From 810086d8fd08445919ae03bf36161037e53a712a Mon Sep 17 00:00:00 2001 From: Neale Ranns Date: Sun, 5 Nov 2017 16:26:46 -0800 Subject: UDP Encapsulation. A UDP-encap object that particiapates in the FIB graph and contributes DPO to teh output chain. It thereofre resembles a tunnel but without the interface. FIB paths (and henace routes) can then be created to egress through the UDP-encap. Said routes can have MPLS labels, hence this also allows MPLSoUPD. Encap is uni-directional. For decap, one still registers with the UDP port dispatcher. Change-Id: I23bd345523b20789a1de1b02022ea1148ca50797 Signed-off-by: Neale Ranns --- src/vnet/udp/udp.api | 67 +++++ src/vnet/udp/udp_api.c | 193 +++++++++++++ src/vnet/udp/udp_encap.c | 617 ++++++++++++++++++++++++++++++++++++++++++ src/vnet/udp/udp_encap.h | 147 ++++++++++ src/vnet/udp/udp_encap_node.c | 280 +++++++++++++++++++ 5 files changed, 1304 insertions(+) create mode 100644 src/vnet/udp/udp.api create mode 100644 src/vnet/udp/udp_api.c create mode 100644 src/vnet/udp/udp_encap.c create mode 100644 src/vnet/udp/udp_encap.h create mode 100644 src/vnet/udp/udp_encap_node.c (limited to 'src/vnet/udp') diff --git a/src/vnet/udp/udp.api b/src/vnet/udp/udp.api new file mode 100644 index 00000000000..25ee96489a2 --- /dev/null +++ b/src/vnet/udp/udp.api @@ -0,0 +1,67 @@ +/* + * 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. + */ + +/** \file + + This file defines vpp UDP control-plane API messages which are generally + called through a shared memory interface. +*/ + +vl_api_version 1.0.0 + +/** \brief Add / del table request + A table can be added multiple times, but need be deleted only once. + @param client_index - opaque cookie to identify the sender + @param context - sender context, to match reply w/ request + @param is_ipv6 - V4 or V6 table + @param table_id - table ID associated with the encap destination +*/ +autoreply define udp_encap_add_del +{ + u32 client_index; + u32 context; + u32 id; + u32 table_id; + u8 is_ip6; + u8 is_add; + u16 src_port; + u16 dst_port; + u8 src_ip[16]; + u8 dst_ip[16]; +}; + +define udp_encap_dump +{ + u32 client_index; + u32 context; +}; + +define udp_encap_details +{ + u32 context; + u32 id; + u32 table_id; + u8 is_ip6; + u16 src_port; + u16 dst_port; + u8 src_ip[16]; + u8 dst_ip[16]; +}; + +/* + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/udp/udp_api.c b/src/vnet/udp/udp_api.c new file mode 100644 index 00000000000..e65235a5396 --- /dev/null +++ b/src/vnet/udp/udp_api.c @@ -0,0 +1,193 @@ +/* + * Copyright (c) 2017 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include +#include + +#include +#include + +#include + +#define vl_typedefs /* define message structures */ +#include +#undef vl_typedefs + +#define vl_endianfun /* define message structures */ +#include +#undef vl_endianfun + +/* instantiate all the print functions we know about */ +#define vl_print(handle, ...) vlib_cli_output (handle, __VA_ARGS__) +#define vl_printfun +#include +#undef vl_printfun + +#include + + +#define foreach_udp_api_msg \ + _(UDP_ENCAP_ADD_DEL, udp_encap_add_del) \ +_(UDP_ENCAP_DUMP, udp_encap_dump) + +static void +send_udp_encap_details (const udp_encap_t * ue, + unix_shared_memory_queue_t * q, u32 context) +{ + vl_api_udp_encap_details_t *mp; + fib_table_t *fib_table; + + mp = vl_msg_api_alloc (sizeof (*mp)); + memset (mp, 0, sizeof (*mp)); + mp->_vl_msg_id = ntohs (VL_API_UDP_ENCAP_DETAILS); + mp->context = context; + + mp->is_ip6 = (ue->ue_ip_proto == FIB_PROTOCOL_IP6); + + if (FIB_PROTOCOL_IP4 == ue->ue_ip_proto) + { + clib_memcpy (mp->src_ip, &ue->ue_hdrs.ip4.ue_ip4.src_address, 