if_index; + u8 sw_if_index_set = 0; + u8 v6_address_set = 0; + vl_api_prefix_t pfx; + u8 use_default = 0; + u8 no_advertise = 0; + u8 off_link = 0; + u8 no_autoconfig = 0; + u8 no_onlink = 0; + u8 is_no = 0; + u32 val_lifetime = 0; + u32 pref_lifetime = 0; + int ret; + + /* Parse args required to build the message */ + while (unformat_check_input (i) != UNFORMAT_END_OF_INPUT) + { + if (unformat (i, "%U", unformat_sw_if_index, vam, &sw_if_index)) + sw_if_index_set = 1; + else if (unformat (i, "sw_if_index %d", &sw_if_index)) + sw_if_index_set = 1; + else if (unformat (i, "%U", unformat_vl_api_prefix, &pfx)) + v6_address_set = 1; + else if (unformat (i, "val_life %d", &val_lifetime)) + ; + else if (unformat (i, "pref_life %d", &pref_lifetime)) + ; + else if (unformat (i, "def")) + use_default = 1; + else if (unformat (i, "noadv")) + no_advertise = 1; + else if (unformat (i, "offl")) + off_link = 1; + else if (unformat (i, "noauto")) + no_autoconfig = 1; + else if (unformat (i, "nolink")) + no_onlink = 1; + else if (unformat (i, "isno")) + is_no = 1; + else + { + clib_warning ("parse error '%U'", format_unformat_error, i); + return -99; + } + } + + if (sw_if_index_set == 0) + { + errmsg ("missing interface name or sw_if_index"); + return -99; + } + if (!v6_address_set) + { + errmsg ("no address set"); + return -99; + } + + /* Construct the API message */ + M (SW_INTERFACE_IP6ND_RA_PREFIX, mp); + + mp->sw_if_index = ntohl (sw_if_index); + clib_memcpy (&mp->prefix, &pfx, sizeof (pfx)); + mp->use_default = use_default; + mp->no_advertise = no_advertise; + mp->off_link = off_link; + mp->no_autoconfig = no_autoconfig; + mp->no_onlink = no_onlink; + mp->is_no = is_no; + mp->val_lifetime = ntohl (val_lifetime); + mp->pref_lifetime = ntohl (pref_lifetime); + + /* send it... */ + S (mp); + + /* Wait for a reply, return good/bad news */ + W (ret); + return ret; +} + +static int +api_sw_interface_ip6nd_ra_config (vat_main_t * vam) +{ + unformat_input_t *i = vam->input; + vl_api_sw_interface_ip6nd_ra_config_t *mp; + u32 sw_if_index; + u8 sw_if_index_set = 0; + u8 suppress = 0; + u8 managed = 0; + u8 other = 0; + u8 ll_option = 0; + u8 send_unicast = 0; + u8 cease = 0; + u8 is_no = 0; + u8 default_router = 0; + u32 max_interval = 0; + u32 min_interval = 0; + u32 lifetime = 0; + u32 initial_count = 0; + u32 initial_interval = 0; + int ret; + + + /* Parse args required to build the message */ + while (unformat_check_input (i) != UNFORMAT_END_OF_INPUT) + { + if (unformat (i, "%U", unformat_sw_if_index, vam, &sw_if_index)) + sw_if_index_set = 1; + else if (unformat (i, "sw_if_index %d", &sw_if_index)) + sw_if_index_set = 1; + else if (unformat (i, "maxint %d", &max_interval)) + ; + else if (unformat (i, "minint %d", &min_interval)) + ; + else if (unformat (i, "life %d", &lifetime)) + ; + else if (unformat (i, "count %d", &initial_count)) + ; + else if (unformat (i, "interval %d", &initial_interval)) + ; + else if (unformat (i, "suppress") || unformat (i, "surpress")) + suppress = 1; + else if (unformat (i, "managed")) + managed = 1; + else if (unformat (i, "other")) + other = 1; + else if (unformat (i, "ll")) + ll_option = 1; + else if (unformat (i, "send")) + send_unicast = 1; + else if (unformat (i, "cease")) + cease = 1; + else if (unformat (i, "isno")) + is_no = 1; + else if (unformat (i, "def")) + default_router = 1; + else + { + clib_warning ("parse error '%U'", format_unformat_error, i); + return -99; + } + } + + if (sw_if_index_set == 0) + { + errmsg ("missing interface name or sw_if_index"); + return -99; + } + + /* Construct the API message */ + M (SW_INTERFACE_IP6ND_RA_CONFIG, mp); + + mp->sw_if_index = ntohl (sw_if_index); + mp->max_interval = ntohl (max_interval); + mp->min_interval = ntohl (min_interval); + mp->lifetime = ntohl (lifetime); + mp->initial_count = ntohl (initial_count); + mp->initial_interval = ntohl (initial_interval); + mp->suppress = suppress; + mp->managed = managed; + mp->other = other; + mp->ll_option = ll_option; + mp->send_unicast = send_unicast; + mp->cease = cease; + mp->is_no = is_no; + mp->default_router = default_router; + + /* send it... */ + S (mp); + + /* Wait for a reply, return good/bad news */ + W (ret); + return ret; +} + +#include <ip6-nd/ip6_nd.api_test.c> + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip6-nd/ip6_ra.c b/src/vnet/ip6-nd/ip6_ra.c new file mode 100644 index 00000000000..ebc2c4be417 --- /dev/null +++ b/src/vnet/ip6-nd/ip6_ra.c @@ -0,0 +1,2261 @@ +/* + * ip/ip6_neighbor.c: IP6 neighbor handling + * + * Copyright (c) 2010 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/ip6-nd/ip6_ra.h> + +#include <vnet/ip/ip.h> +#include <vnet/ip-neighbor/ip_neighbor_dp.h> + +#include <vnet/fib/ip6_fib.h> +#include <vnet/ip/ip6_link.h> + +/** + * @file + * @brief IPv6 Router Advertisements. + * + * The files contains the API and CLI code for managing IPv6 RAs + */ + +/* *INDENT-OFF* */ +/* Router solicitation packet format for ethernet. */ +typedef CLIB_PACKED (struct +{ + ip6_header_t ip; + icmp6_neighbor_discovery_header_t neighbor; + icmp6_neighbor_discovery_ethernet_link_layer_address_option_t + link_layer_option; +}) icmp6_router_solicitation_header_t; + +/* router advertisement packet format for ethernet. */ +typedef CLIB_PACKED (struct +{ + ip6_header_t ip; + icmp6_router_advertisement_header_t router; + icmp6_neighbor_discovery_ethernet_link_layer_address_option_t + link_layer_option; + icmp6_neighbor_discovery_mtu_option_t mtu_option; + icmp6_neighbor_discovery_prefix_information_option_t + prefix[0]; +}) icmp6_router_advertisement_packet_t; +/* *INDENT-ON* */ + +#define DEF_MAX_RADV_INTERVAL 200 +#define DEF_MIN_RADV_INTERVAL .75 * DEF_MAX_RADV_INTERVAL +#define DEF_CURR_HOP_LIMIT 64 +#define DEF_DEF_RTR_LIFETIME 3 * DEF_MAX_RADV_INTERVAL +#define MAX_DEF_RTR_LIFETIME 9000 + +#define MAX_INITIAL_RTR_ADVERT_INTERVAL 16 /* seconds */ +#define MAX_INITIAL_RTR_ADVERTISEMENTS 3 /*transmissions */ +#define MIN_DELAY_BETWEEN_RAS 3 /* seconds */ +#define MAX_DELAY_BETWEEN_RAS 1800 /* seconds */ +#define MAX_RA_DELAY_TIME .5 /* seconds */ + +/* advertised prefix option */ +typedef struct +{ + /* basic advertised information */ + ip6_address_t prefix; + u8 prefix_len; + int adv_on_link_flag; + int adv_autonomous_flag; + u32 adv_valid_lifetime_in_secs; + u32 adv_pref_lifetime_in_secs; + + /* advertised values are computed from these times if decrementing */ + f64 valid_lifetime_expires; + f64 pref_lifetime_expires; + + /* local information */ + int enabled; + int deprecated_prefix_flag; + int decrement_lifetime_flag; + +#define MIN_ADV_VALID_LIFETIME 7203 /* seconds */ +#define DEF_ADV_VALID_LIFETIME 2592000 +#define DEF_ADV_PREF_LIFETIME 604800 + + /* extensions are added here, mobile, DNS etc.. */ +} ip6_radv_prefix_t; + +typedef struct ip6_ra_t_ +{ + /* advertised config information, zero means unspecified */ + u8 curr_hop_limit; + int adv_managed_flag; + int adv_other_flag; + u16 adv_router_lifetime_in_sec; + u32 adv_neighbor_reachable_time_in_msec; + u32 adv_time_in_msec_between_retransmitted_neighbor_solicitations; + + /* mtu option */ + u32 adv_link_mtu; + + /* local information */ + u32 sw_if_index; + int send_radv; /* radv on/off on this interface - set by config */ + int cease_radv; /* we are ceasing to send - set byf config */ + int send_unicast; + int adv_link_layer_address; + int prefix_option; + int failed_device_check; + int ref_count; + + /* prefix option */ + ip6_radv_prefix_t *adv_prefixes_pool; + + /* Hash table mapping address to index in interface advertised prefix pool. */ + mhash_t address_to_prefix_index; + + f64 max_radv_interval; + f64 min_radv_interval; + f64 min_delay_between_radv; + f64 max_delay_between_radv; + f64 max_rtr_default_lifetime; + + f64 last_radv_time; + f64 last_multicast_time; + f64 next_multicast_time; + + + u32 initial_adverts_count; + f64 initial_adverts_interval; + u32 initial_adverts_sent; + + /* stats */ + u32 n_advertisements_sent; + u32 n_solicitations_rcvd; + u32 n_solicitations_dropped; + + /* router solicitations sending state */ + u8 keep_sending_rs; /* when true then next fields are valid */ + icmp6_send_router_solicitation_params_t params; + f64 sleep_interval; + f64 due_time; + u32 n_left; + f64 start_time; + vlib_buffer_t *buffer; + + u32 seed; + +} ip6_ra_t; + +static ip6_link_delegate_id_t ip6_ra_delegate_id; +static ip6_ra_t *ip6_ra_pool; + + +/* vector of registered RA report listeners */ +static ip6_ra_report_notify_t *ip6_ra_listeners; + +static int ip6_ra_publish (ip6_ra_report_t * r); + +void +ip6_ra_report_register (ip6_ra_report_notify_t fn) +{ + ip6_ra_report_notify_t *listener; + vec_foreach (listener, ip6_ra_listeners) + { + if (*listener == fn) + return; + } + + vec_add1 (ip6_ra_listeners, fn); +} + +void +ip6_ra_report_unregister (ip6_ra_report_notify_t fn) +{ + u32 ii; + + vec_foreach_index (ii, ip6_ra_listeners) + { + if (ip6_ra_listeners[ii] == fn) + { + vec_del1 (ip6_ra_listeners, ii); + break; + } + } +} + +static inline ip6_ra_t * +ip6_ra_get_itf (u32 sw_if_index) +{ + index_t rai; + + rai = ip6_link_delegate_get (sw_if_index, ip6_ra_delegate_id); + + if (INDEX_INVALID != rai) + return (pool_elt_at_index (ip6_ra_pool, rai)); + + return (NULL); +} + +/* for "syslogging" - use elog for now */ +#define foreach_log_level \ + _ (DEBUG, "DEBUG") \ + _ (INFO, "INFORMATION") \ + _ (NOTICE, "NOTICE") \ + _ (WARNING, "WARNING") \ + _ (ERR, "ERROR") \ + _ (CRIT, "CRITICAL") \ + _ (ALERT, "ALERT") \ + _ (EMERG, "EMERGENCY") + +typedef enum +{ +#define _(f,s) LOG_##f, + foreach_log_level +#undef _ +} log_level_t; + +static char *log_level_strings[] = { +#define _(f,s) s, + foreach_log_level +#undef _ +}; + +static int logmask = 1 << LOG_DEBUG; + +static void +ip6_neighbor_syslog (vlib_main_t * vm, int priority, char *fmt, ...) +{ + /* just use elog for now */ + u8 *what; + va_list va; + + if ((priority > LOG_EMERG) || !(logmask & (1 << priority))) + return; + + va_start (va, fmt); + if (fmt) + { + what = va_format (0, fmt, &va); + + ELOG_TYPE_DECLARE (e) = + { + .format = "ip6 nd: (%s): %s",.format_args = "T4T4",}; + struct + { + u32 s[2]; + } *ed; + ed = ELOG_DATA (&vm->elog_main, e); + ed->s[0] = elog_string (&vm->elog_main, log_level_strings[priority]); + ed->s[1] = elog_string (&vm->elog_main, (char *) what); + } + va_end (va); + return; +} + +/* ipv6 neighbor discovery - router advertisements */ +typedef enum +{ + ICMP6_ROUTER_SOLICITATION_NEXT_DROP, + ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_RW, + ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_TX, + ICMP6_ROUTER_SOLICITATION_N_NEXT, +} icmp6_router_solicitation_or_advertisement_next_t; + +static_always_inline uword +icmp6_router_solicitation (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + vnet_main_t *vnm = vnet_get_main (); + ip6_main_t *im = &ip6_main; + uword n_packets = frame->n_vectors; + u32 *from, *to_next; + u32 n_left_from, n_left_to_next, next_index; + u32 n_advertisements_sent = 0; + int bogus_length; + + icmp6_neighbor_discovery_option_type_t option_type; + + vlib_node_runtime_t *error_node = + vlib_node_get_runtime (vm, ip6_icmp_input_node.index); + + from = vlib_frame_vector_args (frame); + n_left_from = n_packets; + next_index = node->cached_next_index; + + if (node->flags & VLIB_NODE_FLAG_TRACE) + vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors, + /* stride */ 1, + sizeof (icmp6_input_trace_t)); + + /* source may append his LL address */ + option_type = ICMP6_NEIGHBOR_DISCOVERY_OPTION_source_link_layer_address; + + 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) + { + vlib_buffer_t *p0; + ip6_header_t *ip0; + ip6_ra_t *radv_info = NULL; + + icmp6_neighbor_discovery_header_t *h0; + icmp6_neighbor_discovery_ethernet_link_layer_address_option_t *o0; + + u32 bi0, options_len0, sw_if_index0, next0, error0; + u32 is_solicitation = 1, is_dropped = 0; + u32 is_unspecified, is_link_local; + + bi0 = to_next[0] = from[0]; + + from += 1; + to_next += 1; + n_left_from -= 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, bi0); + ip0 = vlib_buffer_get_current (p0); + h0 = ip6_next_header (ip0); + options_len0 = + clib_net_to_host_u16 (ip0->payload_length) - sizeof (h0[0]); + is_unspecified = ip6_address_is_unspecified (&ip0->src_address); + is_link_local = + ip6_address_is_link_local_unicast (&ip0->src_address); + + error0 = ICMP6_ERROR_NONE; + sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX]; + + /* check if solicitation (not from nd_timer node) */ + if (ip6_address_is_unspecified (&ip0->dst_address)) + is_solicitation = 0; + + /* Check that source address is unspecified, link-local or else on-link. */ + if (!is_unspecified && !is_link_local) + { + u32 src_adj_index0 = ip6_src_lookup_for_packet (im, p0, ip0); + + if (ADJ_INDEX_INVALID != src_adj_index0) + { + ip_adjacency_t *adj0 = adj_get (src_adj_index0); + + error0 = (adj0->rewrite_header.sw_if_index != sw_if_index0 + ? + ICMP6_ERROR_ROUTER_SOLICITATION_SOURCE_NOT_ON_LINK + : error0); + } + else + { + error0 = ICMP6_ERROR_ROUTER_SOLICITATION_SOURCE_NOT_ON_LINK; + } + } + + /* check for source LL option and process */ + o0 = (void *) (h0 + 1); + o0 = ((options_len0 == 8 + && o0->header.type == option_type + && o0->header.n_data_u64s == 1) ? o0 : 0); + + /* if src address unspecified IGNORE any options */ + if (PREDICT_TRUE (error0 == ICMP6_ERROR_NONE && o0 != 0 && + !is_unspecified && !is_link_local)) + { + ip_neighbor_learn_t learn = { + .type = IP46_TYPE_IP6, + .sw_if_index = sw_if_index0, + .ip.ip6 = ip0->src_address, + }; + memcpy (&learn.mac, o0->ethernet_address, sizeof (learn.mac)); + ip_neighbor_learn_dp (&learn); + } + + /* default is to drop */ + next0 = ICMP6_ROUTER_SOLICITATION_NEXT_DROP; + + if (error0 == ICMP6_ERROR_NONE) + { + vnet_sw_interface_t *sw_if0; + ethernet_interface_t *eth_if0; + u32 adj_index0; + + sw_if0 = vnet_get_sup_sw_interface (vnm, sw_if_index0); + ASSERT (sw_if0->type == VNET_SW_INTERFACE_TYPE_HARDWARE); + eth_if0 = + ethernet_get_interface (ðernet_main, sw_if0->hw_if_index); + + /* only support ethernet interface type for now */ + error0 = + (!eth_if0) ? ICMP6_ERROR_ROUTER_SOLICITATION_UNSUPPORTED_INTF + : error0; + + if (error0 == ICMP6_ERROR_NONE) + { + /* adjust the sizeof the buffer to just include the ipv6 header */ + p0->current_length -= + (options_len0 + + sizeof (icmp6_neighbor_discovery_header_t)); + + radv_info = ip6_ra_get_itf (sw_if_index0); + + error0 = ((!radv_info) ? + ICMP6_ERROR_ROUTER_SOLICITATION_RADV_NOT_CONFIG : + error0); + + if (error0 == ICMP6_ERROR_NONE) + { + f64 now = vlib_time_now (vm); + + /* for solicited adverts - need to rate limit */ + if (is_solicitation) + { + if (0 != radv_info->last_radv_time && + (now - radv_info->last_radv_time) < + MIN_DELAY_BETWEEN_RAS) + is_dropped = 1; + else + radv_info->last_radv_time = now; + } + + /* send now */ + icmp6_router_advertisement_header_t rh; + + rh.icmp.type = ICMP6_router_advertisement; + rh.icmp.code = 0; + rh.icmp.checksum = 0; + + rh.current_hop_limit = radv_info->curr_hop_limit; + rh.router_lifetime_in_sec = + clib_host_to_net_u16 + (radv_info->adv_router_lifetime_in_sec); + rh. + time_in_msec_between_retransmitted_neighbor_solicitations + = + clib_host_to_net_u32 (radv_info-> + adv_time_in_msec_between_retransmitted_neighbor_solicitations); + rh.neighbor_reachable_time_in_msec = + clib_host_to_net_u32 (radv_info-> + adv_neighbor_reachable_time_in_msec); + + rh.flags = + (radv_info->adv_managed_flag) ? + ICMP6_ROUTER_DISCOVERY_FLAG_ADDRESS_CONFIG_VIA_DHCP : + 0; + rh.flags |= + ((radv_info->adv_other_flag) ? + ICMP6_ROUTER_DISCOVERY_FLAG_OTHER_CONFIG_VIA_DHCP : + 0); + + + u16 payload_length = + sizeof (icmp6_router_advertisement_header_t); + + if (vlib_buffer_add_data + (vm, &bi0, (void *) &rh, + sizeof (icmp6_router_advertisement_header_t))) + { + /* buffer allocation failed, drop the pkt */ + error0 = ICMP6_ERROR_ALLOC_FAILURE; + goto drop0; + } + + if (radv_info->adv_link_layer_address) + { + icmp6_neighbor_discovery_ethernet_link_layer_address_option_t + h; + + h.header.type = + ICMP6_NEIGHBOR_DISCOVERY_OPTION_source_link_layer_address; + h.header.n_data_u64s = 1; + + /* copy ll address */ + clib_memcpy (&h.ethernet_address[0], + eth_if0->address, 6); + + if (vlib_buffer_add_data + (vm, &bi0, (void *) &h, + sizeof + (icmp6_neighbor_discovery_ethernet_link_layer_address_option_t))) + { + error0 = ICMP6_ERROR_ALLOC_FAILURE; + goto drop0; + } + + payload_length += + sizeof + (icmp6_neighbor_discovery_ethernet_link_layer_address_option_t); + } + + /* add MTU option */ + if (radv_info->adv_link_mtu) + { + icmp6_neighbor_discovery_mtu_option_t h; + + h.unused = 0; + h.mtu = + clib_host_to_net_u32 (radv_info->adv_link_mtu); + h.header.type = ICMP6_NEIGHBOR_DISCOVERY_OPTION_mtu; + h.header.n_data_u64s = 1; + + payload_length += + sizeof (icmp6_neighbor_discovery_mtu_option_t); + + if (vlib_buffer_add_data + (vm, &bi0, (void *) &h, + sizeof + (icmp6_neighbor_discovery_mtu_option_t))) + { + error0 = ICMP6_ERROR_ALLOC_FAILURE; + goto drop0; + } + } + + /* add advertised prefix options */ + ip6_radv_prefix_t *pr_info; + + /* *INDENT-OFF* */ + pool_foreach (pr_info, radv_info->adv_prefixes_pool, + ({ + if(pr_info->enabled && + (!pr_info->decrement_lifetime_flag + || (pr_info->pref_lifetime_expires >0))) + { + /* advertise this prefix */ + icmp6_neighbor_discovery_prefix_information_option_t h; + + h.header.type = ICMP6_NEIGHBOR_DISCOVERY_OPTION_prefix_information; + h.header.n_data_u64s = (sizeof(icmp6_neighbor_discovery_prefix_information_option_t) >> 3); + + h.dst_address_length = pr_info->prefix_len; + + h.flags = (pr_info->adv_on_link_flag) ? ICMP6_NEIGHBOR_DISCOVERY_PREFIX_INFORMATION_FLAG_ON_LINK : 0; + h.flags |= (pr_info->adv_autonomous_flag) ? ICMP6_NEIGHBOR_DISCOVERY_PREFIX_INFORMATION_AUTO : 0; + + if(radv_info->cease_radv && pr_info->deprecated_prefix_flag) + { + h.valid_time = clib_host_to_net_u32(MIN_ADV_VALID_LIFETIME); + h.preferred_time = 0; + } + else + { + if(pr_info->decrement_lifetime_flag) + { + pr_info->adv_valid_lifetime_in_secs = ((pr_info->valid_lifetime_expires > now)) ? + (pr_info->valid_lifetime_expires - now) : 0; + + pr_info->adv_pref_lifetime_in_secs = ((pr_info->pref_lifetime_expires > now)) ? + (pr_info->pref_lifetime_expires - now) : 0; + } + + h.valid_time = clib_host_to_net_u32(pr_info->adv_valid_lifetime_in_secs); + h.preferred_time = clib_host_to_net_u32(pr_info->adv_pref_lifetime_in_secs) ; + } + h.unused = 0; + + clib_warning ("Prefix %U valid %u preferred %u", + format_ip6_address, &pr_info->prefix, + clib_net_to_host_u32 (h.valid_time), + clib_net_to_host_u32 (h.preferred_time)); + + if (h.valid_time == 0) + clib_warning ("WARNING: valid_time 0!!!"); + + clib_memcpy(&h.dst_address, &pr_info->prefix, sizeof(ip6_address_t)); + + payload_length += sizeof( icmp6_neighbor_discovery_prefix_information_option_t); + + if (vlib_buffer_add_data + (vm, &bi0, (void *)&h, + sizeof(icmp6_neighbor_discovery_prefix_information_option_t))) + { + error0 = ICMP6_ERROR_ALLOC_FAILURE; + goto drop0; + } + + } + })); + /* *INDENT-ON* */ + + /* add additional options before here */ + + /* finish building the router advertisement... */ + if (!is_unspecified && radv_info->send_unicast) + { + ip0->dst_address = ip0->src_address; + } + else + { + /* target address is all-nodes mcast addr */ + ip6_set_reserved_multicast_address + (&ip0->dst_address, + IP6_MULTICAST_SCOPE_link_local, + IP6_MULTICAST_GROUP_ID_all_hosts); + } + + /* source address MUST be the link-local address */ + ip6_address_copy (&ip0->src_address, + ip6_get_link_local_address + (radv_info->sw_if_index)); + + ip0->hop_limit = 255; + ip0->payload_length = + clib_host_to_net_u16 (payload_length); + + icmp6_router_advertisement_header_t *rh0 = + (icmp6_router_advertisement_header_t *) (ip0 + 1); + rh0->icmp.checksum = + ip6_tcp_udp_icmp_compute_checksum (vm, p0, ip0, + &bogus_length); + ASSERT (bogus_length == 0); + + /* setup output if and adjacency */ + vnet_buffer (p0)->sw_if_index[VLIB_RX] = + vnet_main.local_interface_sw_if_index; + + if (is_solicitation) + { + ethernet_header_t *eth0; + /* Reuse current MAC header, copy SMAC to DMAC and + * interface MAC to SMAC */ + vlib_buffer_reset (p0); + eth0 = vlib_buffer_get_current (p0); + clib_memcpy (eth0->dst_address, eth0->src_address, + 6); + clib_memcpy (eth0->src_address, eth_if0->address, + 6); + next0 = + is_dropped ? next0 : + ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_TX; + vnet_buffer (p0)->sw_if_index[VLIB_TX] = + sw_if_index0; + } + else + { + adj_index0 = ip6_link_get_mcast_adj (sw_if_index0); + if (adj_index0 == INDEX_INVALID) + error0 = ICMP6_ERROR_DST_LOOKUP_MISS; + else + { + next0 = + is_dropped ? next0 : + ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_RW; + vnet_buffer (p0)->ip.adj_index[VLIB_TX] = + adj_index0; + } + } + p0->flags |= VNET_BUFFER_F_LOCALLY_ORIGINATED; + + radv_info->n_solicitations_dropped += is_dropped; + radv_info->n_solicitations_rcvd += is_solicitation; + + if ((error0 == ICMP6_ERROR_NONE) && !is_dropped) + { + radv_info->n_advertisements_sent++; + n_advertisements_sent++; + } + } + } + } + + drop0: + p0->error = error_node->errors[error0]; + + if (error0 != ICMP6_ERROR_NONE) + vlib_error_count (vm, error_node->node_index, error0, 1); + + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, + bi0, next0); + + } + + vlib_put_next_frame (vm, node, next_index, n_left_to_next); + } + + /* Account for router advertisements sent. */ + vlib_error_count (vm, error_node->node_index, + ICMP6_ERROR_ROUTER_ADVERTISEMENTS_TX, + n_advertisements_sent); + + return frame->n_vectors; +} + +/* *INDENT-OFF* */ +VLIB_REGISTER_NODE (ip6_icmp_router_solicitation_node,static) = +{ + .function = icmp6_router_solicitation, + .name = "icmp6-router-solicitation", + + .vector_size = sizeof (u32), + + .format_trace = format_icmp6_input_trace, + + .n_next_nodes = ICMP6_ROUTER_SOLICITATION_N_NEXT, + .next_nodes = { + [ICMP6_ROUTER_SOLICITATION_NEXT_DROP] = "ip6-drop", + [ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_RW] = "ip6-rewrite-mcast", + [ICMP6_ROUTER_SOLICITATION_NEXT_REPLY_TX] = "interface-output", + }, +}; +/* *INDENT-ON* */ + + /* validate advertised info for consistancy (see RFC-4861 section 6.2.7) - log any inconsistencies, packet will always be dropped */ +static_always_inline uword +icmp6_router_advertisement (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + vnet_main_t *vnm = vnet_get_main (); + uword n_packets = frame->n_vectors; + u32 *from, *to_next; + u32 n_left_from, n_left_to_next, next_index; + u32 n_advertisements_rcvd = 0; + + vlib_node_runtime_t *error_node = + vlib_node_get_runtime (vm, ip6_icmp_input_node.index); + + from = vlib_frame_vector_args (frame); + n_left_from = n_packets; + next_index = node->cached_next_index; + + if (node->flags & VLIB_NODE_FLAG_TRACE) + vlib_trace_frame_buffers_only (vm, node, from, frame->n_vectors, + /* stride */ 1, + sizeof (icmp6_input_trace_t)); + + 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) + { + vlib_buffer_t *p0; + ip6_header_t *ip0; + ip6_ra_t *radv_info = 0; + icmp6_router_advertisement_header_t *h0; + u32 bi0, options_len0, sw_if_index0, next0, error0; + + bi0 = to_next[0] = from[0]; + + from += 1; + to_next += 1; + n_left_from -= 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, bi0); + ip0 = vlib_buffer_get_current (p0); + h0 = ip6_next_header (ip0); + options_len0 = + clib_net_to_host_u16 (ip0->payload_length) - sizeof (h0[0]); + + error0 = ICMP6_ERROR_NONE; + sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX]; + + /* Check that source address is link-local */ + error0 = (!ip6_address_is_link_local_unicast (&ip0->src_address)) ? + ICMP6_ERROR_ROUTER_ADVERTISEMENT_SOURCE_NOT_LINK_LOCAL : error0; + + /* default is to drop */ + next0 = ICMP6_ROUTER_SOLICITATION_NEXT_DROP; + + n_advertisements_rcvd++; + + if (error0 == ICMP6_ERROR_NONE) + { + vnet_sw_interface_t *sw_if0; + ethernet_interface_t *eth_if0; + + sw_if0 = vnet_get_sup_sw_interface (vnm, sw_if_index0); + ASSERT (sw_if0->type == VNET_SW_INTERFACE_TYPE_HARDWARE); + eth_if0 = + ethernet_get_interface (ðernet_main, sw_if0->hw_if_index); + + /* only support ethernet interface type for now */ + error0 = + (!eth_if0) ? ICMP6_ERROR_ROUTER_SOLICITATION_UNSUPPORTED_INTF + : error0; + + if (error0 == ICMP6_ERROR_NONE) + { + /* look up the radv_t information for this interface */ + radv_info = ip6_ra_get_itf (sw_if_index0); + + error0 = ((!radv_info) ? + ICMP6_ERROR_ROUTER_SOLICITATION_RADV_NOT_CONFIG : + error0); + + if (error0 == ICMP6_ERROR_NONE) + { + radv_info->keep_sending_rs = 0; + + ip6_ra_report_t r; + + r.sw_if_index = sw_if_index0; + memcpy (&r.router_address, &ip0->src_address, 16); + r.current_hop_limit = h0->current_hop_limit; + r.flags = h0->flags; + r.router_lifetime_in_sec = + clib_net_to_host_u16 (h0->router_lifetime_in_sec); + r.neighbor_reachable_time_in_msec = + clib_net_to_host_u32 + (h0->neighbor_reachable_time_in_msec); + r.time_in_msec_between_retransmitted_neighbor_solicitations = clib_net_to_host_u32 (h0->time_in_msec_between_retransmitted_neighbor_solicitations); + r.prefixes = 0; + + /* validate advertised information */ + if ((h0->current_hop_limit && radv_info->curr_hop_limit) + && (h0->current_hop_limit != + radv_info->curr_hop_limit)) + { + ip6_neighbor_syslog (vm, LOG_WARNING, + "our AdvCurHopLimit on %U doesn't agree with %U", + format_vnet_sw_if_index_name, + vnm, sw_if_index0, + format_ip6_address, + &ip0->src_address); + } + + if ((h0->flags & + ICMP6_ROUTER_DISCOVERY_FLAG_ADDRESS_CONFIG_VIA_DHCP) + != radv_info->adv_managed_flag) + { + ip6_neighbor_syslog (vm, LOG_WARNING, + "our AdvManagedFlag on %U doesn't agree with %U", + format_vnet_sw_if_index_name, + vnm, sw_if_index0, + format_ip6_address, + &ip0->src_address); + } + + if ((h0->flags & + ICMP6_ROUTER_DISCOVERY_FLAG_OTHER_CONFIG_VIA_DHCP) + != radv_info->adv_other_flag) + { + ip6_neighbor_syslog (vm, LOG_WARNING, + "our AdvOtherConfigFlag on %U doesn't agree with %U", + format_vnet_sw_if_index_name, + vnm, sw_if_index0, + format_ip6_address, + &ip0->src_address); + } + + if ((h0-> + time_in_msec_between_retransmitted_neighbor_solicitations + && radv_info-> + adv_time_in_msec_between_retransmitted_neighbor_solicitations) + && (h0-> + time_in_msec_between_retransmitted_neighbor_solicitations + != + clib_host_to_net_u32 (radv_info-> + adv_time_in_msec_between_retransmitted_neighbor_solicitations))) + { + ip6_neighbor_syslog (vm, LOG_WARNING, + "our AdvRetransTimer on %U doesn't agree with %U", + format_vnet_sw_if_index_name, + vnm, sw_if_index0, + format_ip6_address, + &ip0->src_address); + } + + if ((h0->neighbor_reachable_time_in_msec && + radv_info->adv_neighbor_reachable_time_in_msec) && + (h0->neighbor_reachable_time_in_msec != + clib_host_to_net_u32 + (radv_info->adv_neighbor_reachable_time_in_msec))) + { + ip6_neighbor_syslog (vm, LOG_WARNING, + "our AdvReachableTime on %U doesn't agree with %U", + format_vnet_sw_if_index_name, + vnm, sw_if_index0, + format_ip6_address, + &ip0->src_address); + } + + /* check for MTU or prefix options or .. */ + u8 *opt_hdr = (u8 *) (h0 + 1); + while (options_len0 > 0) + { + icmp6_neighbor_discovery_option_header_t *o0 = + (icmp6_neighbor_discovery_option_header_t *) + opt_hdr; + int opt_len = o0->n_data_u64s << 3; + icmp6_neighbor_discovery_option_type_t option_type = + o0->type; + + if (options_len0 < 2) + { + ip6_neighbor_syslog (vm, LOG_ERR, + "malformed RA packet on %U from %U", + format_vnet_sw_if_index_name, + vnm, sw_if_index0, + format_ip6_address, + &ip0->src_address); + break; + } + + if (opt_len == 0) + { + ip6_neighbor_syslog (vm, LOG_ERR, + " zero length option in RA on %U from %U", + format_vnet_sw_if_index_name, + vnm, sw_if_index0, + format_ip6_address, + &ip0->src_address); + break; + } + else if (opt_len > options_len0) + { + ip6_neighbor_syslog (vm, LOG_ERR, + "option length in RA packet greater than total length on %U from %U", + format_vnet_sw_if_index_name, + vnm, sw_if_index0, + format_ip6_address, + &ip0->src_address); + break; + } + + options_len0 -= opt_len; + opt_hdr += opt_len; + + switch (option_type) + { + case ICMP6_NEIGHBOR_DISCOVERY_OPTION_source_link_layer_address: + { + icmp6_neighbor_discovery_ethernet_link_layer_address_option_t + * h = + (icmp6_neighbor_discovery_ethernet_link_layer_address_option_t + *) (o0); + + if (opt_len < sizeof (*h)) + break; + + memcpy (r.slla, h->ethernet_address, 6); + } + break; + + case ICMP6_NEIGHBOR_DISCOVERY_OPTION_mtu: + { + icmp6_neighbor_discovery_mtu_option_t *h = + (icmp6_neighbor_discovery_mtu_option_t + *) (o0); + + if (opt_len < sizeof (*h)) + break; + + r.mtu = clib_net_to_host_u32 (h->mtu); + + if ((h->mtu && radv_info->adv_link_mtu) && + (h->mtu != + clib_host_to_net_u32 + (radv_info->adv_link_mtu))) + { + ip6_neighbor_syslog (vm, LOG_WARNING, + "our AdvLinkMTU on %U doesn't agree with %U", + format_vnet_sw_if_index_name, + vnm, sw_if_index0, + format_ip6_address, + &ip0->src_address); + } + } + break; + + case ICMP6_NEIGHBOR_DISCOVERY_OPTION_prefix_information: + { + icmp6_neighbor_discovery_prefix_information_option_t + * h = + (icmp6_neighbor_discovery_prefix_information_option_t + *) (o0); + + /* validate advertised prefix options */ + ip6_radv_prefix_t *pr_info; + u32 preferred, valid; + + if (opt_len < sizeof (*h)) + break; + + vec_validate (r.prefixes, + vec_len (r.prefixes)); + ra_report_prefix_info_t *prefix = + vec_elt_at_index (r.prefixes, + vec_len (r.prefixes) - 1); + + preferred = + clib_net_to_host_u32 (h->preferred_time); + valid = clib_net_to_host_u32 (h->valid_time); + + prefix->preferred_time = preferred; + prefix->valid_time = valid; + prefix->flags = h->flags & 0xc0; + prefix->prefix.fp_len = h->dst_address_length; + prefix->prefix.fp_addr.ip6 = h->dst_address; + prefix->prefix.fp_proto = FIB_PROTOCOL_IP6; + + /* look for matching prefix - if we our advertising it, it better be consistant */ + /* *INDENT-OFF* */ + pool_foreach (pr_info, radv_info->adv_prefixes_pool, + ({ + + ip6_address_t mask; + ip6_address_mask_from_width(&mask, pr_info->prefix_len); + + if(pr_info->enabled && + (pr_info->prefix_len == h->dst_address_length) && + ip6_address_is_equal_masked (&pr_info->prefix, &h->dst_address, &mask)) + { + /* found it */ + if(!pr_info->decrement_lifetime_flag && + valid != pr_info->adv_valid_lifetime_in_secs) + { + ip6_neighbor_syslog(vm, LOG_WARNING, + "our ADV validlifetime on %U for %U does not agree with %U", + format_vnet_sw_if_index_name, vnm, sw_if_index0,format_ip6_address, &pr_info->prefix, + format_ip6_address, &h->dst_address); + } + if(!pr_info->decrement_lifetime_flag && + preferred != pr_info->adv_pref_lifetime_in_secs) + { + ip6_neighbor_syslog(vm, LOG_WARNING, + "our ADV preferredlifetime on %U for %U does not agree with %U", + format_vnet_sw_if_index_name, vnm, sw_if_index0,format_ip6_address, &pr_info->prefix, + format_ip6_address, &h->dst_address); + } + } + break; + })); + /* *INDENT-ON* */ + break; + } + default: + /* skip this one */ + break; + } + } + ip6_ra_publish (&r); + } + } + } + + p0->error = error_node->errors[error0]; + + if (error0 != ICMP6_ERROR_NONE) + vlib_error_count (vm, error_node->node_index, error0, 1); + + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, + bi0, next0); + } + + vlib_put_next_frame (vm, node, next_index, n_left_to_next); + } + + /* Account for router advertisements received. */ + vlib_error_count (vm, error_node->node_index, + ICMP6_ERROR_ROUTER_ADVERTISEMENTS_RX, + n_advertisements_rcvd); + + return frame->n_vectors; +} + +/* *INDENT-OFF* */ +VLIB_REGISTER_NODE (ip6_icmp_router_advertisement_node,static) = +{ + .function = icmp6_router_advertisement, + .name = "icmp6-router-advertisement", + + .vector_size = sizeof (u32), + + .format_trace = format_icmp6_input_trace, + + .n_next_nodes = 1, + .next_nodes = { + [0] = "ip6-drop", + }, +}; +/* *INDENT-ON* */ + +static inline f64 +random_f64_from_to (f64 from, f64 to) +{ + static u32 seed = 0; + static u8 seed_set = 0; + if (!seed_set) + { + seed = random_default_seed (); + seed_set = 1; + } + return random_f64 (&seed) * (to - from) + from; +} + +static inline u8 +get_mac_address (u32 sw_if_index, u8 * address) +{ + vnet_main_t *vnm = vnet_get_main (); + vnet_hw_interface_t *hw_if = vnet_get_sup_hw_interface (vnm, sw_if_index); + if (!hw_if->hw_address) + return 1; + clib_memcpy (address, hw_if->hw_address, 6); + return 0; +} + +static inline vlib_buffer_t * +create_buffer_for_rs (vlib_main_t * vm, ip6_ra_t * radv_info) +{ + u32 bi0; + vlib_buffer_t *p0; + icmp6_router_solicitation_header_t *rh; + u16 payload_length; + int bogus_length; + u32 sw_if_index; + + sw_if_index = radv_info->sw_if_index; + + if (vlib_buffer_alloc (vm, &bi0, 1) != 1) + { + clib_warning ("buffer allocation failure"); + return 0; + } + + p0 = vlib_get_buffer (vm, bi0); + VLIB_BUFFER_TRACE_TRAJECTORY_INIT (p0); + p0->flags |= VNET_BUFFER_F_LOCALLY_ORIGINATED; + + vnet_buffer (p0)->sw_if_index[VLIB_RX] = sw_if_index; + vnet_buffer (p0)->sw_if_index[VLIB_TX] = sw_if_index; + + vnet_buffer (p0)->ip.adj_index[VLIB_TX] = + ip6_link_get_mcast_adj (sw_if_index); + + rh = vlib_buffer_get_current (p0); + p0->current_length = sizeof (*rh); + + rh->neighbor.icmp.type = ICMP6_router_solicitation; + rh->neighbor.icmp.code = 0; + rh->neighbor.icmp.checksum = 0; + rh->neighbor.reserved_must_be_zero = 0; + + rh->link_layer_option.header.type = + ICMP6_NEIGHBOR_DISCOVERY_OPTION_source_link_layer_address; + if (0 != get_mac_address (sw_if_index, + rh->link_layer_option.ethernet_address)) + { + clib_warning ("interface with sw_if_index %u has no mac address", + sw_if_index); + vlib_buffer_free (vm, &bi0, 1); + return 0; + } + rh->link_layer_option.header.n_data_u64s = 1; + + payload_length = sizeof (rh->neighbor) + sizeof (u64); + + rh->ip.ip_version_traffic_class_and_flow_label = + clib_host_to_net_u32 (0x6 << 28); + rh->ip.payload_length = clib_host_to_net_u16 (payload_length); + rh->ip.protocol = IP_PROTOCOL_ICMP6; + rh->ip.hop_limit = 255; + ip6_address_copy (&rh->ip.src_address, + ip6_get_link_local_address (radv_info->sw_if_index)); + /* set address ff02::2 */ + rh->ip.dst_address.as_u64[0] = clib_host_to_net_u64 (0xff02ULL << 48); + rh->ip.dst_address.as_u64[1] = clib_host_to_net_u64 (2); + + rh->neighbor.icmp.checksum = ip6_tcp_udp_icmp_compute_checksum (vm, p0, + &rh->ip, + &bogus_length); + + return p0; +} + +static inline void +stop_sending_rs (vlib_main_t * vm, ip6_ra_t * ra) +{ + u32 bi0; + + ra->keep_sending_rs = 0; + if (ra->buffer) + { + bi0 = vlib_get_buffer_index (vm, ra->buffer); + vlib_buffer_free (vm, &bi0, 1); + ra->buffer = 0; + } +} + +static inline bool +check_send_rs (vlib_main_t * vm, ip6_ra_t * radv_info, f64 current_time, + f64 * due_time) +{ + vlib_buffer_t *p0; + vlib_frame_t *f; + u32 *to_next; + u32 next_index; + vlib_buffer_t *c0; + u32 ci0; + + icmp6_send_router_solicitation_params_t *params; + + if (!radv_info->keep_sending_rs) + return false; + + params = &radv_info->params; + + if (radv_info->due_time > current_time) + { + *due_time = radv_info->due_time; + return true; + } + + p0 = radv_info->buffer; + + next_index = ip6_rewrite_mcast_node.index; + + c0 = vlib_buffer_copy (vm, p0); + ci0 = vlib_get_buffer_index (vm, c0); + + f = vlib_get_frame_to_node (vm, next_index); + to_next = vlib_frame_vector_args (f); + to_next[0] = ci0; + f->n_vectors = 1; + vlib_put_frame_to_node (vm, next_index, f); + + if (params->mrc != 0 && --radv_info->n_left == 0) + stop_sending_rs (vm, radv_info); + else + { + radv_info->sleep_interval = + (2 + random_f64_from_to (-0.1, 0.1)) * radv_info->sleep_interval; + if (radv_info->sleep_interval > params->mrt) + radv_info->sleep_interval = + (1 + random_f64_from_to (-0.1, 0.1)) * params->mrt; + + radv_info->due_time = current_time + radv_info->sleep_interval; + + if (params->mrd != 0 + && current_time > radv_info->start_time + params->mrd) + stop_sending_rs (vm, radv_info); + else + *due_time = radv_info->due_time; + } + + return radv_info->keep_sending_rs; +} + +static uword +send_rs_process (vlib_main_t * vm, vlib_node_runtime_t * rt, + vlib_frame_t * f0) +{ + uword *event_data = NULL; + f64 sleep_time = 1e9; + ip6_ra_t *radv_info; + f64 current_time; + f64 due_time; + f64 dt = 0; + + while (true) + { + vlib_process_wait_for_event_or_clock (vm, sleep_time); + vlib_process_get_events (vm, &event_data); + vec_reset_length (event_data); + + current_time = vlib_time_now (vm); + do + { + due_time = current_time + 1e9; + /* *INDENT-OFF* */ + pool_foreach (radv_info, ip6_ra_pool, + ({ + if (check_send_rs (vm, radv_info, current_time, &dt) + && (dt < due_time)) + due_time = dt; + })); + /* *INDENT-ON* */ + current_time = vlib_time_now (vm); + } + while (due_time < current_time); + + sleep_time = due_time - current_time; + } + + return 0; +} + +/* *INDENT-OFF* */ +VLIB_REGISTER_NODE (ip6_rs_process_node) = { + .function = send_rs_process, + .type = VLIB_NODE_TYPE_PROCESS, + .name = "ip6-rs-process", +}; +/* *INDENT-ON* */ + +void +icmp6_send_router_solicitation (vlib_main_t * vm, u32 sw_if_index, u8 stop, + const icmp6_send_router_solicitation_params_t + * params) +{ + ip6_ra_t *ra; + + ASSERT (~0 != sw_if_index); + + ra = ip6_ra_get_itf (sw_if_index); + + if (!ra) + return; + + stop_sending_rs (vm, ra); + + if (!stop) + { + ra->keep_sending_rs = 1; + ra->params = *params; + ra->n_left = params->mrc; + ra->start_time = vlib_time_now (vm); + ra->sleep_interval = (1 + random_f64_from_to (-0.1, 0.1)) * params->irt; + ra->due_time = 0; /* send first packet ASAP */ + ra->buffer = create_buffer_for_rs (vm, ra); + if (!ra->buffer) + ra->keep_sending_rs = 0; + else + vlib_process_signal_event (vm, ip6_rs_process_node.index, 1, 0); + } +} + +static const ethernet_interface_t * +ip6_ra_get_eth_itf (u32 sw_if_index) +{ + const vnet_sw_interface_t *sw; + + /* lookup radv container - ethernet interfaces only */ + sw = vnet_get_sup_sw_interface (vnet_get_main (), sw_if_index); + if (sw->type == VNET_SW_INTERFACE_TYPE_HARDWARE) + return (ethernet_get_interface (ðernet_main, sw->hw_if_index)); + + return (NULL); +} + +/** + * @brief called when IP6 is enabled on an interface + *create and initialize router advertisement parameters with default + * values for this intfc + */ +static void +ip6_ra_link_enable (u32 sw_if_index) +{ + const ethernet_interface_t *eth; + ip6_ra_t *radv_info; + + eth = ip6_ra_get_eth_itf (sw_if_index); + + if (NULL == eth) + return; + + ASSERT (INDEX_INVALID == ip6_link_delegate_get (sw_if_index, + ip6_ra_delegate_id)); + + pool_get_zero (ip6_ra_pool, radv_info); + + radv_info->seed = (u32) clib_cpu_time_now (); + random_u32 (&radv_info->seed); + + radv_info->sw_if_index = sw_if_index; + radv_info->max_radv_interval = DEF_MAX_RADV_INTERVAL; + radv_info->min_radv_interval = DEF_MIN_RADV_INTERVAL; + radv_info->curr_hop_limit = DEF_CURR_HOP_LIMIT; + radv_info->adv_router_lifetime_in_sec = DEF_DEF_RTR_LIFETIME; + + /* send ll address source address option */ + radv_info->adv_link_layer_address = 1; + + radv_info->min_delay_between_radv = MIN_DELAY_BETWEEN_RAS; + radv_info->max_delay_between_radv = MAX_DELAY_BETWEEN_RAS; + radv_info->max_rtr_default_lifetime = MAX_DEF_RTR_LIFETIME; + + radv_info->initial_adverts_count = MAX_INITIAL_RTR_ADVERTISEMENTS; + radv_info->initial_adverts_sent = radv_info->initial_adverts_count - 1; + radv_info->initial_adverts_interval = MAX_INITIAL_RTR_ADVERT_INTERVAL; + + /* deafult is to send */ + radv_info->send_radv = 1; + + /* fill in delegate for this interface that will be needed later */ + radv_info->adv_link_mtu = + vnet_sw_interface_get_mtu (vnet_get_main (), sw_if_index, VNET_MTU_IP6); + + mhash_init (&radv_info->address_to_prefix_index, sizeof (uword), + sizeof (ip6_address_t)); + + ip6_link_delegate_update (sw_if_index, ip6_ra_delegate_id, + radv_info - ip6_ra_pool); +} + +static void +ip6_ra_delegate_disable (index_t rai) +{ + ip6_radv_prefix_t *p; + ip6_ra_t *radv_info; + + radv_info = pool_elt_at_index (ip6_ra_pool, rai); + + /* clean up prefix and MDP pools */ + /* *INDENT-OFF* */ + pool_flush(p, radv_info->adv_prefixes_pool, + ({ + mhash_unset (&radv_info->address_to_prefix_index, &p->prefix, 0); + })); + /* *INDENT-ON* */ + + pool_free (radv_info->adv_prefixes_pool); + + mhash_free (&radv_info->address_to_prefix_index); + + pool_put (ip6_ra_pool, radv_info); +} + +/* send a RA or update the timer info etc.. */ +static uword +ip6_ra_process_timer_event (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + vnet_main_t *vnm = vnet_get_main (); + ip6_ra_t *radv_info; + vlib_frame_t *f = 0; + u32 n_this_frame = 0; + u32 n_left_to_next = 0; + u32 *to_next = 0; + u32 bo0; + icmp6_router_solicitation_header_t *h0; + vlib_buffer_t *b0; + f64 now = vlib_time_now (vm); + + /* Interface ip6 radv info list */ + /* *INDENT-OFF* */ + pool_foreach (radv_info, ip6_ra_pool, + ({ + if( !vnet_sw_interface_is_admin_up (vnm, radv_info->sw_if_index)) + { + radv_info->initial_adverts_sent = radv_info->initial_adverts_count-1; + radv_info->next_multicast_time = now; + radv_info->last_multicast_time = now; + radv_info->last_radv_time = 0; + continue; + } + + /* is it time to send a multicast RA on this interface? */ + if(radv_info->send_radv && (now >= radv_info->next_multicast_time)) + { + u32 n_to_alloc = 1; + u32 n_allocated; + + f64 rfn = (radv_info->max_radv_interval - radv_info->min_radv_interval) * + random_f64 (&radv_info->seed) + radv_info->min_radv_interval; + + /* multicast send - compute next multicast send time */ + if( radv_info->initial_adverts_sent > 0) + { + radv_info->initial_adverts_sent--; + if(rfn > radv_info->initial_adverts_interval) + rfn = radv_info->initial_adverts_interval; + + /* check to see if we are ceasing to send */ + if( radv_info->initial_adverts_sent == 0) + if(radv_info->cease_radv) + radv_info->send_radv = 0; + } + + radv_info->next_multicast_time = rfn + now; + radv_info->last_multicast_time = now; + + /* send advert now - build a "solicted" router advert with unspecified source address */ + n_allocated = vlib_buffer_alloc (vm, &bo0, n_to_alloc); + + if (PREDICT_FALSE(n_allocated == 0)) + { + clib_warning ("buffer allocation failure"); + continue; + } + b0 = vlib_get_buffer (vm, bo0); + b0->current_length = sizeof( icmp6_router_solicitation_header_t); + b0->error = ICMP6_ERROR_NONE; + vnet_buffer (b0)->sw_if_index[VLIB_RX] = radv_info->sw_if_index; + + h0 = vlib_buffer_get_current (b0); + + clib_memset (h0, 0, sizeof (icmp6_router_solicitation_header_t)); + + h0->ip.ip_version_traffic_class_and_flow_label = clib_host_to_net_u32 (0x6 << 28); + h0->ip.payload_length = clib_host_to_net_u16 (sizeof (icmp6_router_solicitation_header_t) + - STRUCT_OFFSET_OF (icmp6_router_solicitation_header_t, neighbor)); + h0->ip.protocol = IP_PROTOCOL_ICMP6; + h0->ip.hop_limit = 255; + + /* set src/dst address as "unspecified" this marks this packet as internally generated rather than recieved */ + h0->ip.src_address.as_u64[0] = 0; + h0->ip.src_address.as_u64[1] = 0; + + h0->ip.dst_address.as_u64[0] = 0; + h0->ip.dst_address.as_u64[1] = 0; + + h0->neighbor.icmp.type = ICMP6_router_solicitation; + + if (PREDICT_FALSE(f == 0)) + { + f = vlib_get_frame_to_node (vm, ip6_icmp_router_solicitation_node.index); + to_next = vlib_frame_vector_args (f); + n_left_to_next = VLIB_FRAME_SIZE; + n_this_frame = 0; + } + + n_this_frame++; + n_left_to_next--; + to_next[0] = bo0; + to_next += 1; + + if (PREDICT_FALSE(n_left_to_next == 0)) + { + f->n_vectors = n_this_frame; + vlib_put_frame_to_node (vm, ip6_icmp_router_solicitation_node.index, f); + f = 0; + } + } + })); + /* *INDENT-ON* */ + + if (f) + { + ASSERT (n_this_frame); + f->n_vectors = n_this_frame; + vlib_put_frame_to_node (vm, ip6_icmp_router_solicitation_node.index, f); + } + return 0; +} + +static void +ip6_ra_handle_report (const ip6_ra_report_t * rap) +{ + u32 ii; + + vec_foreach_index (ii, ip6_ra_listeners) ip6_ra_listeners[ii] (rap); +} + +static uword +ip6_ra_event_process (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + ip6_ra_report_t *r, *rs; + uword event_type; + + /* init code here */ + + while (1) + { + vlib_process_wait_for_event_or_clock (vm, 1. /* seconds */ ); + + rs = vlib_process_get_event_data (vm, &event_type); + + if (NULL == rs) + { + /* No events found: timer expired. */ + /* process interface list and send RAs as appropriate, update timer info */ + ip6_ra_process_timer_event (vm, node, frame); + } + else + { + vec_foreach (r, rs) ip6_ra_handle_report (r); + vec_reset_length (rs); + } + } + return frame->n_vectors; +} + +VLIB_REGISTER_NODE (ip6_ra_process_node) = +{ +.function = ip6_ra_event_process,.name = "ip6-ra-process",.type = + VLIB_NODE_TYPE_PROCESS,}; + +static void +ip6_ra_signal_report (ip6_ra_report_t * r) +{ + vlib_main_t *vm = vlib_get_main (); + ip6_ra_report_t *q; + + if (!vec_len (ip6_ra_listeners)) + return; + + q = vlib_process_signal_event_data (vm, + ip6_ra_process_node.index, + 0, 1, sizeof *q); + *q = *r; +} + +static int +ip6_ra_publish (ip6_ra_report_t * r) +{ + void vl_api_rpc_call_main_thread (void *fp, u8 * data, u32 data_length); + vl_api_rpc_call_main_thread (ip6_ra_signal_report, (u8 *) r, sizeof *r); + return 0; +} + +/* API support functions */ +int +ip6_ra_config (vlib_main_t * vm, u32 sw_if_index, + u8 suppress, u8 managed, u8 other, + u8 ll_option, u8 send_unicast, u8 cease, + u8 use_lifetime, u32 lifetime, + u32 initial_count, u32 initial_interval, + u32 max_interval, u32 min_interval, u8 is_no) +{ + ip6_ra_t *radv_info; + + /* look up the radv_t information for this interface */ + radv_info = ip6_ra_get_itf (sw_if_index); + + if (!radv_info) + return (VNET_API_ERROR_IP6_NOT_ENABLED); + + if ((max_interval != 0) && (min_interval == 0)) + min_interval = .75 * max_interval; + + max_interval = + (max_interval != + 0) ? ((is_no) ? DEF_MAX_RADV_INTERVAL : max_interval) : + radv_info->max_radv_interval; + min_interval = + (min_interval != + 0) ? ((is_no) ? DEF_MIN_RADV_INTERVAL : min_interval) : + radv_info->min_radv_interval; + lifetime = + (use_lifetime != + 0) ? ((is_no) ? DEF_DEF_RTR_LIFETIME : lifetime) : + radv_info->adv_router_lifetime_in_sec; + + if (lifetime) + { + if (lifetime > MAX_DEF_RTR_LIFETIME) + lifetime = MAX_DEF_RTR_LIFETIME; + + if (lifetime <= max_interval) + return VNET_API_ERROR_INVALID_VALUE; + } + + if (min_interval != 0) + { + if ((min_interval > .75 * max_interval) || (min_interval < 3)) + return VNET_API_ERROR_INVALID_VALUE; + } + + if ((initial_count > MAX_INITIAL_RTR_ADVERTISEMENTS) || + (initial_interval > MAX_INITIAL_RTR_ADVERT_INTERVAL)) + return VNET_API_ERROR_INVALID_VALUE; + + /* + if "flag" is set and is_no is true then restore default value else set value corresponding to "flag" + if "flag" is clear don't change corresponding value + */ + radv_info->send_radv = + (suppress != 0) ? ((is_no != 0) ? 1 : 0) : radv_info->send_radv; + radv_info->adv_managed_flag = + (managed != 0) ? ((is_no) ? 0 : 1) : radv_info->adv_managed_flag; + radv_info->adv_other_flag = + (other != 0) ? ((is_no) ? 0 : 1) : radv_info->adv_other_flag; + radv_info->adv_link_layer_address = + (ll_option != 0) ? ((is_no) ? 1 : 0) : radv_info->adv_link_layer_address; + radv_info->send_unicast = + (send_unicast != 0) ? ((is_no) ? 0 : 1) : radv_info->send_unicast; + radv_info->cease_radv = + (cease != 0) ? ((is_no) ? 0 : 1) : radv_info->cease_radv; + + radv_info->min_radv_interval = min_interval; + radv_info->max_radv_interval = max_interval; + radv_info->adv_router_lifetime_in_sec = lifetime; + + radv_info->initial_adverts_count = + (initial_count != + 0) ? ((is_no) ? MAX_INITIAL_RTR_ADVERTISEMENTS : initial_count) : + radv_info->initial_adverts_count; + radv_info->initial_adverts_interval = + (initial_interval != + 0) ? ((is_no) ? MAX_INITIAL_RTR_ADVERT_INTERVAL : initial_interval) : + radv_info->initial_adverts_interval; + + /* restart */ + if ((cease != 0) && (is_no)) + radv_info->send_radv = 1; + + radv_info->initial_adverts_sent = radv_info->initial_adverts_count - 1; + radv_info->next_multicast_time = vlib_time_now (vm); + radv_info->last_multicast_time = vlib_time_now (vm); + radv_info->last_radv_time = 0; + + return (0); +} + + +int +ip6_ra_prefix (vlib_main_t * vm, u32 sw_if_index, + ip6_address_t * prefix_addr, u8 prefix_len, + u8 use_default, u32 val_lifetime, u32 pref_lifetime, + u8 no_advertise, u8 off_link, u8 no_autoconfig, + u8 no_onlink, u8 is_no) +{ + ip6_ra_t *radv_info; + + /* look up the radv_t information for this interface */ + radv_info = ip6_ra_get_itf (sw_if_index); + + if (!radv_info) + return (VNET_API_ERROR_IP6_NOT_ENABLED); + + f64 now = vlib_time_now (vm); + + /* prefix info add, delete or update */ + ip6_radv_prefix_t *prefix; + + /* lookup prefix info for this address on this interface */ + uword *p = mhash_get (&radv_info->address_to_prefix_index, prefix_addr); + + prefix = p ? pool_elt_at_index (radv_info->adv_prefixes_pool, p[0]) : 0; + + if (is_no) + { + /* delete */ + if (!prefix) + return VNET_API_ERROR_INVALID_VALUE; /* invalid prefix */ + + if (prefix->prefix_len != prefix_len) + return VNET_API_ERROR_INVALID_VALUE_2; + + /* FIXME - Should the DP do this or the CP ? */ + /* do specific delete processing here before returning */ + /* try to remove from routing table */ + + mhash_unset (&radv_info->address_to_prefix_index, prefix_addr, + /* old_value */ 0); + pool_put (radv_info->adv_prefixes_pool, prefix); + + radv_info->initial_adverts_sent = radv_info->initial_adverts_count - 1; + radv_info->next_multicast_time = vlib_time_now (vm); + radv_info->last_multicast_time = vlib_time_now (vm); + radv_info->last_radv_time = 0; + return (0); + } + + /* adding or changing */ + if (!prefix) + { + /* add */ + u32 pi; + pool_get_zero (radv_info->adv_prefixes_pool, prefix); + pi = prefix - radv_info->adv_prefixes_pool; + mhash_set (&radv_info->address_to_prefix_index, prefix_addr, pi, + /* old_value */ 0); + + clib_memset (prefix, 0x0, sizeof (ip6_radv_prefix_t)); + + prefix->prefix_len = prefix_len; + clib_memcpy (&prefix->prefix, prefix_addr, sizeof (ip6_address_t)); + + /* initialize default values */ + prefix->adv_on_link_flag = 1; /* L bit set */ + prefix->adv_autonomous_flag = 1; /* A bit set */ + prefix->adv_valid_lifetime_in_secs = DEF_ADV_VALID_LIFETIME; + prefix->adv_pref_lifetime_in_secs = DEF_ADV_PREF_LIFETIME; + prefix->enabled = 1; + prefix->decrement_lifetime_flag = 1; + prefix->deprecated_prefix_flag = 1; + + if (off_link == 0) + { + /* FIXME - Should the DP do this or the CP ? */ + /* insert prefix into routing table as a connected prefix */ + } + + if (use_default) + goto restart; + } + else + { + + if (prefix->prefix_len != prefix_len) + return VNET_API_ERROR_INVALID_VALUE_2; + + if (off_link != 0) + { + /* FIXME - Should the DP do this or the CP ? */ + /* remove from routing table if already there */ + } + } + + if ((val_lifetime == ~0) || (pref_lifetime == ~0)) + { + prefix->adv_valid_lifetime_in_secs = ~0; + prefix->adv_pref_lifetime_in_secs = ~0; + prefix->decrement_lifetime_flag = 0; + } + else + { + prefix->adv_valid_lifetime_in_secs = val_lifetime;; + prefix->adv_pref_lifetime_in_secs = pref_lifetime; + } + + /* copy remaining */ + prefix->enabled = !