vpp - Vector Packet Processing

Age	Commit message (Collapse)	Author	Files	Lines
2017-10-15	Revert "Enforce FIB table creation before use"	Florin Coras	1	-0/+2
	This reverts commit f9342023c19887da656133e2688a90d70383b0c5. Reverting to unblock master. No idea why jjb +1ed this patch! On closer inspection it looks like it -1ed it and subsequently changed opinion. CSIT tests should be fixed before re-merging. Change-Id: I26608912a962c52083073e16c7c9d2cc44a3cc8d Signed-off-by: Florin Coras <fcoras@cisco.com>
2017-10-13	Enforce FIB table creation before use	Neale Ranns	1	-2/+0
	last i the serise of the use of the FIB table create/delete API. VPP now forces the tables to have been explicitly creted before they are used. Change-Id: Ifde3b1bbb76697a01ab71bce4f5264e6d1725467 Signed-off-by: Neale Ranns <nranns@cisco.com>
2017-10-05	Distributed Virtual Router Support	Neale Ranns	1	-0/+178
	A distributed virtual router works by attmpeting to switch a packet, but on failing to find a local consumer (i.e. the packet is destined to a locally attached host) then the packet is sent unmodified 'upstream' to where the rest of the 'distributed' router is present. When L3 switching a packet this means the L2 header must not be modifed. This patch adds a 'l2-bridge' object to the L3 FIB which re-injects packets from the L3 path back into the L2 path - use with extreme caution. Change-Id: I069724eb45956647d7980cbe40a80a788ee6ee82 Signed-off-by: Neale Ranns <nranns@cisco.com>

/* * 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. */ /* * node.c: srp packet processing * * Copyright (c) 2011 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 <vlib/vlib.h> #include <vnet/ip/ip_packet.h> /* for ip_csum_fold */ #include <vnet/srp/srp.h> srp_main_t srp_main; typedef struct { u8 packet_data[32]; } srp_input_trace_t; static u8 * format_srp_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 *); srp_input_trace_t * t = va_arg (*va, srp_input_trace_t *); s = format (s, "%U", format_srp_header, t->packet_data); return s; } typedef enum { SRP_INPUT_NEXT_ERROR, SRP_INPUT_NEXT_ETHERNET_INPUT, SRP_INPUT_NEXT_CONTROL, SRP_INPUT_N_NEXT, } srp_input_next_t; typedef struct { u8 next_index; u8 buffer_advance; u16 error; } srp_input_disposition_t; static srp_input_disposition_t srp_input_disposition_by_mode[8] = { [SRP_MODE_reserved0] = { .next_index = SRP_INPUT_NEXT_ERROR, .error = SRP_ERROR_UNKNOWN_MODE, }, [SRP_MODE_reserved1] = { .next_index = SRP_INPUT_NEXT_ERROR, .error = SRP_ERROR_UNKNOWN_MODE, }, [SRP_MODE_reserved2] = { .next_index = SRP_INPUT_NEXT_ERROR, .error = SRP_ERROR_UNKNOWN_MODE, }, [SRP_MODE_reserved3] = { .next_index = SRP_INPUT_NEXT_ERROR, .error = SRP_ERROR_UNKNOWN_MODE, }, [SRP_MODE_keep_alive] = { .next_index = SRP_INPUT_NEXT_ERROR, .error = SRP_ERROR_KEEP_ALIVE_DROPPED, }, [SRP_MODE_data] = { .next_index = SRP_INPUT_NEXT_ETHERNET_INPUT, .buffer_advance = sizeof (srp_header_t), }, [SRP_MODE_control_pass_to_host] = { .next_index = SRP_INPUT_NEXT_CONTROL, }, [SRP_MODE_control_locally_buffered_for_host] = { .next_index = SRP_INPUT_NEXT_CONTROL, }, }; static uword srp_input (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * from_frame) { vnet_main_t * vnm = vnet_get_main(); srp_main_t * sm = &srp_main; u32 n_left_from, next_index, * from, * to_next; from = vlib_frame_vector_args (from_frame); n_left_from = from_frame->n_vectors; if (node->flags & VLIB_NODE_FLAG_TRACE) vlib_trace_frame_buffers_only (vm, node, from, n_left_from, sizeof (from[0]), sizeof (srp_input_trace_t)); 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 >= 4 && n_left_to_next >= 2) { u32 bi0, bi1, sw_if_index0, sw_if_index1; vlib_buffer_t * b0, * b1; u8 next0, next1, error0, error1; srp_header_t * s0, * s1; srp_input_disposition_t * d0, * d1; vnet_hw_interface_t * hi0, * hi1; srp_interface_t * si0, * si1; /* Prefetch next iteration. */ { vlib_buffer_t * b2, * b3; b2 = vlib_get_buffer (vm, from[2]); b3 = vlib_get_buffer (vm, from[3]); vlib_prefetch_buffer_header (b2, LOAD); vlib_prefetch_buffer_header (b3, LOAD); CLIB_PREFETCH (b2->data, sizeof (srp_header_t), LOAD); CLIB_PREFETCH (b3->data, sizeof (srp_header_t), LOAD); } bi0 = from[0]; bi1 = from[1]; to_next[0] = bi0; to_next[1] = bi1; from += 2; to_next += 2; n_left_to_next -= 2; n_left_from -= 2; b0 = vlib_get_buffer (vm, bi0); b1 = vlib_get_buffer (vm, bi1); s0 = (void *) (b0->data + b0->current_data); s1 = (void *) (b1->data + b1->current_data); /* Data packets are always assigned to side A (outer ring) interface. */ sw_if_index0 = vnet_buffer (b0)->sw_if_index[VLIB_RX]; sw_if_index1 = vnet_buffer (b1)->sw_if_index[VLIB_RX]; hi0 = vnet_get_sup_hw_interface (vnm, sw_if_index0); hi1 = vnet_get_sup_hw_interface (vnm, sw_if_index1); si0 = pool_elt_at_index (sm->interface_pool, hi0->hw_instance); si1 = pool_elt_at_index (sm->interface_pool, hi1->hw_instance); sw_if_index0 = (s0->mode == SRP_MODE_data ? si0->rings[SRP_RING_OUTER].sw_if_index : sw_if_index0); sw_if_index1 = (s1->mode == SRP_MODE_data ? si1->rings[SRP_RING_OUTER].sw_if_index : sw_if_index1); vnet_buffer (b0)->sw_if_index[VLIB_RX] = sw_if_index0; vnet_buffer (b1)->sw_if_index[VLIB_RX] = sw_if_index1; d0 = srp_input_disposition_by_mode + s0->mode; d1 = srp_input_disposition_by_mode + s1->mode; next0 = d0->next_index; next1 = d1->next_index; error0 = d0->error; error1 = d1->error; vlib_buffer_advance (b0, d0->buffer_advance); vlib_buffer_advance (b1, d1->buffer_advance); b0->error = node->errors[error0]; b1->error = node->errors[error1]; 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, sw_if_index0; vlib_buffer_t * b0; u8 next0, error0; srp_header_t * s0; srp_input_disposition_t * d0; srp_interface_t * si0; vnet_hw_interface_t * hi0; bi0 = from[0]; to_next[0] = bi0; from += 1; to_next += 1; n_left_to_next -= 1; n_left_from -= 1; b0 = vlib_get_buffer (vm, bi0); s0 = (void *) (b0->data + b0->current_data); /* Data packets are always assigned to side A (outer ring) interface. */ sw_if_index0 = vnet_buffer (b0)->sw_if_index[VLIB_RX]; hi0 = vnet_get_sup_hw_interface (vnm, sw_if_index0); si0 = pool_elt_at_index (sm->interface_pool, hi0->hw_instance); sw_if_index0 = (s0->mode == SRP_MODE_data ? si0->rings[SRP_RING_OUTER].sw_if_index : sw_if_index0); vnet_buffer (b0)->sw_if_index[VLIB_RX] = sw_if_index0; d0 = srp_input_disposition_by_mode + s0->mode; next0 = d0->next_index; error0 = d0->error; vlib_buffer_advance (b0, d0->buffer_advance); b0->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); } return from_frame->n_vectors; } static char * srp_error_strings[] = { #define _(f,s) s, foreach_srp_error #undef _ }; static vlib_node_registration_t srp_input_node = { .function = srp_input, .name = "srp-input", /* Takes a vector of packets. */ .vector_size = sizeof (u32), .n_errors = SRP_N_ERROR, .error_strings = srp_error_strings, .n_next_nodes = SRP_INPUT_N_NEXT, .next_nodes = { [SRP_INPUT_NEXT_ERROR] = "error-drop", [SRP_INPUT_NEXT_ETHERNET_INPUT] = "ethernet-input", [SRP_INPUT_NEXT_CONTROL] = "srp-control", }, .format_buffer = format_srp_header_with_length, .format_trace = format_srp_input_trace, .unformat_buffer = unformat_srp_header, }; static uword srp_topology_packet (vlib_main_t * vm, u32 sw_if_index, u8 ** contents) { vnet_main_t * vnm = vnet_get_main(); vnet_hw_interface_t * hi = vnet_get_sup_hw_interface (vnm, sw_if_index); srp_topology_header_t * t; srp_topology_mac_binding_t * mb; u32 nb, nmb; t = (void *) *contents; nb = clib_net_to_host_u16 (t->n_bytes_of_data_that_follows); nmb = (nb - sizeof (t->originator_address)) / sizeof (mb[0]); if (vec_len (*contents) < sizeof (t[0]) + nmb * sizeof (mb[0])) return SRP_ERROR_TOPOLOGY_BAD_LENGTH; /* Fill in our source MAC address. */ clib_memcpy (t->ethernet.src_address, hi->hw_address, vec_len (hi->hw_address)); /* Make space for our MAC binding. */ vec_resize (*contents, sizeof (srp_topology_mac_binding_t)); t = (void *) *contents; t->n_bytes_of_data_that_follows = clib_host_to_net_u16 (nb + sizeof (mb[0])); mb = t->bindings + nmb; mb->flags = ((t->srp.is_inner_ring ? SRP_TOPOLOGY_MAC_BINDING_FLAG_IS_INNER_RING : 0) | (/* is wrapped FIXME */ 0)); clib_memcpy (mb->address, hi->hw_address, vec_len (hi->hw_address)); t->control.checksum = ~ip_csum_fold (ip_incremental_checksum (0, &t->control, vec_len (*contents) - STRUCT_OFFSET_OF (srp_generic_control_header_t, control))); { vlib_frame_t * f = vlib_get_frame_to_node (vm, hi->output_node_index); vlib_buffer_t * b; u32 * to_next = vlib_frame_vector_args (f); u32 bi; bi = vlib_buffer_add_data (vm, VLIB_BUFFER_DEFAULT_FREE_LIST_INDEX, /* buffer to append to */ 0, *contents, vec_len (*contents)); 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; to_next[0] = bi; f->n_vectors = 1; vlib_put_frame_to_node (vm, hi->output_node_index, f); } return SRP_ERROR_CONTROL_PACKETS_PROCESSED; } typedef uword (srp_control_handler_function_t) (vlib_main_t * vm, u32 sw_if_index, u8 ** contents); static uword srp_control_input (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * from_frame) { u32 n_left_from, next_index, * from, * to_next; vlib_node_runtime_t * error_node; static u8 * contents; error_node = vlib_node_get_runtime (vm, srp_input_node.index); from = vlib_frame_vector_args (from_frame); n_left_from = from_frame->n_vectors; if (node->flags & VLIB_NODE_FLAG_TRACE) vlib_trace_frame_buffers_only (vm, node, from, n_left_from, sizeof (from[0]), sizeof (srp_input_trace_t)); next_index = node->cached_next_index; while (n_left_from > 0) { u32 n_left_to_next; vlib_get_next_frame (vm, node, next_index, to_next, n_left_to_next); while (n_left_from > 0 && n_left_to_next > 0) { u32 bi0, l2_len0, l3_len0; vlib_buffer_t * b0; u8 next0, error0; srp_generic_control_header_t * s0; bi0 = from[0]; to_next[0] = bi0; from += 1; to_next += 1; n_left_to_next -= 1; n_left_from -= 1; b0 = vlib_get_buffer (vm, bi0); s0 = (void *) (b0->data + b0->current_data); l2_len0 = vlib_buffer_length_in_chain (vm, b0); l3_len0 = l2_len0 - STRUCT_OFFSET_OF (srp_generic_control_header_t, control); error0 = SRP_ERROR_CONTROL_PACKETS_PROCESSED; error0 = s0->control.version != 0 ? SRP_ERROR_CONTROL_VERSION_NON_ZERO : error0; { u16 save0 = s0->control.checksum; u16 computed0; s0->control.checksum = 0; computed0 = ~ip_csum_fold (ip_incremental_checksum (0, &s0->control, l3_len0)); error0 = save0 != computed0 ? SRP_ERROR_CONTROL_BAD_CHECKSUM : error0; } if (error0 == SRP_ERROR_CONTROL_PACKETS_PROCESSED) { static srp_control_handler_function_t * t[SRP_N_CONTROL_PACKET_TYPE] = { [SRP_CONTROL_PACKET_TYPE_topology] = srp_topology_packet, }; srp_control_handler_function_t * f; f = 0; if (s0->control.type < ARRAY_LEN (t)) f = t[s0->control.type]; if (f) { vec_validate (contents, l2_len0 - 1); vlib_buffer_contents (vm, bi0, contents); error0 = f (vm, vnet_buffer (b0)->sw_if_index[VLIB_RX], &contents); } else error0 = SRP_ERROR_UNKNOWN_CONTROL; } b0->error = error_node->errors[error0]; next0 = 0; 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 from_frame->n_vectors; } static vlib_node_registration_t srp_control_input_node = { .function = srp_control_input, .name = "srp-control", /* Takes a vector of packets. */ .vector_size = sizeof (u32), .n_next_nodes = 1, .next_nodes = { [0] = "error-drop", }, .format_buffer = format_srp_header_with_length, .format_trace = format_srp_input_trace, .unformat_buffer = unformat_srp_header, }; static u8 * format_srp_ips_request_type (u8 * s, va_list * args) { u32 x = va_arg (*args, u32); char * t = 0; switch (x) { #define _(f,n) case SRP_IPS_REQUEST_##f: t = #f; break; foreach_srp_ips_request_type #undef _ default: return format (s, "unknown 0x%x", x); } return format (s, "%U", format_c_identifier, t); } static u8 * format_srp_ips_status (u8 * s, va_list * args) { u32 x = va_arg (*args, u32); char * t = 0; switch (x) { #define _(f,n) case SRP_IPS_STATUS_##f: t = #f; break; foreach_srp_ips_status #undef _ default: return format (s, "unknown 0x%x", x); } return format (s, "%U", format_c_identifier, t); } static u8 * format_srp_ips_state (u8 * s, va_list * args) { u32 x = va_arg (*args, u32); char * t = 0; switch (x) { #define _(f) case SRP_IPS_STATE_##f: t = #f; break; foreach_srp_ips_state #undef _ default: return format (s, "unknown 0x%x", x); } return format (s, "%U", format_c_identifier, t); } static u8 * format_srp_ring (u8 * s, va_list * args) { u32 ring = va_arg (*args, u32); return format (s, "%s", ring == SRP_RING_INNER ? "inner" : "outer"); } static u8 * format_srp_ips_header (u8 * s, va_list * args) { srp_ips_header_t * h = va_arg (*args, srp_ips_header_t *); s = format (s, "%U, %U, %U, %s-path", format_srp_ips_request_type, h->request_type, format_ethernet_address, h->originator_address, format_srp_ips_status, h->status, h->is_long_path ? "long" : "short"); return s; } static u8 * format_srp_interface (u8 * s, va_list * args) { srp_interface_t * si = va_arg (*args, srp_interface_t *); srp_interface_ring_t * ir; s = format (s, "address %U, IPS state %U", format_ethernet_address, si->my_address, format_srp_ips_state, si->current_ips_state); for (ir = si->rings; ir < si->rings + SRP_N_RING; ir++) if (ir->rx_neighbor_address_valid) s = format (s, ", %U neighbor %U", format_srp_ring, ir->ring, format_ethernet_address, ir->rx_neighbor_address); return s; } u8 * format_srp_device (u8 * s, va_list * args) { u32 hw_if_index = va_arg (*args, u32); CLIB_UNUSED (int verbose) = va_arg (*args, int); vnet_main_t * vnm = vnet_get_main(); srp_main_t * sm = &srp_main; vnet_hw_interface_t * hi = vnet_get_hw_interface (vnm, hw_if_index); srp_interface_t * si = pool_elt_at_index (sm->interface_pool, hi->hw_instance); return format (s, "%U", format_srp_interface, si); } always_inline srp_interface_t * srp_get_interface (u32 sw_if_index, srp_ring_type_t * ring) { vnet_main_t * vnm = vnet_get_main(); srp_main_t * sm = &srp_main; vnet_hw_interface_t * hi = vnet_get_sup_hw_interface (vnm, sw_if_index); srp_interface_t * si; ASSERT (hi->hw_class_index == srp_hw_interface_class.index); si = pool_elt_at_index (sm->interface_pool, hi->hw_instance); ASSERT (si->rings[SRP_RING_INNER].hw_if_index == hi->hw_if_index || si->rings[SRP_RING_OUTER].hw_if_index == hi->hw_if_index); if (ring) *ring = (hi->hw_if_index == si->rings[SRP_RING_INNER].hw_if_index ? SRP_RING_INNER : SRP_RING_OUTER); return si; } static void init_ips_packet (srp_interface_t * si, srp_ring_type_t tx_ring, srp_ips_header_t * i) { memset (i, 0, sizeof (i[0])); i->srp.ttl = 1; i->srp.is_inner_ring = tx_ring; i->srp.priority = 7; i->srp.mode = SRP_MODE_control_locally_buffered_for_host; srp_header_compute_parity (&i->srp); clib_memcpy (&i->ethernet.src_address, &si->my_address, sizeof (si->my_address)); i->ethernet.type = clib_host_to_net_u16 (ETHERNET_TYPE_SRP_CONTROL); /* Checksum will be filled in later. */ i->control.version = 0; i->control.type = SRP_CONTROL_PACKET_TYPE_ips; i->control.ttl = 255; clib_memcpy (&i->originator_address, &si->my_address, sizeof (si->my_address)); } static void tx_ips_packet (srp_interface_t * si, srp_ring_type_t tx_ring, srp_ips_header_t * i) { srp_main_t * sm = &srp_main; vnet_main_t * vnm = vnet_get_main(); vlib_main_t * vm = sm->vlib_main; vnet_hw_interface_t * hi = vnet_get_hw_interface (vnm, si->rings[tx_ring].hw_if_index); vlib_frame_t * f; vlib_buffer_t * b; u32 * to_next, bi; if (! vnet_sw_interface_is_admin_up (vnm, hi->sw_if_index)) return; if (hi->hw_class_index != srp_hw_interface_class.index) return; i->control.checksum = ~ip_csum_fold (ip_incremental_checksum (0, &i->control, sizeof (i[0]) - STRUCT_OFFSET_OF (srp_ips_header_t, control))); bi = vlib_buffer_add_data (vm, VLIB_BUFFER_DEFAULT_FREE_LIST_INDEX, /* buffer to append to */ 0, i, sizeof (i[0])); /* FIXME trace. */ if (0) clib_warning ("%U %U", format_vnet_sw_if_index_name, vnm, hi->sw_if_index, format_srp_ips_header, i); b = vlib_get_buffer (vm, bi); vnet_buffer (b)->sw_if_index[VLIB_RX] = vnet_buffer (b)->sw_if_index[VLIB_TX] = hi->sw_if_index; f = vlib_get_frame_to_node (vm, hi->output_node_index); 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 serialize_srp_interface_state_msg (serialize_main_t * m, va_list * va) { srp_interface_t * si = va_arg (*va, srp_interface_t *); srp_main_t * sm = &srp_main; int r; ASSERT (! pool_is_free (sm->interface_pool, si)); serialize_integer (m, si - sm->interface_pool, sizeof (u32)); serialize_likely_small_unsigned_integer (m, si->current_ips_state); for (r = 0; r < SRP_N_RING; r++) { srp_interface_ring_t * ir = &si->rings[r]; void * p; serialize_likely_small_unsigned_integer (m, ir->rx_neighbor_address_valid); if (ir->rx_neighbor_address_valid) { p = serialize_get (m, sizeof (ir->rx_neighbor_address)); clib_memcpy (p, ir->rx_neighbor_address, sizeof (ir->rx_neighbor_address)); } serialize_likely_small_unsigned_integer (m, ir->waiting_to_restore); if (ir->waiting_to_restore) serialize (m, serialize_f64, ir->wait_to_restore_start_time); } } static void unserialize_srp_interface_state_msg (serialize_main_t * m, va_list * va) { CLIB_UNUSED (mc_main_t * mcm) = va_arg (*va, mc_main_t *); srp_main_t * sm = &srp_main; srp_interface_t * si; u32 si_index, r; unserialize_integer (m, &si_index, sizeof (u32)); si = pool_elt_at_index (sm->interface_pool, si_index); si->current_ips_state = unserialize_likely_small_unsigned_integer (m); for (r = 0; r < SRP_N_RING; r++) { srp_interface_ring_t * ir = &si->rings[r]; void * p; ir->rx_neighbor_address_valid = unserialize_likely_small_unsigned_integer (m); if (ir->rx_neighbor_address_valid) { p = unserialize_get (m, sizeof (ir->rx_neighbor_address)); clib_memcpy (ir->rx_neighbor_address, p, sizeof (ir->rx_neighbor_address)); } ir->waiting_to_restore = unserialize_likely_small_unsigned_integer (m); if (ir->waiting_to_restore) unserialize (m, unserialize_f64, &ir->wait_to_restore_start_time); } } MC_SERIALIZE_MSG (srp_interface_state_msg, static) = { .name = "vnet_srp_interface_state", .serialize = serialize_srp_interface_state_msg, .unserialize = unserialize_srp_interface_state_msg, }; static int requests_switch (srp_ips_request_type_t r) { static u8 t[16] = { [SRP_IPS_REQUEST_forced_switch] = 1, [SRP_IPS_REQUEST_manual_switch] = 1, [SRP_IPS_REQUEST_signal_fail] = 1, [SRP_IPS_REQUEST_signal_degrade] = 1, }; return (int) r < ARRAY_LEN (t) ? t[r] : 0; } /* Called when an IPS control packet is received on given interface. */ void srp_ips_rx_packet (u32 sw_if_index, srp_ips_header_t * h) { vnet_main_t * vnm = vnet_get_main(); vlib_main_t * vm = srp_main.vlib_main; srp_ring_type_t rx_ring; srp_interface_t * si = srp_get_interface (sw_if_index, &rx_ring); srp_interface_ring_t * ir = &si->rings[rx_ring]; int si_needs_broadcast = 0; /* FIXME trace. */ if (0) clib_warning ("%U %U %U", format_time_interval, "h:m:s:u", vlib_time_now (vm), format_vnet_sw_if_index_name, vnm, sw_if_index, format_srp_ips_header, h); /* Ignore self-generated IPS packets. */ if (! memcmp (h->originator_address, si->my_address, sizeof (h->originator_address))) goto done; /* Learn neighbor address from short path messages. */ if (! h->is_long_path) { if (ir->rx_neighbor_address_valid && memcmp (ir->rx_neighbor_address, h->originator_address, sizeof (ir->rx_neighbor_address))) { ASSERT (0); } ir->rx_neighbor_address_valid = 1; clib_memcpy (ir->rx_neighbor_address, h->originator_address, sizeof (ir->rx_neighbor_address)); } switch (si->current_ips_state) { case SRP_IPS_STATE_idle: /* Received {REQ,NEIGHBOR,W,S} in idle state: wrap. */ if (requests_switch (h->request_type) && ! h->is_long_path && h->status == SRP_IPS_STATUS_wrapped) { srp_ips_header_t to_tx[2]; si_needs_broadcast = 1; si->current_ips_state = SRP_IPS_STATE_wrapped; si->hw_wrap_function (si->rings[SRP_SIDE_A].hw_if_index, /* enable_wrap */ 1); si->hw_wrap_function (si->rings[SRP_SIDE_B].hw_if_index, /* enable_wrap */ 1); init_ips_packet (si, rx_ring ^ 0, &to_tx[0]); to_tx[0].request_type = SRP_IPS_REQUEST_idle; to_tx[0].status = SRP_IPS_STATUS_wrapped; to_tx[0].is_long_path = 0; tx_ips_packet (si, rx_ring ^ 0, &to_tx[0]); init_ips_packet (si, rx_ring ^ 1, &to_tx[1]); to_tx[1].request_type = h->request_type; to_tx[1].status = SRP_IPS_STATUS_wrapped; to_tx[1].is_long_path = 1; tx_ips_packet (si, rx_ring ^ 1, &to_tx[1]); } break; case SRP_IPS_STATE_wrapped: if (! h->is_long_path && h->request_type == SRP_IPS_REQUEST_idle && h->status == SRP_IPS_STATUS_idle) { si_needs_broadcast = 1; si->current_ips_state = SRP_IPS_STATE_idle; si->hw_wrap_function (si->rings[SRP_SIDE_A].hw_if_index, /* enable_wrap */ 0); si->hw_wrap_function (si->rings[SRP_SIDE_B].hw_if_index, /* enable_wrap */ 0); } break; case SRP_IPS_STATE_pass_thru: /* FIXME */ break; default: abort (); break; } done: if (vm->mc_main && si_needs_broadcast) mc_serialize (vm->mc_main, &srp_interface_state_msg, si); } /* Preform local IPS request on given interface. */ void srp_ips_local_request (u32 sw_if_index, srp_ips_request_type_t request) { vnet_main_t * vnm = vnet_get_main(); srp_main_t * sm = &srp_main; vlib_main_t * vm = sm->vlib_main; srp_ring_type_t rx_ring; srp_interface_t * si = srp_get_interface (sw_if_index, &rx_ring); srp_interface_ring_t * ir = &si->rings[rx_ring]; int si_needs_broadcast = 0; if (request == SRP_IPS_REQUEST_wait_to_restore) { if (si->current_ips_state != SRP_IPS_STATE_wrapped) return; if (! ir->waiting_to_restore) { ir->wait_to_restore_start_time = vlib_time_now (sm->vlib_main); ir->waiting_to_restore = 1; si_needs_broadcast = 1; } } else { /* FIXME handle local signal fail. */ si_needs_broadcast = ir->waiting_to_restore; ir->wait_to_restore_start_time = 0; ir->waiting_to_restore = 0; } /* FIXME trace. */ if (0) clib_warning ("%U %U", format_vnet_sw_if_index_name, vnm, sw_if_index, format_srp_ips_request_type, request); if (vm->mc_main && si_needs_broadcast) mc_serialize (vm->mc_main, &srp_interface_state_msg, si); } static void maybe_send_ips_message (srp_interface_t * si) { srp_main_t * sm = &srp_main; srp_ips_header_t to_tx[2]; srp_ring_type_t rx_ring = SRP_RING_OUTER; srp_interface_ring_t * r0 = &si->rings[rx_ring ^ 0]; srp_interface_ring_t * r1 = &si->rings[rx_ring ^ 1]; f64 now = vlib_time_now (sm->vlib_main); if (! si->ips_process_enable) return; if (si->current_ips_state == SRP_IPS_STATE_wrapped && r0->waiting_to_restore && r1->waiting_to_restore && now >= r0->wait_to_restore_start_time + si->config.wait_to_restore_idle_delay && now >= r1->wait_to_restore_start_time + si->config.wait_to_restore_idle_delay) { si->current_ips_state = SRP_IPS_STATE_idle; r0->waiting_to_restore = r1->waiting_to_restore = 0; r0->wait_to_restore_start_time = r1->wait_to_restore_start_time = 0; } if (si->current_ips_state != SRP_IPS_STATE_idle) return; init_ips_packet (si, rx_ring ^ 0, &to_tx[0]); init_ips_packet (si, rx_ring ^ 1, &to_tx[1]); if (si->current_ips_state == SRP_IPS_STATE_idle) { to_tx[0].request_type = to_tx[1].request_type = SRP_IPS_REQUEST_idle; to_tx[0].status = to_tx[1].status = SRP_IPS_STATUS_idle; to_tx[0].is_long_path = to_tx[1].is_long_path = 0; } else if (si->current_ips_state == SRP_IPS_STATE_wrapped) { to_tx[0].request_type = (si->rings[rx_ring ^ 0].waiting_to_restore ? SRP_IPS_REQUEST_wait_to_restore : SRP_IPS_REQUEST_signal_fail); to_tx[1].request_type = (si->rings[rx_ring ^ 1].waiting_to_restore ? SRP_IPS_REQUEST_wait_to_restore : SRP_IPS_REQUEST_signal_fail); to_tx[0].status = to_tx[1].status = SRP_IPS_STATUS_wrapped; to_tx[0].is_long_path = 0; to_tx[1].is_long_path = 1; } tx_ips_packet (si, rx_ring ^ 0, &to_tx[0]); tx_ips_packet (si, rx_ring ^ 1, &to_tx[1]); } static uword srp_ips_process (vlib_main_t * vm, vlib_node_runtime_t * rt, vlib_frame_t * f) { srp_main_t * sm = &srp_main; srp_interface_t * si; while (1) { pool_foreach (si, sm->interface_pool, ({ maybe_send_ips_message (si); })); vlib_process_suspend (vm, 1.0); } return 0; } vlib_node_registration_t srp_ips_process_node = { .function = srp_ips_process, .type = VLIB_NODE_TYPE_PROCESS, .name = "srp-ips-process", .state = VLIB_NODE_STATE_DISABLED, }; static clib_error_t * srp_init (vlib_main_t * vm) { srp_main_t * sm = &srp_main; sm->default_data_ttl = 255; sm->vlib_main = vm; vlib_register_node (vm, &srp_ips_process_node); vlib_register_node (vm, &srp_input_node); vlib_register_node (vm, &srp_control_input_node); srp_setup_node (vm, srp_input_node.index); return 0; } VLIB_INIT_FUNCTION (srp_init);