4); + clib_memcpy (mp->dst_ip, &ue->ue_hdrs.ip4.ue_ip4.dst_address, 4); + mp->src_port = htons (ue->ue_hdrs.ip4.ue_udp.src_port); + mp->dst_port = htons (ue->ue_hdrs.ip4.ue_udp.dst_port); + } + else + { + clib_memcpy (mp->src_ip, &ue->ue_hdrs.ip6.ue_ip6.src_address, 16); + clib_memcpy (mp->dst_ip, &ue->ue_hdrs.ip6.ue_ip6.dst_address, 16); + mp->src_port = htons (ue->ue_hdrs.ip6.ue_udp.src_port); + mp->dst_port = htons (ue->ue_hdrs.ip6.ue_udp.dst_port); + } + + fib_table = fib_table_get (ue->ue_fib_index, ue->ue_ip_proto); + mp->table_id = htonl (fib_table->ft_table_id); + mp->id = htonl (ue->ue_id); + + vl_msg_api_send_shmem (q, (u8 *) & mp); +} + +static void +vl_api_udp_encap_dump_t_handler (vl_api_udp_encap_dump_t * mp, + vlib_main_t * vm) +{ + unix_shared_memory_queue_t *q; + udp_encap_t *ue; + + q = vl_api_client_index_to_input_queue (mp->client_index); + if (q == 0) + return; + + /* *INDENT-OFF* */ + pool_foreach(ue, udp_encap_pool, + ({ + send_udp_encap_details(ue, q, mp->context); + })); + /* *INDENT-OFF* */ +} + +static void +vl_api_udp_encap_add_del_t_handler (vl_api_udp_encap_add_del_t * mp, + vlib_main_t * vm) +{ + vl_api_udp_encap_add_del_reply_t *rmp; + ip46_address_t src_ip, dst_ip; + u32 fib_index, table_id, ue_id; + fib_protocol_t fproto; + int rv = 0; + + ue_id = ntohl(mp->id); + table_id = ntohl(mp->table_id); + fproto = (mp->is_ip6 ? FIB_PROTOCOL_IP6 : FIB_PROTOCOL_IP4); + + fib_index = fib_table_find(fproto, table_id); + + if (~0 == fib_index) + { + rv = VNET_API_ERROR_NO_SUCH_TABLE; + goto done; + } + + if (FIB_PROTOCOL_IP4 == fproto) + { + clib_memcpy(&src_ip.ip4, mp->src_ip, 4); + clib_memcpy(&dst_ip.ip4, mp->dst_ip, 4); + } + else + { + clib_memcpy(&src_ip.ip6, mp->src_ip, 16); + clib_memcpy(&dst_ip.ip6, mp->dst_ip, 16); + } + + if (mp->is_add) + { + udp_encap_add_and_lock(ue_id, fproto, fib_index, + &src_ip, &dst_ip, + ntohs(mp->src_port), + ntohs(mp->dst_port), + UDP_ENCAP_FIXUP_NONE); + } + else + { + udp_encap_unlock(ue_id); + } + + done: + REPLY_MACRO (VL_API_UDP_ENCAP_ADD_DEL_REPLY); +} + +#define vl_msg_name_crc_list +#include +#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_udp; +#undef _ +} + +static clib_error_t * +udp_api_hookup (vlib_main_t * vm) +{ + api_main_t *am = &api_main; + +#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), 1); + foreach_udp_api_msg; +#undef _ + + /* + * Set up the (msg_name, crc, message-id) table + */ + setup_message_id_table (am); + + return 0; +} + +VLIB_API_INIT_FUNCTION (udp_api_hookup); + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ diff --git a/src/vnet/udp/udp_encap.c b/src/vnet/udp/udp_encap.c new file mode 100644 index 00000000000..98b824ba866 --- /dev/null +++ b/src/vnet/udp/udp_encap.c @@ -0,0 +1,617 @@ +/* + * Copyright (c) 2017 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include +#include +#include +#include + +/** + * Registered DPO types for the IP header encapsulated, v4 or v6. + */ +dpo_type_t udp_encap_dpo_types[FIB_PROTOCOL_MAX]; + +/** + * Hash DB to map from client ID to VPP index. + */ +uword *udp_encap_db; + +/** + * Pool of encaps + */ +udp_encap_t *udp_encap_pool; + +static udp_encap_t * +udp_encap_get_w_id (u32 id) +{ + udp_encap_t *ue = NULL; + index_t uei; + + uei = udp_encap_find (id); + + if (INDEX_INVALID != uei) + { + ue = udp_encap_get (uei); + } + + return (ue); +} + +static void +udp_encap_restack (udp_encap_t * ue) +{ + dpo_stack (udp_encap_dpo_types[ue->ue_ip_proto], + fib_proto_to_dpo (ue->ue_ip_proto), + &ue->ue_dpo, + fib_entry_contribute_ip_forwarding (ue->ue_fib_entry_index)); +} + +index_t +udp_encap_add_and_lock (u32 id, + fib_protocol_t proto, + index_t fib_index, + const ip46_address_t * src_ip, + const ip46_address_t * dst_ip, + u16 src_port, + u16 dst_port, udp_encap_fixup_flags_t flags) +{ + udp_encap_t *ue; + index_t uei; + + uei = udp_encap_find (id); + + if (INDEX_INVALID == uei) + { + u8 pfx_len = 0; + + pool_get (udp_encap_pool, ue); + uei = ue - udp_encap_pool; + + hash_set (udp_encap_db, id, uei); + + fib_node_init (&ue->ue_fib_node, FIB_NODE_TYPE_UDP_ENCAP); + fib_node_lock (&ue->ue_fib_node); + ue->ue_fib_index = fib_index; + ue->ue_flags = flags; + ue->ue_id = id; + ue->ue_ip_proto = proto; + + switch (proto) + { + case FIB_PROTOCOL_IP4: + pfx_len = 32; + ue->ue_hdrs.ip4.ue_ip4.ip_version_and_header_length = 0x45; + ue->ue_hdrs.ip4.ue_ip4.ttl = 254; + ue->ue_hdrs.ip4.ue_ip4.protocol = IP_PROTOCOL_UDP; + ue->ue_hdrs.ip4.ue_ip4.src_address.as_u32 = src_ip->ip4.as_u32; + ue->ue_hdrs.ip4.ue_ip4.dst_address.as_u32 = dst_ip->ip4.as_u32; + ue->ue_hdrs.ip4.ue_ip4.checksum = + ip4_header_checksum (&ue->ue_hdrs.ip4.ue_ip4); + ue->ue_hdrs.ip4.ue_udp.src_port = clib_host_to_net_u16 (src_port); + ue->ue_hdrs.ip4.ue_udp.dst_port = clib_host_to_net_u16 (dst_port); + + break; + case FIB_PROTOCOL_IP6: + pfx_len = 128; + ue->ue_hdrs.ip6.ue_ip6.ip_version_traffic_class_and_flow_label = + clib_host_to_net_u32 (6 << 28); + ue->ue_hdrs.ip6.ue_ip6.hop_limit = 255; + ue->ue_hdrs.ip6.ue_ip6.protocol = IP_PROTOCOL_UDP; + ue->ue_hdrs.ip6.ue_ip6.src_address.as_u64[0] = + src_ip->ip6.as_u64[0]; + ue->ue_hdrs.ip6.ue_ip6.src_address.as_u64[1] = + src_ip->ip6.as_u64[1]; + ue->ue_hdrs.ip6.ue_ip6.dst_address.as_u64[0] = + dst_ip->ip6.as_u64[0]; + ue->ue_hdrs.ip6.ue_ip6.dst_address.as_u64[1] = + dst_ip->ip6.as_u64[1]; + ue->ue_hdrs.ip6.ue_udp.src_port = clib_host_to_net_u16 (src_port); + ue->ue_hdrs.ip6.ue_udp.dst_port = clib_host_to_net_u16 (dst_port); + + break; + default: + ASSERT (0); + } + + /* + * track the destination address + */ + fib_prefix_t dst_pfx = { + .fp_proto = proto, + .fp_len = pfx_len, + .fp_addr = *dst_ip, + }; + + ue->ue_fib_entry_index = + fib_table_entry_special_add (fib_index, + &dst_pfx, + FIB_SOURCE_RR, FIB_ENTRY_FLAG_NONE); + ue->ue_fib_sibling = + fib_entry_child_add (ue->ue_fib_entry_index, + FIB_NODE_TYPE_UDP_ENCAP, uei); + + udp_encap_restack (ue); + } + else + { + /* + * existing entry. updates not supported yet + */ + uei = INDEX_INVALID; + } + return (uei); +} + +void +udp_encap_contribute_forwarding (u32 id, dpo_proto_t proto, dpo_id_t * dpo) +{ + index_t uei; + + uei = udp_encap_find (id); + + if (INDEX_INVALID == uei) + { + dpo_copy (dpo, drop_dpo_get (proto)); + } + else + { + udp_encap_t *ue; + + ue = udp_encap_get (uei); + + dpo_set (dpo, udp_encap_dpo_types[ue->ue_ip_proto], proto, uei); + } +} + +index_t +udp_encap_find (u32 id) +{ + uword *p; + + p = hash_get (udp_encap_db, id); + + if (NULL != p) + return p[0]; + + return INDEX_INVALID; +} + +void +udp_encap_lock (u32 id) +{ + udp_encap_t *ue; + + ue = udp_encap_get_w_id (id); + + if (NULL != ue) + { + fib_node_lock (&ue->ue_fib_node); + } +} + +void +udp_encap_unlock_w_index (index_t uei) +{ + udp_encap_t *ue; + + if (INDEX_INVALID == uei) + { + return; + } + + ue = udp_encap_get (uei); + + if (NULL != ue) + { + fib_node_unlock (&ue->ue_fib_node); + } +} + +void +udp_encap_unlock (u32 id) +{ + udp_encap_t *ue; + + ue = udp_encap_get_w_id (id); + + if (NULL != ue) + { + fib_node_unlock (&ue->ue_fib_node); + } +} + +static void +udp_encap_dpo_lock (dpo_id_t * dpo) +{ + udp_encap_t *ue; + + ue = udp_encap_get (dpo->dpoi_index); + + fib_node_lock (&ue->ue_fib_node); +} + +static void +udp_encap_dpo_unlock (dpo_id_t * dpo) +{ + udp_encap_t *ue; + + ue = udp_encap_get (dpo->dpoi_index); + + fib_node_unlock (&ue->ue_fib_node); +} + +static u8 * +format_udp_encap_i (u8 * s, va_list * args) +{ + index_t uei = va_arg (*args, index_t); + u32 indent = va_arg (*args, u32); + u32 details = va_arg (*args, u32); + udp_encap_t *ue; + + ue = udp_encap_get (uei); + + // FIXME + s = format (s, "udp-ecap:[%d]: id:%d ip-fib-index:%d", + uei, ue->ue_id, ue->ue_fib_index); + if (FIB_PROTOCOL_IP4 == ue->ue_ip_proto) + { + s = format (s, "ip:[src:%U, dst:%U] udp:[src:%d, dst:%d]", + format_ip4_address, + &ue->ue_hdrs.ip4.ue_ip4.src_address, + format_ip4_address, + &ue->ue_hdrs.ip4.ue_ip4.dst_address, + clib_net_to_host_u16 (ue->ue_hdrs.ip4.ue_udp.src_port), + clib_net_to_host_u16 (ue->ue_hdrs.ip4.ue_udp.dst_port)); + } + else + { + s = format (s, "ip:[src:%U, dst:%U] udp:[src:%d dst:%d]", + format_ip6_address, + &ue->ue_hdrs.ip6.ue_ip6.src_address, + format_ip6_address, + &ue->ue_hdrs.ip6.ue_ip6.dst_address, + clib_net_to_host_u16 (ue->ue_hdrs.ip6.ue_udp.src_port), + clib_net_to_host_u16 (ue->ue_hdrs.ip6.ue_udp.dst_port)); + } + if (details) + { + s = format (s, " locks:%d", ue->ue_fib_node.fn_locks); + s = format (s, "\n%UStacked on:", format_white_space, indent + 1); + s = format (s, "\n%U%U", + format_white_space, indent + 2, + format_dpo_id, &ue->ue_dpo, indent + 3); + } + return (s); +} + +static u8 * +format_udp_encap_dpo (u8 * s, va_list * args) +{ + index_t uei = va_arg (*args, index_t); + u32 indent = va_arg (*args, u32); + + return (format (s, "%U", format_udp_encap_i, uei, indent, 1)); +} + +u8 * +format_udp_encap (u8 * s, va_list * args) +{ + u32 id = va_arg (*args, u32); + u32 details = va_arg (*args, u32); + index_t uei; + + uei = udp_encap_find (id); + + if (INDEX_INVALID == uei) + { + return (format (s, "Invalid udp-encap ID: %d", id)); + } + + return (format (s, "%U", format_udp_encap_i, uei, 0, details)); +} + +static udp_encap_t * +udp_encap_from_fib_node (fib_node_t * node) +{ +#if (CLIB_DEBUG > 0) + ASSERT (FIB_NODE_TYPE_UDP_ENCAP == node->fn_type); +#endif + return ((udp_encap_t *) (((char *) node) - + STRUCT_OFFSET_OF (udp_encap_t, ue_fib_node))); +} + +/** + * Function definition to backwalk a FIB node + */ +static fib_node_back_walk_rc_t +udp_encap_fib_back_walk (fib_node_t * node, fib_node_back_walk_ctx_t * ctx) +{ + udp_encap_restack (udp_encap_from_fib_node (node)); + + return (FIB_NODE_BACK_WALK_CONTINUE); +} + +/** + * Function definition to get a FIB node from its index + */ +static fib_node_t * +udp_encap_fib_node_get (fib_node_index_t index) +{ + udp_encap_t *ue; + + ue = pool_elt_at_index (udp_encap_pool, index); + + return (&ue->ue_fib_node); +} + +/** + * Function definition to inform the FIB node that its last lock has gone. + */ +static void +udp_encap_fib_last_lock_gone (fib_node_t * node) +{ + udp_encap_t *ue; + + ue = udp_encap_from_fib_node (node); + + /** + * reset the stacked DPO to unlock it + */ + dpo_reset (&ue->ue_dpo); + hash_unset (udp_encap_db, ue->ue_id); + + fib_entry_child_remove (ue->ue_fib_entry_index, ue->ue_fib_sibling); + fib_table_entry_delete_index (ue->ue_fib_entry_index, FIB_SOURCE_RR); + + + pool_put (udp_encap_pool, ue); +} + +const static char *const udp4_encap_ip4_nodes[] = { + "udp4-encap", + NULL, +}; + +const static char *const udp4_encap_ip6_nodes[] = { + "udp4-encap", + NULL, +}; + +const static char *const udp4_encap_mpls_nodes[] = { + "udp4-encap", + NULL, +}; + +const static char *const udp6_encap_ip4_nodes[] = { + "udp6-encap", + NULL, +}; + +const