(no_advertise != 0); + prefix->adv_on_link_flag = !((off_link != 0) || (no_onlink != 0)); + prefix->adv_autonomous_flag = !(no_autoconfig != 0); + +restart: + /* restart */ + /* fill in the expiration times */ + prefix->valid_lifetime_expires = now + prefix->adv_valid_lifetime_in_secs; + prefix->pref_lifetime_expires = now + prefix->adv_pref_lifetime_in_secs; + + radv_info->initial_adverts_sent = radv_info->initial_adverts_count - 1; + radv_info->next_multicast_time = vlib_time_now (vm); + radv_info->last_multicast_time = vlib_time_now (vm); + radv_info->last_radv_time = 0; + + return (0); +} + +clib_error_t * +ip6_ra_cmd (vlib_main_t * vm, + unformat_input_t * main_input, vlib_cli_command_t * cmd) +{ + vnet_main_t *vnm = vnet_get_main (); + clib_error_t *error = 0; + u8 is_no = 0; + u8 suppress = 0, managed = 0, other = 0; + u8 suppress_ll_option = 0, send_unicast = 0, cease = 0; + u8 use_lifetime = 0; + u32 sw_if_index, ra_lifetime = 0, ra_initial_count = + 0, ra_initial_interval = 0; + u32 ra_max_interval = 0, ra_min_interval = 0; + + unformat_input_t _line_input, *line_input = &_line_input; + + int add_radv_info = 1; + ip6_address_t ip6_addr; + u32 addr_len; + + + /* Get a line of input. */ + if (!unformat_user (main_input, unformat_line_input, line_input)) + return 0; + + /* get basic radv info for this interface */ + if (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + + if (unformat_user (line_input, + unformat_vnet_sw_interface, vnm, &sw_if_index)) + { + if (!ip6_ra_get_eth_itf (sw_if_index)) + { + error = + clib_error_return (0, "Interface must be of ethernet type"); + goto done; + } + + if (!ip6_link_is_enabled (sw_if_index)) + { + error = clib_error_return (0, "IP6 nt enabler interface %U'", + format_unformat_error, line_input); + goto done; + } + } + else + { + error = clib_error_return (0, "invalid interface name %U'", + format_unformat_error, line_input); + goto done; + } + } + + /* get the rest of the command */ + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (line_input, "no")) + is_no = 1; + else if (unformat (line_input, "prefix %U/%d", + unformat_ip6_address, &ip6_addr, &addr_len)) + { + add_radv_info = 0; + break; + } + else if (unformat (line_input, "ra-managed-config-flag")) + { + managed = 1; + break; + } + else if (unformat (line_input, "ra-other-config-flag")) + { + other = 1; + break; + } + else if (unformat (line_input, "ra-suppress") || + unformat (line_input, "ra-surpress")) + { + suppress = 1; + break; + } + else if (unformat (line_input, "ra-suppress-link-layer") || + unformat (line_input, "ra-surpress-link-layer")) + { + suppress_ll_option = 1; + break; + } + else if (unformat (line_input, "ra-send-unicast")) + { + send_unicast = 1; + break; + } + else if (unformat (line_input, "ra-lifetime")) + { + if (!unformat (line_input, "%d", &ra_lifetime)) + { + error = unformat_parse_error (line_input); + goto done; + } + use_lifetime = 1; + break; + } + else if (unformat (line_input, "ra-initial")) + { + if (!unformat + (line_input, "%d %d", &ra_initial_count, &ra_initial_interval)) + { + error = unformat_parse_error (line_input); + goto done; + } + break; + } + else if (unformat (line_input, "ra-interval")) + { + if (!unformat (line_input, "%d", &ra_max_interval)) + { + error = unformat_parse_error (line_input); + goto done; + } + + if (!unformat (line_input, "%d", &ra_min_interval)) + ra_min_interval = 0; + break; + } + else if (unformat (line_input, "ra-cease")) + { + cease = 1; + break; + } + else + { + error = unformat_parse_error (line_input); + goto done; + } + } + + if (add_radv_info) + { + ip6_ra_config (vm, sw_if_index, + suppress, managed, other, + suppress_ll_option, send_unicast, cease, + use_lifetime, ra_lifetime, + ra_initial_count, ra_initial_interval, + ra_max_interval, ra_min_interval, is_no); + } + else + { + u32 valid_lifetime_in_secs = 0; + u32 pref_lifetime_in_secs = 0; + u8 use_prefix_default_values = 0; + u8 no_advertise = 0; + u8 off_link = 0; + u8 no_autoconfig = 0; + u8 no_onlink = 0; + + /* get the rest of the command */ + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (line_input, "default")) + { + use_prefix_default_values = 1; + break; + } + else if (unformat (line_input, "infinite")) + { + valid_lifetime_in_secs = ~0; + pref_lifetime_in_secs = ~0; + break; + } + else if (unformat (line_input, "%d %d", &valid_lifetime_in_secs, + &pref_lifetime_in_secs)) + break; + else + break; + } + + + /* get the rest of the command */ + while (!use_prefix_default_values && + unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (line_input, "no-advertise")) + no_advertise = 1; + else if (unformat (line_input, "off-link")) + off_link = 1; + else if (unformat (line_input, "no-autoconfig")) + no_autoconfig = 1; + else if (unformat (line_input, "no-onlink")) + no_onlink = 1; + else + { + error = unformat_parse_error (line_input); + goto done; + } + } + + ip6_ra_prefix (vm, sw_if_index, + &ip6_addr, addr_len, + use_prefix_default_values, + valid_lifetime_in_secs, + pref_lifetime_in_secs, + no_advertise, off_link, no_autoconfig, no_onlink, is_no); + } + +done: + unformat_free (line_input); + + return error; +} + +static u8 * +format_ip6_ra (u8 * s, va_list * args) +{ + index_t rai = va_arg (*args, index_t); + u32 indent = va_arg (*args, u32); + ip6_radv_prefix_t *p; + ip6_ra_t *radv_info; + + radv_info = pool_elt_at_index (ip6_ra_pool, rai); + + s = format (s, "%UAdvertised Prefixes:\n", format_white_space, indent); + + indent += 2; + + /* *INDENT-OFF* */ + pool_foreach (p, radv_info->adv_prefixes_pool, + ({ + s = format (s, "%Uprefix %U, length %d\n", + format_white_space, indent+2, + format_ip6_address, &p->prefix, p->prefix_len); + })); + /* *INDENT-ON* */ + + s = format (s, "%UMTU is %d\n", + format_white_space, indent, radv_info->adv_link_mtu); + s = format (s, "%UICMP error messages are unlimited\n", + format_white_space, indent); + s = format (s, "%UICMP redirects are disabled\n", + format_white_space, indent); + s = format (s, "%UICMP unreachables are not sent\n", + format_white_space, indent); + s = format (s, "%UND DAD is disabled\n", format_white_space, indent); + //s = format (s, "%UND reachable time is %d milliseconds\n",); + s = format (s, "%UND advertised reachable time is %d\n", + format_white_space, indent, + radv_info->adv_neighbor_reachable_time_in_msec); + s = format (s, + "%UND advertised retransmit interval is %d (msec)\n", + format_white_space, indent, + radv_info-> + adv_time_in_msec_between_retransmitted_neighbor_solicitations); + s = + format (s, + "%UND router advertisements are sent every %0.1f seconds (min interval is %0.1f)\n", + format_white_space, indent, radv_info->max_radv_interval, + radv_info->min_radv_interval); + s = + format (s, "%UND router advertisements live for %d seconds\n", + format_white_space, indent, + radv_info->adv_router_lifetime_in_sec); + s = + format (s, "%UHosts %s stateless autoconfig for addresses\n", + format_white_space, indent, + (radv_info->adv_managed_flag) ? "use" : " don't use"); + s = + format (s, "%UND router advertisements sent %d\n", format_white_space, + indent, radv_info->n_advertisements_sent); + s = + format (s, "%UND router solicitations received %d\n", format_white_space, + indent, radv_info->n_solicitations_rcvd); + s = + format (s, "%UND router solicitations dropped %d\n", format_white_space, + indent, radv_info->n_solicitations_dropped); + + return (s); +} + + +/*? + * This command is used to configure the neighbor discovery + * parameters on a given interface. Use the '<em>show ip6 interface</em>' + * command to display some of the current neighbor discovery parameters + * on a given interface. This command has three formats: + * + * + * <b>Format 1 - Router Advertisement Options:</b> (Only one can be entered in a single command) + * + * '<em><b>ip6 nd <interface> [no] [ra-managed-config-flag] | [ra-other-config-flag] | [ra-suppress] | [ra-suppress-link-layer] | [ra-send-unicast] | [ra-lifetime <lifetime>] | [ra-initial <cnt> <interval>] | [ra-interval <max-interval> [<min-interval>]] | [ra-cease]</b></em>' + * + * Where: + * + * <em>[no] ra-managed-config-flag</em> - Advertises in ICMPv6 + * router-advertisement messages to use stateful address + * auto-configuration to obtain address information (sets the M-bit). + * Default is the M-bit is not set and the '<em>no</em>' option + * returns it to this default state. + * + * <em>[no] ra-other-config-flag</em> - Indicates in ICMPv6 + * router-advertisement messages that hosts use stateful auto + * configuration to obtain nonaddress related information (sets + * the O-bit). Default is the O-bit is not set and the '<em>no</em>' + * option returns it to this default state. + * + * <em>[no] ra-suppress</em> - Disables sending ICMPv6 router-advertisement + * messages. The '<em>no</em>' option implies to enable sending ICMPv6 + * router-advertisement messages. + * + * <em>[no] ra-suppress-link-layer</em> - Indicates not to include the + * optional source link-layer address in the ICMPv6 router-advertisement + * messages. Default is to include the optional source link-layer address + * and the '<em>no</em>' option returns it to this default state. + * + * <em>[no] ra-send-unicast</em> - Use the source address of the + * router-solicitation message if availiable. The default is to use + * multicast address of all nodes, and the '<em>no</em>' option returns + * it to this default state. + * + * <em>[no] ra-lifetime <lifetime></em> - Advertises the lifetime of a + * default router in ICMPv6 router-advertisement messages. The range is + * from 0 to 9000 seconds. '<em><lifetime></em>' must be greater than + * '<em><max-interval></em>'. The default value is 600 seconds and the + * '<em>no</em>' option returns it to this default value. + * + * <em>[no] ra-initial <cnt> <interval></em> - Number of initial ICMPv6 + * router-advertisement messages sent and the interval between each + * message. Range for count is 1 - 3 and default is 3. Range for interval + * is 1 to 16 seconds, and default is 16 seconds. The '<em>no</em>' option + * returns both to their default value. + * + * <em>[no] ra-interval <max-interval> [<min-interval>]</em> - Configures the + * interval between sending ICMPv6 router-advertisement messages. The + * range for max-interval is from 4 to 200 seconds. min-interval can not + * be more than 75% of max-interval. If not set, min-interval will be + * set to 75% of max-interval. The range for min-interval is from 3 to + * 150 seconds. The '<em>no</em>' option returns both to their default + * value. + * + * <em>[no] ra-cease</em> - Cease sending ICMPv6 router-advertisement messages. + * The '<em>no</em>' options implies to start (or restart) sending + * ICMPv6 router-advertisement messages. + * + * + * <b>Format 2 - Prefix Options:</b> + * + * '<em><b>ip6 nd <interface> [no] prefix <ip6-address>/<width> [<valid-lifetime> <pref-lifetime> | infinite] [no-advertise] [off-link] [no-autoconfig] [no-onlink]</b></em>' + * + * Where: + * + * <em>no</em> - All additional flags are ignored and the prefix is deleted. + * + * <em><valid-lifetime> <pref-lifetime></em> - '<em><valid-lifetime></em>' is the + * length of time in seconds during what the prefix is valid for the purpose of + * on-link determination. Range is 7203 to 2592000 seconds and default is 2592000 + * seconds (30 days). '<em><pref-lifetime></em>' is the prefered-lifetime and is the + * length of time in seconds during what addresses generated from the prefix remain + * preferred. Range is 0 to 604800 seconds and default is 604800 seconds (7 days). + * + * <em>infinite</em> - Both '<em><valid-lifetime></em>' and '<em><<pref-lifetime></em>' + * are inifinte, no timeout. + * + * <em>no-advertise</em> - Do not send full router address in prefix + * advertisement. Default is to advertise (i.e. - This flag is off by default). + * + * <em>off-link</em> - Prefix is off-link, clear L-bit in packet. Default is on-link + * (i.e. - This flag is off and L-bit in packet is set by default and this prefix can + * be used for on-link determination). '<em>no-onlink</em>' also controls the L-bit. + * + * <em>no-autoconfig</em> - Do not use prefix for autoconfiguration, clear A-bit in packet. + * Default is autoconfig (i.e. - This flag is off and A-bit in packet is set by default. + * + * <em>no-onlink</em> - Do not use prefix for onlink determination, clear L-bit in packet. + * Default is on-link (i.e. - This flag is off and L-bit in packet is set by default and + * this prefix can be used for on-link determination). '<em>off-link</em>' also controls + * the L-bit. + * + * + * <b>Format 3: - Default of Prefix:</b> + * + * '<em><b>ip6 nd <interface> [no] prefix <ip6-address>/<width> default</b></em>' + * + * When a new prefix is added (or existing one is being overwritten) <em>default</em> + * uses default values for the prefix. If <em>no</em> is used, the <em>default</em> + * is ignored and the prefix is deleted. + * + * + * @cliexpar + * Example of how set a router advertisement option: + * @cliexcmd{ip6 nd GigabitEthernet2/0/0 ra-interval 100 20} + * Example of how to add a prefix: + * @cliexcmd{ip6 nd GigabitEthernet2/0/0 prefix fe80::fe:28ff:fe9c:75b3/64 infinite no-advertise} + * Example of how to delete a prefix: + * @cliexcmd{ip6 nd GigabitEthernet2/0/0 no prefix fe80::fe:28ff:fe9c:75b3/64} +?*/ +/* *INDENT-OFF* */ +VLIB_CLI_COMMAND (ip6_nd_command, static) = +{ + .path = "ip6 nd", + .short_help = "ip6 nd <interface> ...", + .function = ip6_ra_cmd, +}; +/* *INDENT-ON* */ + +/** + * VFT for registering as a delegate to an IP6 link + */ +const static ip6_link_delegate_vft_t ip6_ra_delegate_vft = { + .ildv_disable = ip6_ra_delegate_disable, + .ildv_enable = ip6_ra_link_enable, + .ildv_format = format_ip6_ra, +}; + +static clib_error_t * +ip6_ra_init (vlib_main_t * vm) +{ + vlib_call_init_function (vm, icmp6_init); + + icmp6_register_type (vm, ICMP6_router_solicitation, + ip6_icmp_router_solicitation_node.index); + icmp6_register_type (vm, ICMP6_router_advertisement, + ip6_icmp_router_advertisement_node.index); + + ip6_ra_delegate_id = ip6_link_delegate_register (&ip6_ra_delegate_vft); + + return (NULL); +} + +/* *INDENT-OFF* */ +VLIB_INIT_FUNCTION (ip6_ra_init) = +{ + .runs_after = VLIB_INITS("icmp6_init"), +}; +/* *INDENT-ON* */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip6-nd/ip6_ra.h b/src/vnet/ip6-nd/ip6_ra.h new file mode 100644 index 00000000000..4efd92e6968 --- /dev/null +++ b/src/vnet/ip6-nd/ip6_ra.h @@ -0,0 +1,89 @@ +/* + * + * ip6_neighboor.h: ip6 neighbor structures + * + * Copyright (c) 2016 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef __IP6_RA_H__ +#define __IP6_RA_H__ + +#include <vnet/fib/fib_types.h> + +extern int ip6_ra_config (vlib_main_t * vm, u32 sw_if_index, + u8 suppress, u8 managed, u8 other, + u8 ll_option, u8 send_unicast, u8 cease, + u8 use_lifetime, u32 lifetime, + u32 initial_count, u32 initial_interval, + u32 max_interval, u32 min_interval, u8 is_no); + +extern int ip6_ra_prefix (vlib_main_t * vm, u32 sw_if_index, + ip6_address_t * prefix_addr, u8 prefix_len, + u8 use_default, u32 val_lifetime, + u32 pref_lifetime, u8 no_advertise, + u8 off_link, u8 no_autoconfig, + u8 no_onlink, u8 is_no); + +typedef struct +{ + u32 irt; + u32 mrt; + u32 mrc; + u32 mrd; +} icmp6_send_router_solicitation_params_t; + +extern void icmp6_send_router_solicitation (vlib_main_t * vm, + u32 sw_if_index, + u8 stop, + const + icmp6_send_router_solicitation_params_t + * params); + +typedef struct +{ + fib_prefix_t prefix; + u8 flags; + u32 valid_time; + u32 preferred_time; +} ra_report_prefix_info_t; + +typedef struct +{ + ip6_address_t router_address; + u32 sw_if_index; + u8 current_hop_limit; + u8 flags; + u16 router_lifetime_in_sec; + u32 neighbor_reachable_time_in_msec; + u32 time_in_msec_between_retransmitted_neighbor_solicitations; + u8 slla[6]; + u32 mtu; + ra_report_prefix_info_t *prefixes; +} ip6_ra_report_t; + + +typedef void (*ip6_ra_report_notify_t) (const ip6_ra_report_t * rap); + +extern void ip6_ra_report_register (ip6_ra_report_notify_t fn); +extern void ip6_ra_report_unregister (ip6_ra_report_notify_t fn); + +#endif /* included_ip6_neighbor_h */ + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip/rd_cp.api b/src/vnet/ip6-nd/rd_cp.api index 58480789d16..58480789d16 100644 --- a/src/vnet/ip/rd_cp.api +++ b/src/vnet/ip6-nd/rd_cp.api diff --git a/src/vnet/ip/rd_cp.c b/src/vnet/ip6-nd/rd_cp.c index c9c387e07c0..ee7c323f8b1 100644 --- a/src/vnet/ip/rd_cp.c +++ b/src/vnet/ip6-nd/rd_cp.c @@ -13,29 +13,15 @@ * limitations under the License. */ -#include <vnet/vnet.h> -#include <vlibmemory/api.h> -#include <vnet/vnet_msg_enum.h> +#include <vnet/ip6-nd/ip6_ra.h> + #include <vnet/ip/ip6.h> +#include <vnet/ip/ip6_link.h> #include <vnet/ethernet/ethernet.h> -#include <vnet/ip/ip6_neighbor.h> #include <vnet/fib/fib_table.h> #include <signal.h> #include <math.h> -#define vl_typedefs /* define message structures */ -#include <vnet/vnet_all_api_h.h> -#undef vl_typedefs - -#define vl_endianfun /* define message structures */ -#include <vnet/vnet_all_api_h.h> -#undef vl_endianfun - -#include <vlibapi/api_helper_macros.h> - -#define foreach_rd_cp_msg \ -_(IP6_ND_ADDRESS_AUTOCONFIG, ip6_nd_address_autoconfig) - typedef struct { u32 sw_if_index; @@ -68,7 +54,6 @@ typedef struct /* binary API client */ u8 api_connected; - svm_queue_t *vl_input_queue; u32 my_client_index; /* logging */ @@ -77,7 +62,6 @@ typedef struct /* convenience */ vlib_main_t *vlib_main; vnet_main_t *vnet_main; - api_main_t *api_main; u32 node_index; } rd_cp_main_t; @@ -110,13 +94,13 @@ static void interrupt_process (void); static int add_slaac_address (vlib_main_t * vm, u32 sw_if_index, u8 address_length, - ip6_address_t * address, f64 due_time) + const ip6_address_t * address, f64 due_time) { rd_cp_main_t *rm = &rd_cp_main; slaac_address_t *slaac_address; clib_error_t *rv = 0; - pool_get (rm->slaac_address_pool, slaac_address); + pool_get_zero (rm->slaac_address_pool, slaac_address); slaac_address->sw_if_index = sw_if_index; slaac_address->address_length = address_length; @@ -132,12 +116,12 @@ add_slaac_address (vlib_main_t * vm, u32 sw_if_index, u8 address_length, static void add_default_route (vlib_main_t * vm, u32 sw_if_index, - ip6_address_t * next_hop_address, f64 due_time) + const ip6_address_t * next_hop_address, f64 due_time) { rd_cp_main_t *rm = &rd_cp_main; default_route_t *default_route; - pool_get (rm->default_route_pool, default_route); + pool_get_zero (rm->default_route_pool, default_route); default_route->sw_if_index = sw_if_index; default_route->router_address = *next_hop_address; @@ -232,17 +216,6 @@ get_interface_mac_address (u32 sw_if_index, u8 mac[]) return 0; } -static u32 -ip6_enable (u32 sw_if_index) -{ - rd_cp_main_t *rm = &rd_cp_main; - clib_error_t *rv; - - rv = enable_ip6_interface (rm->vlib_main, sw_if_index); - - return rv != 0; -} - static u8 ip6_prefixes_equal (ip6_address_t * prefix1, ip6_address_t * prefix2, u8 len) { @@ -261,13 +234,11 @@ ip6_prefixes_equal (ip6_address_t * prefix1, ip6_address_t * prefix2, u8 len) #define PREFIX_FLAG_A (1 << 6) #define PREFIX_FLAG_L (1 << 7) -static clib_error_t * -ip6_ra_report_handler (void *data) +static void +ip6_ra_report_handler (const ip6_ra_report_t * r) { rd_cp_main_t *rm = &rd_cp_main; vlib_main_t *vm = rm->vlib_main; - clib_error_t *error = 0; - ra_report_t *r = data; interface_config_t *if_config; default_route_t *default_route; slaac_address_t *slaac_address; @@ -284,7 +255,7 @@ ip6_ra_report_handler (void *data) sw_if_index = r->sw_if_index; if (sw_if_index >= vec_len (rm->config_by_sw_if_index)) - return 0; + return; if_config = &rm->config_by_sw_if_index[sw_if_index]; if (if_config->install_default_routes) @@ -326,11 +297,11 @@ ip6_ra_report_handler (void *data) if (get_interface_mac_address (sw_if_index, mac) != 0) { vlib_log_warn (rm->log_class, "Error getting MAC address"); - return clib_error_return (0, "Error getting MAC address"); + return; } if (!if_config->enabled) - return 0; + return; n_prefixes = vec_len (r->prefixes); for (i = 0; i < n_prefixes; i++) @@ -413,11 +384,9 @@ ip6_ra_report_handler (void *data) interrupt_process (); - return error; + return; } -VNET_IP6_NEIGHBOR_RA_FUNCTION (ip6_ra_report_handler); - static uword rd_cp_process (vlib_main_t * vm, vlib_node_runtime_t * rt, vlib_frame_t * f) { @@ -459,7 +428,7 @@ rd_cp_process (vlib_main_t * vm, vlib_node_runtime_t * rt, vlib_frame_t * f) u32 sw_if_index = slaac_address->sw_if_index; remove_slaac_address (vm, slaac_address); /* make sure ip6 stays enabled */ - ip6_enable (sw_if_index); + ip6_link_enable (sw_if_index); } })); pool_foreach_index (index, rm->default_route_pool, @@ -502,8 +471,9 @@ interrupt_process (void) RD_CP_EVENT_INTERRUPT, 0); } -static int -set_address_autoconfig (u32 sw_if_index, u8 enable, u8 install_default_routes) +int +rd_cp_set_address_autoconfig (u32 sw_if_index, + u8 enable, u8 install_default_routes) { rd_cp_main_t *rm = &rd_cp_main; vlib_main_t *vm = rm->vlib_main; @@ -534,7 +504,7 @@ set_address_autoconfig (u32 sw_if_index, u8 enable, u8 install_default_routes) if_config = &rm->config_by_sw_if_index[sw_if_index]; if (!if_config->enabled && enable) - ip6_enable (sw_if_index); + ip6_link_enable (sw_if_index); if ((!if_config->enabled && enable) || (!if_config->install_default_routes && install_default_routes)) @@ -593,7 +563,8 @@ ip6_nd_address_autoconfig (vlib_main_t * vm, if (sw_if_index != ~0) { - if (set_address_autoconfig (sw_if_index, enable, default_route) != 0) + if (rd_cp_set_address_autoconfig (sw_if_index, enable, default_route) != + 0) error = clib_error_return (0, "Invalid sw_if_index"); } else @@ -625,68 +596,20 @@ VLIB_CLI_COMMAND (ip6_nd_address_autoconfig_command, static) = { }; /* *INDENT-ON* */ -static void -vl_api_ip6_nd_address_autoconfig_t_handler (vl_api_ip6_nd_address_autoconfig_t - * mp) -{ - vl_api_ip6_nd_address_autoconfig_reply_t *rmp; - u32 sw_if_index; - int rv = 0; - - VALIDATE_SW_IF_INDEX (mp); - - sw_if_index = ntohl (mp->sw_if_index); - - rv = - set_address_autoconfig (sw_if_index, mp->enable, - mp->install_default_routes); - - BAD_SW_IF_INDEX_LABEL; - - REPLY_MACRO (VL_API_IP6_ND_ADDRESS_AUTOCONFIG_REPLY); -} - -#define vl_msg_name_crc_list -#include <vnet/ip/rd_cp.api.h> -#undef vl_msg_name_crc_list - -static void -setup_message_id_table (api_main_t * am) -{ -#define _(id,n,crc) vl_msg_api_add_msg_name_crc (am, #n "_" #crc, id); - foreach_vl_msg_name_crc_rd_cp; -#undef _ -} - static clib_error_t * rd_cp_init (vlib_main_t * vm) { rd_cp_main_t *rm = &rd_cp_main; - api_main_t *am = vlibapi_get_main (); rm->vlib_main = vm; rm->vnet_main = vnet_get_main (); - rm->api_main = am; rm->node_index = rd_cp_process_node.index; rm->log_class = vlib_log_register_class ("rd_cp", 0); -#define _(N,n) \ - vl_msg_api_set_handlers(VL_API_##N, #n, \ - vl_api_##n##_t_handler, \ - vl_noop_handler, \ - vl_api_##n##_t_endian, \ - vl_api_##n##_t_print, \ - sizeof(vl_api_##n##_t), 0/* do NOT trace! */); - foreach_rd_cp_msg; -#undef _ - - /* - * Set up the (msg_name, crc, message-id) table - */ - setup_message_id_table (am); + ip6_ra_report_register (ip6_ra_report_handler); - return 0; + return (NULL); } VLIB_INIT_FUNCTION (rd_cp_init); diff --git a/src/vnet/ip6-nd/rd_cp.h b/src/vnet/ip6-nd/rd_cp.h new file mode 100644 index 00000000000..2fe43da621b --- /dev/null +++ b/src/vnet/ip6-nd/rd_cp.h @@ -0,0 +1,33 @@ +/* + * Copyright (c) 2018 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#ifndef __RD_CP_H__ +#define __RD_CP_H__ + +#include <vnet/ip6-nd/ip6_nd.h> + +extern int rd_cp_set_address_autoconfig (u32 sw_if_index, + u8 enable, + u8 install_default_routes); + +#endif + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/ip6-nd/rd_cp_api.c