static char *const udp6_encap_ip6_nodes[] = { + "udp6-encap", + NULL, +}; + +const static char *const udp6_encap_mpls_nodes[] = { + "udp6-encap", + NULL, +}; + +const static char *const *const udp4_encap_nodes[DPO_PROTO_NUM] = { + [DPO_PROTO_IP4] = udp4_encap_ip4_nodes, + [DPO_PROTO_IP6] = udp4_encap_ip6_nodes, + [DPO_PROTO_MPLS] = udp4_encap_mpls_nodes, +}; + +const static char *const *const udp6_encap_nodes[DPO_PROTO_NUM] = { + [DPO_PROTO_IP4] = udp6_encap_ip4_nodes, + [DPO_PROTO_IP6] = udp6_encap_ip6_nodes, + [DPO_PROTO_MPLS] = udp6_encap_mpls_nodes, +}; + +/* + * Virtual function table registered by UDP encaps + * for participation in the FIB object graph. + */ +const static fib_node_vft_t udp_encap_fib_vft = { + .fnv_get = udp_encap_fib_node_get, + .fnv_last_lock = udp_encap_fib_last_lock_gone, + .fnv_back_walk = udp_encap_fib_back_walk, +}; + +const static dpo_vft_t udp_encap_dpo_vft = { + .dv_lock = udp_encap_dpo_lock, + .dv_unlock = udp_encap_dpo_unlock, + .dv_format = format_udp_encap_dpo, + //.dv_mem_show = replicate_mem_show, +}; + +clib_error_t * +udp_encap_init (vlib_main_t * vm) +{ + udp_encap_db = hash_create (0, sizeof (index_t)); + + fib_node_register_type (FIB_NODE_TYPE_UDP_ENCAP, &udp_encap_fib_vft); + + udp_encap_dpo_types[FIB_PROTOCOL_IP4] = + dpo_register_new_type (&udp_encap_dpo_vft, udp4_encap_nodes); + udp_encap_dpo_types[FIB_PROTOCOL_IP6] = + dpo_register_new_type (&udp_encap_dpo_vft, udp6_encap_nodes); + + return (NULL); +} + +VLIB_INIT_FUNCTION (udp_encap_init); + +clib_error_t * +udp_encap_cli (vlib_main_t * vm, + unformat_input_t * main_input, vlib_cli_command_t * cmd) +{ + unformat_input_t _line_input, *line_input = &_line_input; + clib_error_t *error = NULL; + ip46_address_t src_ip, dst_ip; + u32 table_id, ue_id; + u32 src_port, dst_port; + udp_encap_fixup_flags_t flags; + fib_protocol_t fproto; + u8 is_del; + + is_del = 0; + table_id = 0; + flags = UDP_ENCAP_FIXUP_NONE; + fproto = FIB_PROTOCOL_MAX; + dst_port = 0; + ue_id = ~0; + + /* Get a line of input. */ + if (!unformat_user (main_input, unformat_line_input, line_input)) + return 0; + + while (unformat_check_input (line_input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (line_input, "id %d", &ue_id)) + ; + else if (unformat (line_input, "add")) + is_del = 0; + else if (unformat (line_input, "del")) + is_del = 1; + else if (unformat (line_input, "%U %U", + unformat_ip4_address, + &src_ip.ip4, unformat_ip4_address, &dst_ip.ip4)) + fproto = FIB_PROTOCOL_IP4; + else if (unformat (line_input, "%U %U", + unformat_ip6_address, + &src_ip.ip6, unformat_ip6_address, &dst_ip.ip6)) + fproto = FIB_PROTOCOL_IP6; + else if (unformat (line_input, "%d %d", &src_port, &dst_port)) + ; + else if (unformat (line_input, "%d", &dst_port)) + ; + else if (unformat (line_input, "table-id %d", &table_id)) + ; + else if (unformat (line_input, "src-port-is-entropy")) + flags |= UDP_ENCAP_FIXUP_UDP_SRC_PORT_ENTROPY; + else + { + error = unformat_parse_error (line_input); + goto done; + } + } + + if (~0 == ue_id) + { + error = + clib_error_return (0, "An ID for the UDP encap instance is required"); + goto done; + } + + if (!is_del && fproto != FIB_PROTOCOL_MAX) + { + u32 fib_index; + index_t uei; + + fib_index = fib_table_find (fproto, table_id); + + if (~0 == fib_index) + { + error = clib_error_return (0, "Nonexistent table id %d", table_id); + goto done; + } + + uei = udp_encap_add_and_lock (ue_id, fproto, fib_index, + &src_ip, &dst_ip, + src_port, dst_port, flags); + + if (INDEX_INVALID == uei) + { + error = + clib_error_return (0, "update to existing encap not