b/src/vnet/ip6-nd/rd_cp_api.c new file mode 100644 index 00000000000..1f0d8587970 --- /dev/null +++ b/src/vnet/ip6-nd/rd_cp_api.c @@ -0,0 +1,73 @@ +/* + * Copyright (c) 2018 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/ip6-nd/rd_cp.h> + +#include <vlibapi/api.h> +#include <vlibmemory/api.h> + +/* define message IDs */ +#include <vnet/format_fns.h> +#include <vnet/ip6-nd/rd_cp.api_enum.h> +#include <vnet/ip6-nd/rd_cp.api_types.h> + +/** + * Base message ID fot the plugin + */ +static u32 rd_cp_base_msg_id; +#define REPLY_MSG_ID_BASE rd_cp_base_msg_id + +#include <vlibapi/api_helper_macros.h> + + +static void +vl_api_ip6_nd_address_autoconfig_t_handler (vl_api_ip6_nd_address_autoconfig_t + * mp) +{ + vl_api_ip6_nd_address_autoconfig_reply_t *rmp; + u32 sw_if_index; + int rv = 0; + + VALIDATE_SW_IF_INDEX (mp); + + sw_if_index = ntohl (mp->sw_if_index); + + rv = rd_cp_set_address_autoconfig (sw_if_index, + mp->enable, mp->install_default_routes); + + BAD_SW_IF_INDEX_LABEL; + + REPLY_MACRO (VL_API_IP6_ND_ADDRESS_AUTOCONFIG_REPLY); +} + +#include <vnet/ip6-nd/rd_cp.api.c> + +static clib_error_t * +rd_cp_api_init (vlib_main_t * vm) +{ + rd_cp_base_msg_id = setup_message_id_table (); + + return (NULL); +} + +VLIB_INIT_FUNCTION (rd_cp_api_init); + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/l2/l2.api b/src/vnet/l2/l2.api index b0031ba16cc..06663595cf7 100644 --- a/src/vnet/l2/l2.api +++ b/src/vnet/l2/l2.api @@ -18,6 +18,7 @@ option version = "2.2.2"; import "vnet/ip/ip_types.api"; import "vnet/ethernet/ethernet_types.api"; +import "vnet/interface_types.api"; /** \brief Reply to l2_xconnect_dump @param context - sender context which was passed in the request @@ -154,7 +155,7 @@ autoreply define want_l2_macs_events u32 learn_limit; u8 scan_delay; u8 max_macs_in_event; - u8 enable_disable; + bool enable_disable; u32 pid; }; @@ -598,6 +599,43 @@ autoreply define bvi_delete u32 sw_if_index; }; +/** \brief Register for IP4 ARP resolution event on receing ARP reply or + MAC/IP info from ARP requests in L2 BDs + @param client_index - opaque cookie to identify the sender + @param context - sender context, to match reply w/ request + @param enable - 1 => register for events, 0 => cancel registration + @param pid - sender's pid +*/ +autoreply define want_l2_arp_term_events +{ + u32 client_index; + u32 context; + bool enable; + u32 pid; +}; + +/** \brief Tell client about an IP4 ARP resolution event or + MAC/IP info from ARP requests in L2 BDs + @param client_index - opaque cookie to identify the sender + @param pid - client pid registered to receive notification + @param ip - IP address of new ARP term entry + @param sw_if_index - interface of new ARP term entry + @param mac - MAC address of new ARP term entry +*/ +define l2_arp_term_event +{ + u32 client_index; + u32 pid; + vl_api_address_t ip; + vl_api_interface_index_t sw_if_index; + vl_api_mac_address_t mac; +}; + +service { + rpc want_l2_arp_term_events returns want_l2_arp_term_events_reply + events l2_arp_term_event; +}; + /* * Local Variables: * eval: (c-set-style "gnu") diff --git a/src/vnet/l2/l2_api.c b/src/vnet/l2/l2_api.c index e8b103ac2c7..fa852742f9f 100644 --- a/src/vnet/l2/l2_api.c +++ b/src/vnet/l2/l2_api.c @@ -28,6 +28,7 @@ #include <vnet/l2/l2_learn.h> #include <vnet/l2/l2_bd.h> #include <vnet/l2/l2_bvi.h> +#include <vnet/l2/l2_arp_term.h> #include <vnet/ip/ip_types_api.h> #include <vnet/ethernet/ethernet_types_api.h> @@ -77,7 +78,8 @@ _(L2_INTERFACE_PBB_TAG_REWRITE, l2_interface_pbb_tag_rewrite) \ _(BRIDGE_DOMAIN_SET_MAC_AGE, bridge_domain_set_mac_age) \ _(SW_INTERFACE_SET_VPATH, sw_interface_set_vpath) \ _(BVI_CREATE, bvi_create) \ -_(BVI_DELETE, bvi_delete) +_(BVI_DELETE, bvi_delete) \ +_(WANT_L2_ARP_TERM_EVENTS, want_l2_arp_term_events) static void send_l2_xconnect_details (vl_api_registration_t * reg, u32 context, @@ -1045,6 +1047,169 @@ vl_api_bvi_delete_t_handler (vl_api_bvi_delete_t * mp) REPLY_MACRO (VL_API_BVI_DELETE_REPLY); } +static bool +l2_arp_term_publish_event_is_equal (const l2_arp_term_publish_event_t * e1, + const l2_arp_term_publish_event_t * e2) +{ + if (e1 == NULL || e2 == NULL) + return false; + return (ip46_address_is_equal (&e1->ip, &e2->ip) && + (e1->sw_if_index == e2->sw_if_index) && + (mac_address_equal (&e1->mac, &e2->mac))); +} + +static uword +l2_arp_term_process (vlib_main_t * vm, vlib_node_runtime_t * rt, + vlib_frame_t * f) +{ + /* These cross the longjmp boundary (vlib_process_wait_for_event) + * and need to be volatile - to prevent them from being optimized into + * a register - which could change during suspension */ + volatile f64 last = vlib_time_now (vm); + volatile l2_arp_term_publish_event_t last_event = { }; + + l2_arp_term_main_t *l2am = &l2_arp_term_main; + + while (1) + { + uword event_type = L2_ARP_TERM_EVENT_PUBLISH; + vpe_client_registration_t *reg; + f64 now; + + vlib_process_wait_for_event (vm); + + vlib_process_get_event_data (vm, &event_type); + now = vlib_time_now (vm); + + if (event_type == L2_ARP_TERM_EVENT_PUBLISH) + { + l2_arp_term_publish_event_t *event; + + vec_foreach (event, l2am->publish_events) + { + /* dampen duplicate events - cast away volatile */ + if (l2_arp_term_publish_event_is_equal + (event, (l2_arp_term_publish_event_t *) & last_event) && + (now - last) < 10.0) + { + continue; + } + last_event = *event; + last = now; + + /* *INDENT-OFF* */ + pool_foreach(reg, vpe_api_main.l2_arp_term_events_registrations, + ({ + vl_api_registration_t *vl_reg; + vl_reg = vl_api_client_index_to_registration (reg->client_index); + ASSERT (vl_reg != NULL); + + if (reg && vl_api_can_send_msg (vl_reg)) + { + vl_api_l2_arp_term_event_t * vevent; + vevent = vl_msg_api_alloc (sizeof *vevent); + clib_memset (vevent, 0, sizeof *vevent); + vevent->_vl_msg_id = htons (VL_API_L2_ARP_TERM_EVENT); + vevent->client_index = reg->client_index; + vevent->pid = reg->client_pid; + ip_address_encode(&event->ip, + event->type, + &vevent->ip); + vevent->sw_if_index = htonl(event->sw_if_index); + mac_address_encode(&event->mac, vevent->mac); + vl_api_send_msg (vl_reg, (u8 *) vevent); + } + })); + /* *INDENT-ON* */ + } + vec_reset_length (l2am->publish_events); + } + } + + return 0; +} + +/* *INDENT-OFF* */ +VLIB_REGISTER_NODE (l2_arp_term_process_node) = { + .function = l2_arp_term_process, + .type = VLIB_NODE_TYPE_PROCESS, + .name = "l2-arp-term-publisher", +}; +/* *INDENT-ON* */ + +static void +vl_api_want_l2_arp_term_events_t_handler (vl_api_want_l2_arp_term_events_t * + mp) +{ + vl_api_want_l2_arp_term_events_reply_t *rmp; + vpe_api_main_t *am = &vpe_api_main; + vpe_client_registration_t *rp; + int rv = 0; + uword *p; + + p = hash_get (am->l2_arp_term_events_registration_hash, mp->client_index); + + if (p) + { + if (mp->enable) + { + clib_warning ("pid %d: already enabled...", mp->pid); + rv = VNET_API_ERROR_INVALID_REGISTRATION; + goto reply; + } + else + { + rp = pool_elt_at_index (am->l2_arp_term_events_registrations, p[0]); + pool_put (am->l2_arp_term_events_registrations, rp); + hash_unset (am->l2_arp_term_events_registration_hash, + mp->client_index); + if (pool_elts (am->l2_arp_term_events_registrations) == 0) + l2_arp_term_set_publisher_node (false); + goto reply; + } + } + if (mp->enable == 0) + { + clib_warning ("pid %d: already disabled...", mp->pid); + rv = VNET_API_ERROR_INVALID_REGISTRATION; + goto reply; + } + pool_get (am->l2_arp_term_events_registrations, rp); + rp->client_index = mp->client_index; + rp->client_pid = mp->pid; + hash_set (am->l2_arp_term_events_registration_hash, rp->client_index, + rp - am->l2_arp_term_events_registrations); + l2_arp_term_set_publisher_node (true); + +reply: + REPLY_MACRO (VL_API_WANT_L2_ARP_TERM_EVENTS_REPLY); +} + +static clib_error_t * +want_l2_arp_term_events_reaper (u32 client_index) +{ + vpe_client_registration_t *rp; + vpe_api_main_t *am; + uword *p; + + am = &vpe_api_main; + + /* remove from the registration hash */ + p = hash_get (am->l2_arp_term_events_registration_hash, client_index); + + if (p) + { + rp = pool_elt_at_index (am->l2_arp_term_events_registrations, p[0]); + pool_put (am->l2_arp_term_events_registrations, rp); + hash_unset (am->l2_arp_term_events_registration_hash, client_index); + if (pool_elts (am->l2_arp_term_events_registrations) == 0) + l2_arp_term_set_publisher_node (false); + } + return (NULL); +} + +VL_MSG_API_REAPER_FUNCTION (want_l2_arp_term_events_reaper); + /* * l2_api_hookup * Add vpe's API message handlers to the table. diff --git a/src/vnet/l2/l2_arp_term.c b/src/vnet/l2/l2_arp_term.c new file mode 100644 index 00000000000..982fd9f05ea --- /dev/null +++ b/src/vnet/l2/l2_arp_term.c @@ -0,0 +1,481 @@ +/* + * l2/l2_arp_term.c: IP v4 ARP L2 BD termination + * + * Copyright (c) 2010 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 <vlibmemory/api.h> + +#include <vnet/l2/l2_arp_term.h> +#include <vnet/l2/l2_input.h> +#include <vnet/l2/feat_bitmap.h> + +#include <vnet/ip/ip4_packet.h> +#include <vnet/ethernet/arp_packet.h> + +static const u8 vrrp_prefix[] = { 0x00, 0x00, 0x5E, 0x00, 0x01 }; + +l2_arp_term_main_t l2_arp_term_main; + +/* + * ARP/ND Termination in a L2 Bridge Domain based on IP4/IP6 to MAC + * hash tables mac_by_ip4 and mac_by_ip6 for each BD. + */ +typedef enum +{ + ARP_TERM_NEXT_L2_OUTPUT, + ARP_TERM_NEXT_DROP, + ARP_TERM_N_NEXT, +} arp_term_next_t; + +u32 arp_term_next_node_index[32]; + +typedef struct +{ + u8 packet_data[64]; +} ethernet_arp_input_trace_t; + +#define foreach_ethernet_arp_error \ + _ (replies_sent, "ARP replies sent") \ + _ (l2_type_not_ethernet, "L2 type not ethernet") \ + _ (l3_type_not_ip4, "L3 type not IP4") \ + _ (l3_src_address_not_local, "IP4 source address not local to subnet") \ + _ (l3_dst_address_not_local, "IP4 destination address not local to subnet") \ + _ (l3_dst_address_unset, "IP4 destination address is unset") \ + _ (l3_src_address_is_local, "IP4 source address matches local interface") \ + _ (l3_src_address_learned, "ARP request IP4 source address learned") \ + _ (replies_received, "ARP replies received") \ + _ (opcode_not_request, "ARP opcode not request") \ + _ (proxy_arp_replies_sent, "Proxy ARP replies sent") \ + _ (l2_address_mismatch, "ARP hw addr does not match L2 frame src addr") \ + _ (gratuitous_arp, "ARP probe or announcement dropped") \ + _ (interface_no_table, "Interface is not mapped to an IP table") \ + _ (interface_not_ip_enabled, "Interface is not IP enabled") \ + _ (unnumbered_mismatch, "RX interface is unnumbered to different subnet") \ + +typedef enum +{ +#define _(sym,string) ETHERNET_ARP_ERROR_##sym, + foreach_ethernet_arp_error +#undef _ + ETHERNET_ARP_N_ERROR, +} ethernet_arp_reply_error_t; + +static char *ethernet_arp_error_strings[] = { +#define _(sym,string) string, + foreach_ethernet_arp_error +#undef _ +}; + +static u8 * +format_arp_term_input_trace (u8 * s, va_list * va) +{ + CLIB_UNUSED (vlib_main_t * vm) = va_arg (*va, vlib_main_t *); + CLIB_UNUSED (vlib_node_t * node) = va_arg (*va, vlib_node_t *); + ethernet_arp_input_trace_t *t = va_arg (*va, ethernet_arp_input_trace_t *); + + /* arp-term trace data saved is either arp or ip6/icmp6 packet: + - for arp, the 1st 16-bit field is hw type of value of 0x0001. + - for ip6, the first nibble has value of 6. */ + s = format (s, "%U", t->packet_data[0] == 0 ? + format_ethernet_arp_header : format_ip6_header, + t->packet_data, sizeof (t->packet_data)); + + return s; +} + +void +l2_arp_term_set_publisher_node (bool on) +{ + l2_arp_term_main_t *l2am = &l2_arp_term_main; + + l2am->publish = on; +} + +static int +l2_arp_term_publish (l2_arp_term_publish_event_t * ctx) +{ + l2_arp_term_main_t *l2am = &l2_arp_term_main; + + vec_add1 (l2am->publish_events, *ctx); + + vlib_process_signal_event (vlib_get_main (), + l2_arp_term_process_node.index, + L2_ARP_TERM_EVENT_PUBLISH, 0); + + return 0; +} + +static inline void +l2_arp_term_publish_v4_dp (u32 sw_if_index, + const ethernet_arp_ip4_over_ethernet_address_t * a) +{ + l2_arp_term_main_t *l2am = &l2_arp_term_main; + + if (!l2am->publish) + return; + + l2_arp_term_publish_event_t args = { + .sw_if_index = sw_if_index, + .type = IP46_TYPE_IP4, + .ip.ip4 = a->ip4, + .mac = a->mac, + }; + + vl_api_rpc_call_main_thread (l2_arp_term_publish, (u8 *) & args, + sizeof (args)); +} + +static inline void +l2_arp_term_publish_v6_dp (u32 sw_if_index, + const ip6_address_t * addr, + const mac_address_t * mac) +{ + l2_arp_term_main_t *l2am = &l2_arp_term_main; + + if (!l2am->publish) + return; + + l2_arp_term_publish_event_t args = { + .sw_if_index = sw_if_index, + .type = IP46_TYPE_IP6, + .ip.ip6 = *addr, + .mac = *mac, + }; + + vl_api_rpc_call_main_thread (l2_arp_term_publish, (u8 *) & args, + sizeof (args)); +} + +static inline int +vnet_ip6_nd_term (vlib_main_t * vm, + vlib_node_runtime_t * node, + vlib_buffer_t * p0, + ethernet_header_t * eth, + ip6_header_t * ip, u32 sw_if_index, u16 bd_index) +{ + icmp6_neighbor_solicitation_or_advertisement_header_t *ndh; + mac_address_t mac; + + mac_address_from_bytes (&mac, eth->src_address); + ndh = ip6_next_header (ip); + if (ndh->icmp.type != ICMP6_neighbor_solicitation && + ndh->icmp.type != ICMP6_neighbor_advertisement) + return 0; + + if (PREDICT_FALSE ((node->flags & VLIB_NODE_FLAG_TRACE) && + (p0->flags & VLIB_BUFFER_IS_TRACED))) + { + u8 *t0 = vlib_add_trace (vm, node, p0, + sizeof (icmp6_input_trace_t)); + clib_memcpy (t0, ip, sizeof (icmp6_input_trace_t)); + } + + /* Check if anyone want ND events for L2 BDs */ + if (PREDICT_FALSE (!ip6_address_is_link_local_unicast (&ip->src_address))) + { + l2_arp_term_publish_v6_dp (sw_if_index, &ip->src_address, &mac); + } + + /* Check if MAC entry exsist for solicited target IP */ + if (ndh->icmp.type == ICMP6_neighbor_solicitation) + { + icmp6_neighbor_discovery_ethernet_link_layer_address_option_t *opt; + l2_bridge_domain_t *bd_config; + u8 *macp; + + opt = (void *) (ndh + 1); + if ((opt->header.type != + ICMP6_NEIGHBOR_DISCOVERY_OPTION_source_link_layer_address) || + (opt->header.n_data_u64s != 1)) + return 0; /* source link layer address option not present */ + + bd_config = vec_elt_at_index (l2input_main.bd_configs, bd_index); + macp = + (u8 *) hash_get_mem (bd_config->mac_by_ip6, &ndh->target_address); + if (macp) + { /* found ip-mac entry, generate eighbor advertisement response */ + int bogus_length; + vlib_node_runtime_t *error_node = + vlib_node_get_runtime (vm, ip6_icmp_input_node.index); + ip->dst_address = ip->src_address; + ip->src_address = ndh->target_address; + ip->hop_limit = 255; + opt->header.type = + ICMP6_NEIGHBOR_DISCOVERY_OPTION_target_link_layer_address; + clib_memcpy (opt->ethernet_address, macp, 6); + ndh->icmp.type = ICMP6_neighbor_advertisement; + ndh->advertisement_flags = clib_host_to_net_u32 + (ICMP6_NEIGHBOR_ADVERTISEMENT_FLAG_SOLICITED | + ICMP6_NEIGHBOR_ADVERTISEMENT_FLAG_OVERRIDE); + ndh->icmp.checksum = 0; + ndh->icmp.checksum = + ip6_tcp_udp_icmp_compute_checksum (vm, p0, ip, &bogus_length); + clib_memcpy (eth->dst_address, eth->src_address, 6); + clib_memcpy (eth->src_address, macp, 6); + vlib_error_count (vm, error_node->node_index, + ICMP6_ERROR_NEIGHBOR_ADVERTISEMENTS_TX, 1); + return 1; + } + } + + return 0; + +} + +static uword +arp_term_l2bd (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + l2input_main_t *l2im = &l2input_main; + u32 n_left_from, next_index, *from, *to_next; + u32 n_replies_sent = 0; + u16 last_bd_index = ~0; + l2_bridge_domain_t *last_bd_config = 0; + l2_input_config_t *cfg0; + + from = vlib_frame_vector_args (frame); + n_left_from = frame->n_vectors; + next_index = node->cached_next_index; + + while (n_left_from > 0) + { + u32 n_left_to_next; + + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); + + while (n_left_from > 0 && n_left_to_next > 0) + { + vlib_buffer_t *p0; + ethernet_header_t *eth0; + ethernet_arp_header_t *arp0; + ip6_header_t *iph0; + u8 *l3h0; + u32 pi0, error0, next0, sw_if_index0; + u16 ethertype0; + u16 bd_index0; + u32 ip0; + u8 *macp0; + + pi0 = from[0]; + to_next[0] = pi0; + from += 1; + to_next += 1; + n_left_from -= 1; + n_left_to_next -= 1; + + p0 = vlib_get_buffer (vm, pi0); + // Terminate only local (SHG == 0) ARP + if (vnet_buffer (p0)->l2.shg != 0) + goto next_l2_feature; + + eth0 = vlib_buffer_get_current (p0); + l3h0 = (u8 *) eth0 + vnet_buffer (p0)->l2.l2_len; + ethertype0 = clib_net_to_host_u16 (*(u16 *) (l3h0 - 2)); + arp0 = (ethernet_arp_header_t *) l3h0; + + if (ethertype0 != ETHERNET_TYPE_ARP) + goto check_ip6_nd; + + if ((arp0->opcode != + clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_request)) && + (arp0->opcode != + clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_reply))) + goto check_ip6_nd; + + /* Must be ARP request/reply packet here */ + if (PREDICT_FALSE ((node->flags & VLIB_NODE_FLAG_TRACE) && + (p0->flags & VLIB_BUFFER_IS_TRACED))) + { + u8 *t0 = vlib_add_trace (vm, node, p0, + sizeof (ethernet_arp_input_trace_t)); + clib_memcpy_fast (t0, l3h0, + sizeof (ethernet_arp_input_trace_t)); + } + + error0 = 0; + error0 = + (arp0->l2_type != + clib_net_to_host_u16 (ETHERNET_ARP_HARDWARE_TYPE_ethernet) + ? ETHERNET_ARP_ERROR_l2_type_not_ethernet : error0); + error0 = + (arp0->l3_type != + clib_net_to_host_u16 (ETHERNET_TYPE_IP4) ? + ETHERNET_ARP_ERROR_l3_type_not_ip4 : error0); + + sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX]; + + if (error0) + goto drop; + + /* Trash ARP packets whose ARP-level source addresses do not + match, or if requester address is mcast */ + if (PREDICT_FALSE + (!ethernet_mac_address_equal (eth0->src_address, + arp0->ip4_over_ethernet[0]. + mac.bytes)) + || ethernet_address_cast (arp0->ip4_over_ethernet[0].mac.bytes)) + { + /* VRRP virtual MAC may be different to SMAC in ARP reply */ + if (!ethernet_mac_address_equal + (arp0->ip4_over_ethernet[0].mac.bytes, vrrp_prefix)) + { + error0 = ETHERNET_ARP_ERROR_l2_address_mismatch; + goto drop; + } + } + if (PREDICT_FALSE + (ip4_address_is_multicast (&arp0->ip4_over_ethernet[0].ip4))) + { + error0 = ETHERNET_ARP_ERROR_l3_src_address_not_local; + goto drop; + } + + /* Check if anyone want ARP request events for L2 BDs */ + l2_arp_term_publish_v4_dp (sw_if_index0, + &arp0->ip4_over_ethernet[0]); + + /* lookup BD mac_by_ip4 hash table for MAC entry */ + ip0 = arp0->ip4_over_ethernet[1].ip4.as_u32; + bd_index0 = vnet_buffer (p0)->l2.bd_index; + if (PREDICT_FALSE ((bd_index0 != last_bd_index) + || (last_bd_index == (u16) ~ 0))) + { + last_bd_index = bd_index0; + last_bd_config = vec_elt_at_index (l2im->bd_configs, bd_index0); + } + macp0 = (u8 *) hash_get (last_bd_config->mac_by_ip4, ip0); + + if (PREDICT_FALSE (!macp0)) + goto next_l2_feature; /* MAC not found */ + if (PREDICT_FALSE (arp0->ip4_over_ethernet[0].ip4.as_u32 == + arp0->ip4_over_ethernet[1].ip4.as_u32)) + goto next_l2_feature; /* GARP */ + + /* MAC found, send ARP reply - + Convert ARP request packet to ARP reply */ + arp0->opcode = clib_host_to_net_u16 (ETHERNET_ARP_OPCODE_reply); + arp0->ip4_over_ethernet[1] = arp0->ip4_over_ethernet[0]; + arp0->ip4_over_ethernet[0].ip4.as_u32 = ip0; + mac_address_from_bytes (&arp0->ip4_over_ethernet[0].mac, macp0); + clib_memcpy_fast (eth0->dst_address, eth0->src_address, 6); + clib_memcpy_fast (eth0->src_address, macp0, 6); + n_replies_sent += 1; + + output_response: + /* For BVI, need to use l2-fwd node to send ARP reply as + l2-output node cannot output packet to BVI properly */ + cfg0 = vec_elt_at_index (l2im->configs, sw_if_index0); + if (PREDICT_FALSE (cfg0->bvi)) + { + vnet_buffer (p0)->l2.feature_bitmap |= L2INPUT_FEAT_FWD; + vnet_buffer (p0)->sw_if_index[VLIB_RX] = 0; + goto next_l2_feature; + } + + /* Send ARP/ND reply back out input interface through l2-output */ + vnet_buffer (p0)->sw_if_index[VLIB_TX] = sw_if_index0; + next0 = ARP_TERM_NEXT_L2_OUTPUT; + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, pi0, + next0); + continue; + + check_ip6_nd: + /* IP6 ND event notification or solicitation handling to generate + local response instead of flooding */ + iph0 = (ip6_header_t *) l3h0; + if (PREDICT_FALSE (ethertype0 == ETHERNET_TYPE_IP6 && + iph0->protocol == IP_PROTOCOL_ICMP6 && + !ip6_address_is_unspecified + (&iph0->src_address))) + { + sw_if_index0 = vnet_buffer (p0)->sw_if_index[VLIB_RX]; + if (vnet_ip6_nd_term + (vm, node, p0, eth0, iph0, sw_if_index0, + vnet_buffer (p0)->l2.bd_index)) + goto output_response; + } + + next_l2_feature: + { + next0 = vnet_l2_feature_next (p0, arp_term_next_node_index, + L2INPUT_FEAT_ARP_TERM); + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, + pi0, next0); + continue; + } + + drop: + if (0 == arp0->ip4_over_ethernet[0].ip4.as_u32 || + (arp0->ip4_over_ethernet[0].ip4.as_u32 == + arp0->ip4_over_ethernet[1].ip4.as_u32)) + { + error0 = ETHERNET_ARP_ERROR_gratuitous_arp; + } + next0 = ARP_TERM_NEXT_DROP; + p0->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); + } + + vlib_error_count (vm, node->node_index, + ETHERNET_ARP_ERROR_replies_sent, n_replies_sent); + return frame->n_vectors; +} + +/* *INDENT-OFF* */ +VLIB_REGISTER_NODE (arp_term_l2bd_node, static) = { + .function = arp_term_l2bd, + .name = "arp-term-l2bd", + .vector_size = sizeof (u32), + .n_errors = ETHERNET_ARP_N_ERROR, + .error_strings = ethernet_arp_error_strings, + .n_next_nodes = ARP_TERM_N_NEXT, + .next_nodes = { + [ARP_TERM_NEXT_L2_OUTPUT] = "l2-output", + [ARP_TERM_NEXT_DROP] = "error-drop", + }, + .format_buffer = format_ethernet_arp_header, + .format_trace = format_arp_term_input_trace, +}; +/* *INDENT-ON* */ + +clib_error_t * +arp_term_init (vlib_main_t * vm) +{ + // Initialize the feature next-node indexes + feat_bitmap_init_next_nodes (vm, + arp_term_l2bd_node.index, + L2INPUT_N_FEAT, + l2input_get_feat_names (), + arp_term_next_node_index); + return 0; +} + +VLIB_INIT_FUNCTION (arp_term_init); + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/l2/l2_arp_term.h b/src/vnet/l2/l2_arp_term.h new file mode 100644 index 00000000000..5426d03af69 --- /dev/null +++ b/src/vnet/l2/l2_arp_term.h @@ -0,0 +1,54 @@ +/* + * l2/l2_arp_term.c: IP v4 ARP L2 BD termination + * + * Copyright (c) 2010 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include <vnet/ip/ip46_address.h> +#include <vnet/ethernet/arp_packet.h> + +typedef struct l2_arp_term_publish_ctx_t_ +{ + u32 sw_if_index; + ip46_type_t type; + ip46_address_t ip; + mac_address_t mac; +} l2_arp_term_publish_event_t; + +enum +{ + L2_ARP_TERM_EVENT_PUBLISH, +}; + +typedef struct l2_arp_term_main_t_ +{ + bool publish; + + l2_arp_term_publish_event_t *publish_events; + +} l2_arp_term_main_t; + +extern l2_arp_term_main_t l2_arp_term_main; +extern vlib_node_registration_t l2_arp_term_process_node; + +extern void l2_arp_term_set_publisher_node (bool on); + + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/l2/l2_bd.h b/src/vnet/l2/l2_bd.h index 360880839ec..35c0e0bf0a1 100644 --- a/src/vnet/l2/l2_bd.h +++ b/src/vnet/l2/l2_bd.h @@ -20,7 +20,7 @@ #include <vlib/vlib.h> #include <vnet/vnet.h> -#include <vnet/ip/ip6_packet.h> +#include <vnet/ip/ip46_address.h> #include <vnet/ethernet/mac_address.h> typedef enum l2_bd_port_type_t_ diff --git a/src/vnet/tcp/tcp.c b/src/vnet/tcp/tcp.c index 7dce97ea0c2..915675c11ed 100644 --- a/src/vnet/tcp/tcp.c +++ b/src/vnet/tcp/tcp.c @@ -23,9 +23,8 @@ #include <vnet/fib/fib.h> #include <vnet/dpo/load_balance.h> #include <vnet/dpo/receive_dpo.h> -#include <vnet/ip/ip6_neighbor.h> +#include <vnet/ip-neighbor/ip_neighbor.h> #include <math.h> -#include <vnet/ethernet/arp.h> tcp_main_t tcp_main; @@ -1863,7 +1862,6 @@ tcp_configure_v4_source_address_range (vlib_main_t * vm, ip4_address_t * start, ip4_address_t * end, u32 table_id) { - vnet_main_t *vnm = vnet_get_main (); u32 start_host_byte_order, end_host_byte_order; fib_prefix_t prefix; fib_node_index_t fei; @@ -1899,12 +1897,12 @@ tcp_configure_v4_source_address_range (vlib_main_t * vm, sw_if_index = fib_entry_get_resolving_interface (fei); /* Configure proxy arp across the range */ - rv = vnet_proxy_arp_add_del (start, end, fib_index, 0 /* is_del */ ); + rv = ip4_neighbor_proxy_add (fib_index, start, end); if (rv) return rv; - rv = vnet_proxy_arp_enable_disable (vnm, sw_if_index, 1); + rv = ip4_neighbor_proxy_enable (sw_if_index); if (rv) return rv; @@ -1989,7 +1987,7 @@ tcp_configure_v6_source_address_range (vlib_main_t * vm, return VNET_API_ERROR_NO_MATCHING_INTERFACE; /* Add a proxy neighbor discovery entry for this address */ - ip6_neighbor_proxy_add_del (sw_if_index, start, 0 /* is_del */ ); + ip6_neighbor_proxy_add (sw_if_index, start); /* Add a receive adjacency for this address */ receive_dpo_add_or_lock (DPO_PROTO_IP6, ~0 /* sw_if_index */ , diff --git a/src/vnet/vnet_all_api_h.h b/src/vnet/vnet_all_api_h.h index 7395d1c4756..a0bf8f72d14 100644 --- a/src/vnet/vnet_all_api_h.h +++ b/src/vnet/vnet_all_api_h.h @@ -44,7 +44,6 @@ #include <vnet/l2tp/l2tp.api.h> #include <vnet/span/span.api.h> #include <vnet/ip/ip.api.h> -#include <vnet/ip/rd_cp.api.h> #include <vnet/vxlan/vxlan.api.h> #include <vnet/geneve/geneve.api.h> #include <vnet/lldp/lldp.api.h> |