supported %d", + ue_id); + goto done; + } + } + else if (is_del) + { + udp_encap_unlock (ue_id); + } + else + { + error = + clib_error_return (0, + "Some IP addresses would be usefull, don't you think?", + ue_id); + } + +done: + unformat_free (line_input); + return error; +} + +clib_error_t * +udp_encap_show (vlib_main_t * vm, + unformat_input_t * input, vlib_cli_command_t * cmd) +{ + u32 ue_id; + + ue_id = ~0; + + /* Get a line of input. */ + while (unformat_check_input (input) != UNFORMAT_END_OF_INPUT) + { + if (unformat (input, "%d", &ue_id)) + ; + } + + if (~0 == ue_id) + { + udp_encap_t *ue; + + /* *INDENT-OFF* */ + pool_foreach(ue, udp_encap_pool, + ({ + vlib_cli_output(vm, "%U", format_udp_encap, ue->ue_id, 0); + })); + /* *INDENT-ON* */ + } + else + { + vlib_cli_output (vm, "%U", format_udp_encap, ue_id, 1); + } + + return NULL; +} + +/* *INDENT-OFF* */ +VLIB_CLI_COMMAND (udp_encap_add_command, static) = { + .path = "udp encap", + .short_help = "udp encap [add|del] [] [src-port-is-entropy] [table-id ]", + .function = udp_encap_cli, + .is_mp_safe = 1, +}; +VLIB_CLI_COMMAND (udp_encap_show_command, static) = { + .path = "show udp encap", + .short_help = "show udp encap [ID]", + .function = udp_encap_show, + .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/udp/udp_encap.h b/src/vnet/udp/udp_encap.h new file mode 100644 index 00000000000..b8f329dcbee --- /dev/null +++ b/src/vnet/udp/udp_encap.h @@ -0,0 +1,147 @@ +/* + * 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 __UDP_ENCAP_H__ +#define __UDP_ENCAP_H__ + +#include +#include +#include + +/** + * UDP encapsualtion. + * A representation of the encapsulation of packets in UDP-over-IP. + * This is encapsulation only, there is no tunnel interface, hence + * it is uni-directional. For decap register a handler with the UDP port + * dispatcher. + */ + +/** + * Fixup behaviour. Actions performed on the encap in the data-plance + */ +typedef enum udp_encap_fixup_flags_t_ +{ + UDP_ENCAP_FIXUP_NONE = 0, + /** + * UDP source port contains an entropy/hash value for load-balancing by downstream peers. + */ + UDP_ENCAP_FIXUP_UDP_SRC_PORT_ENTROPY = (1 << 0), +} udp_encap_fixup_flags_t; + +/** + * The UDP encap represenation + */ +typedef struct udp_encap_t_ +{ + /** + * The first cacheline contains the data used in the data-plane + */ + CLIB_CACHE_LINE_ALIGN_MARK (cacheline0); + + /** + * The headers to paint, in packet painting order + */ + union + { + struct + { + ip4_header_t ue_ip4; + udp_header_t ue_udp; + } __attribute__ ((packed)) ip4; + struct + { + ip6_header_t ue_ip6; + udp_header_t ue_udp; + } __attribute__ ((packed)) ip6; + } __attribute__ ((packed)) ue_hdrs; + + /** + * Flags controlling fixup behaviour + */ + udp_encap_fixup_flags_t ue_flags; + + /** + * The DPO used to forward to the next node in the VLIB graph + */ + dpo_id_t ue_dpo; + + /** + * the protocol of the IP header imposed + */ + fib_protocol_t ue_ip_proto; + + CLIB_CACHE_LINE_ALIGN_MARK (cacheline1); + + /** + * linkage into the FIB graph + */ + fib_node_t ue_fib_node; + + /** + * The ID given by the user/client. + * This ID is used by the client for modifications. + */ + u32 ue_id; + + /** + * Tracking information for the IP destination + */ + fib_node_index_t ue_fib_entry_index; + u32 ue_fib_sibling; + + /** + * The FIB index in which the encap destination resides + */ + index_t ue_fib_index; +} udp_encap_t; + +extern index_t udp_encap_add_and_lock (u32 id, + fib_protocol_t proto, + index_t fib_index, + const ip46_address_t * src_ip, + const ip46_address_t * dst_ip, + u16 src_port, + u16 dst_port, + udp_encap_fixup_flags_t flags); + +extern index_t udp_encap_find (u32 id); +extern void udp_encap_lock (u32 id); +extern void udp_encap_unlock (u32 id); +extern u8 *format_udp_encap (u8 * s, va_list * args); +extern void udp_encap_unlock_w_index (index_t uei); +extern void udp_encap_contribute_forwarding (u32 id, + dpo_proto_t proto, + dpo_id_t * dpo); + +/** + * Pool of encaps + */ +extern udp_encap_t *udp_encap_pool; + +static inline udp_encap_t * +udp_encap_get (index_t uei) +{ + return (pool_elt_at_index (udp_encap_pool, uei)); +} + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ + +#endif diff --git a/src/vnet/udp/udp_encap_node.c b/src/vnet/udp/udp_encap_node.c new file mode 100644 index 00000000000..09a76b530f6 --- /dev/null +++ b/src/vnet/udp/udp_encap_node.c @@ -0,0 +1,280 @@ +/* + * Copyright (c) 2017 Cisco and/or its affiliates. + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at: + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +#include + +typedef struct udp4_encap_trace_t_ +{ + udp_header_t udp; + ip4_header_t ip; +} udp4_encap_trace_t; + +typedef struct udp6_encap_trace_t_ +{ + udp_header_t udp; + ip6_header_t ip; +} udp6_encap_trace_t; + +static u8 * +format_udp4_encap_trace (u8 * s, va_list * args) +{ + CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *); + CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *); + udp4_encap_trace_t *t; + + t = va_arg (*args, udp4_encap_trace_t *); + + s = format (s, "%U\n %U", + format_ip4_header, &t->ip, sizeof (t->ip), + format_udp_header, &t->udp, sizeof (t->udp)); + return (s); +} + +static u8 * +format_udp6_encap_trace (u8 * s, va_list * args) +{ + CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *); + CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *); + udp6_encap_trace_t *t; + + t = va_arg (*args, udp6_encap_trace_t *); + + s = format (s, "%U\n %U", + format_ip6_header, &t->ip, sizeof (t->ip), + format_udp_header, &t->udp, sizeof (t->udp)); + return (s); +} + +always_inline uword +udp_encap_inline (vlib_main_t * vm, + vlib_node_runtime_t * node, + vlib_frame_t * frame, int is_encap_v6) +{ + u32 *from = vlib_frame_vector_args (frame); + u32 n_left_from, n_left_to_next, *to_next, next_index; + + n_left_from = frame->n_vectors; + next_index = node->cached_next_index; + + while (n_left_from > 0) + { + vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); + + while (n_left_from >= 4 && n_left_to_next >= 2) + { + vlib_buffer_t *b0, *b1; + udp_encap_t *ue0, *ue1; + u32 bi0, next0, uei0; + u32 bi1, next1, uei1; + + /* Prefetch next iteration. */ + { + vlib_buffer_t *p2, *p3; + + p2 = vlib_get_buffer (vm, from[2]); + p3 = vlib_get_buffer (vm, from[3]); + + vlib_prefetch_buffer_header (p2, STORE); + vlib_prefetch_buffer_header (p3, STORE); + } + + bi0 = to_next[0] = from[0]; + bi1 = to_next[1] = from[1]; + + from += 2; + n_left_from -= 2; + to_next += 2; + n_left_to_next -= 2; + + b0 = vlib_get_buffer (vm, bi0); + b1 = vlib_get_buffer (vm, bi1); + + uei0 = vnet_buffer (b0)->ip.adj_index[VLIB_TX]; + uei1 = vnet_buffer (b1)->ip.adj_index[VLIB_TX]; + + /* Rewrite packet header and updates lengths. */ + ue0 = udp_encap_get (uei0); + ue1 = udp_encap_get (uei1); + + /* Paint */ + if (is_encap_v6) + { + const u8 n_bytes = + sizeof (udp_header_t) + sizeof (ip6_header_t); + ip_udp_encap_two (vm, b0, b1, (u8 *) & ue0->ue_hdrs, + (u8 *) & ue1->ue_hdrs, n_bytes, 0); + if (PREDICT_FALSE (b0->flags & VLIB_BUFFER_IS_TRACED)) + { + udp6_encap_trace_t *tr = + vlib_add_trace (vm, node, b0, sizeof (*tr)); + tr->udp = ue0->ue_hdrs.ip6.ue_udp; + tr->ip = ue0->ue_hdrs.ip6.ue_ip6; + } + if (PREDICT_FALSE (b1->flags & VLIB_BUFFER_IS_TRACED)) + { + udp6_encap_trace_t *tr = + vlib_add_trace (vm, node, b1, sizeof (*tr)); + tr->udp = ue1->ue_hdrs.ip6.ue_udp; + tr->ip = ue1->ue_hdrs.ip6.ue_ip6; + } + } + else + { + const u8 n_bytes = + sizeof (udp_header_t) + sizeof (ip4_header_t); + + ip_udp_encap_two (vm, b0, b1, + (u8 *) & ue0->ue_hdrs, + (u8 *) & ue1->ue_hdrs, n_bytes, 1); + + if (PREDICT_FALSE (b0->flags & VLIB_BUFFER_IS_TRACED)) + { + udp4_encap_trace_t *tr = + vlib_add_trace (vm, node, b0, sizeof (*tr)); + tr->udp = ue0->ue_hdrs.ip4.ue_udp; + tr->ip = ue0->ue_hdrs.ip4.ue_ip4; + } + if (PREDICT_FALSE (b1->flags & VLIB_BUFFER_IS_TRACED)) + { + udp4_encap_trace_t *tr = + vlib_add_trace (vm, node, b1, sizeof (*tr)); + tr->udp = ue1->ue_hdrs.ip4.ue_udp; + tr->ip = ue1->ue_hdrs.ip4.ue_ip4; + } + } + + next0 = ue0->ue_dpo.dpoi_next_node; + next1 = ue1->ue_dpo.dpoi_next_node; + vnet_buffer (b0)->ip.adj_index[VLIB_TX] = ue0->ue_dpo.dpoi_index; + vnet_buffer (b1)->ip.adj_index[VLIB_TX] = ue1->ue_dpo.dpoi_index; + + vlib_validate_buffer_enqueue_x2 (vm, node, next_index, + to_next, n_left_to_next, + bi0, bi1, next0, next1); + } + + while (n_left_from > 0 && n_left_to_next > 0) + { + u32 bi0, next0, uei0; + vlib_buffer_t *b0; + udp_encap_t *ue0; + + bi0 = to_next[0] = from[0]; + + from += 1; + n_left_from -= 1; + to_next += 1; + n_left_to_next -= 1; + + b0 = vlib_get_buffer (vm, bi0); + + uei0 = vnet_buffer (b0)->ip.adj_index[VLIB_TX]; + + /* Rewrite packet header and updates lengths. */ + ue0 = udp_encap_get (uei0); + + /* Paint */ + if (is_encap_v6) + { + const u8 n_bytes = + sizeof (udp_header_t) + sizeof (ip6_header_t); + ip_udp_encap_one (vm, b0, (u8 *) & ue0->ue_hdrs.ip6, n_bytes, + 0); + + if (PREDICT_FALSE (b0->flags & VLIB_BUFFER_IS_TRACED)) + { + udp6_encap_trace_t *tr = + vlib_add_trace (vm, node, b0, sizeof (*tr)); + tr->udp = ue0->ue_hdrs.ip6.ue_udp; + tr->ip = ue0->ue_hdrs.ip6.ue_ip6; + } + } + else + { + const u8 n_bytes = + sizeof (udp_header_t) + sizeof (ip4_header_t); + + ip_udp_encap_one (vm, b0, (u8 *) & ue0->ue_hdrs.ip4, n_bytes, + 1); + + if (PREDICT_FALSE (b0->flags & VLIB_BUFFER_IS_TRACED)) + { + udp4_encap_trace_t *tr = + vlib_add_trace (vm, node, b0, sizeof (*tr)); + tr->udp = ue0->ue_hdrs.ip4.ue_udp; + tr->ip = ue0->ue_hdrs.ip4.ue_ip4; + } + } + + next0 = ue0->ue_dpo.dpoi_next_node; + vnet_buffer (b0)->ip.adj_index[VLIB_TX] = ue0->ue_dpo.dpoi_index; + + vlib_validate_buffer_enqueue_x1 (vm, node, next_index, + to_next, n_left_to_next, + bi0, next0); + } + + vlib_put_next_frame (vm, node, next_index, n_left_to_next); + } + + return frame->n_vectors; +} + +static uword +udp4_encap (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + return udp_encap_inline (vm, node, frame, 0); +} + +static uword +udp6_encap (vlib_main_t * vm, + vlib_node_runtime_t * node, vlib_frame_t * frame) +{ + return udp_encap_inline (vm, node, frame, 1); +} + +/* *INDENT-OFF* */ +VLIB_REGISTER_NODE (udp4_encap_node) = { + .function = udp4_encap, + .name = "udp4-encap", + .vector_size = sizeof (u32), + + .format_trace = format_udp4_encap_trace, + + .n_next_nodes = 0, +}; +VLIB_NODE_FUNCTION_MULTIARCH (udp4_encap_node, udp4_encap); + +VLIB_REGISTER_NODE (udp6_encap_node) = { + .function = udp6_encap, + .name = "udp6-encap", + .vector_size = sizeof (u32), + + .format_trace = format_udp6_encap_trace, + + .n_next_nodes = 0, +}; +VLIB_NODE_FUNCTION_MULTIARCH (udp6_encap_node, udp6_encap); +/* *INDENT-ON* */ + + +/* + * fd.io coding-style-patch-verification: ON + * + * Local Variables: + * eval: (c-set-style "gnu") + * End: + */ -- cgit 1.2.3-korg