diff options
Diffstat (limited to 'src/vnet/bier/bier_lookup.c')
-rw-r--r-- | src/vnet/bier/bier_lookup.c | 367 |
1 files changed, 367 insertions, 0 deletions
diff --git a/src/vnet/bier/bier_lookup.c b/src/vnet/bier/bier_lookup.c new file mode 100644 index 00000000000..4cf29f886e4 --- /dev/null +++ b/src/vnet/bier/bier_lookup.c @@ -0,0 +1,367 @@ +/* + * 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 <vnet/buffer.h> +#include <vnet/vnet.h> + +#include <vnet/bier/bier_fmask.h> +#include <vnet/bier/bier_hdr_inlines.h> +#include <vnet/bier/bier_table.h> +#include <vnet/bier/bier_fmask.h> + +/** + * Struct maintining the per-worker thread data for BIER lookups + */ +typedef struct bier_lookup_main_t_ +{ + /* per-cpu vector of cloned packets */ + u32 **blm_clones; + /* per-cpu vector of BIER fmasks */ + u32 **blm_fmasks; +} bier_lookup_main_t; + +/** + * Single instance of the lookup main + */ +static bier_lookup_main_t bier_lookup_main; + +static char * bier_lookup_error_strings[] = { +#define bier_error(n,s) s, +#include <vnet/bier/bier_lookup_error.def> +#undef bier_error +}; + +/* + * Keep these values sematically the same as BIER lookup + */ +#define foreach_bier_lookup_next \ + _(DROP, "bier-drop") \ + _(OUTPUT, "bier-output") + +typedef enum { +#define _(s,n) BIER_LOOKUP_NEXT_##s, + foreach_bier_lookup_next +#undef _ + BIER_LOOKUP_N_NEXT, +} bier_lookup_next_t; + +typedef enum { +#define bier_error(n,s) BIER_LOOKUP_ERROR_##n, +#include <vnet/bier/bier_lookup_error.def> +#undef bier_error + BIER_LOOKUP_N_ERROR, +} bier_lookup_error_t; + +vlib_node_registration_t bier_lookup_node; + +/** + * @brief Packet trace recoed for a BIER lookup + */ +typedef struct bier_lookup_trace_t_ +{ + u32 next_index; + index_t bt_index; + index_t bfm_index; +} bier_lookup_trace_t; + +static uword +bier_lookup (vlib_main_t * vm, + vlib_node_runtime_t * node, + vlib_frame_t * from_frame) +{ + u32 n_left_from, next_index, * from, * to_next; + bier_lookup_main_t *blm = &bier_lookup_main; + u32 thread_index = vlib_get_thread_index(); + + from = vlib_frame_vector_args (from_frame); + n_left_from = from_frame->n_vectors; + next_index = BIER_LOOKUP_NEXT_DROP; + + 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) + { + bier_bit_mask_bucket_t buckets_copy[BIER_HDR_BUCKETS_256]; + u32 next0, bi0, n_bytes, bti0, bfmi0; + const bier_fmask_t *bfm0; + const bier_table_t *bt0; + u16 index, num_buckets; + const bier_hdr_t *bh0; + bier_bit_string_t bbs; + vlib_buffer_t *b0; + bier_bp_t fbs; + int bucket; + + bi0 = from[0]; + from += 1; + n_left_from -= 1; + + b0 = vlib_get_buffer (vm, bi0); + bh0 = vlib_buffer_get_current (b0); + bti0 = vnet_buffer(b0)->ip.adj_index[VLIB_TX]; + + /* + * default to drop so that if no bits are matched then + * that is where we go - DROP. + */ + next0 = BIER_LOOKUP_NEXT_DROP; + + /* + * At the imposition or input node, + * we stored the BIER Table index in the TX adjacency + */ + bt0 = bier_table_get(vnet_buffer(b0)->ip.adj_index[VLIB_TX]); + + /* + * we should only forward via one for the ECMP tables + */ + ASSERT(!bier_table_is_main(bt0)); + + /* + * number of integer sized buckets + */ + n_bytes = bier_hdr_len_id_to_num_buckets(bt0->bt_id.bti_hdr_len); + vnet_buffer(b0)->bier.n_bytes = n_bytes; + vnet_buffer(b0)->sw_if_index[VLIB_TX] = ~0; + num_buckets = n_bytes / sizeof(int); + bier_bit_string_init(&bbs, + bt0->bt_id.bti_hdr_len, + buckets_copy); + memcpy(bbs.bbs_buckets, bh0->bh_bit_string, bbs.bbs_len); + + /* + * reset the fmask and clone storage vectors + */ + vec_reset_length (blm->blm_fmasks[thread_index]); + vec_reset_length (blm->blm_clones[thread_index]); + + /* + * Loop through the buckets in the header + */ + for (index = 0; index < num_buckets; index++) { + /* + * loop through each bit in the bucket + */ + bucket = ((int*)bbs.bbs_buckets)[index]; + + while (bucket) { + fbs = bier_find_first_bit_string_set(bucket); + fbs += (((num_buckets - 1) - index) * + BIER_BIT_MASK_BITS_PER_INT); + + bfmi0 = bier_table_fwd_lookup(bt0, fbs); + + /* + * whatever happens, the bit we just looked for + * MUST be cleared from the packet + * otherwise we could be in this loop a while ... + */ + bier_bit_string_clear_bit(&bbs, fbs); + + if (PREDICT_TRUE(INDEX_INVALID != bfmi0)) + { + bfm0 = bier_fmask_get(bfmi0); + vnet_buffer (b0)->ip.adj_index[VLIB_TX] = bfmi0; + + /* + * use the bit-string on the fmask to reset + * the bits in the header we are walking + */ + bier_bit_string_clear_string( + &bfm0->bfm_bits.bfmb_input_reset_string, + &bbs); + bucket = ((int*)bbs.bbs_buckets)[index]; + + /* + * the fmask is resolved so replicate a + * packet its way + */ + next0 = BIER_LOOKUP_NEXT_OUTPUT; + + vec_add1 (blm->blm_fmasks[thread_index], bfmi0); + } else { + /* + * go to the next bit-position set + */ + bucket = ((int*)bbs.bbs_buckets)[index]; + continue; + } + } + } + + /* + * Full mask now processed. + * Create the number of clones we need based on the number + * of fmasks we are sending to. + */ + u8 num_cloned, clone; + u32 n_clones; + + n_clones = vec_len(blm->blm_fmasks[thread_index]); + + if (PREDICT_TRUE(0 != n_clones)) + { + ASSERT(n_clones < 256); + num_cloned = vlib_buffer_clone(vm, bi0, + blm->blm_clones[thread_index], + n_clones, 128); + + if (num_cloned != n_clones) + { + vlib_node_increment_counter + (vm, node->node_index, + BIER_LOOKUP_ERROR_BUFFER_ALLOCATION_FAILURE, 1); + } + + for (clone = 0; clone < num_cloned; clone++) + { + vlib_buffer_t *c0; + u32 ci0; + + ci0 = blm->blm_clones[thread_index][clone]; + c0 = vlib_get_buffer(vm, ci0); + + to_next[0] = ci0; + to_next += 1; + n_left_to_next -= 1; + + if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED)) + { + bier_lookup_trace_t *tr; + + vlib_trace_buffer (vm, node, next0, c0, 0); + tr = vlib_add_trace (vm, node, c0, sizeof (*tr)); + tr->bt_index = bti0; + tr->bfm_index = blm->blm_fmasks[thread_index][clone]; + + c0->flags |= VLIB_BUFFER_IS_TRACED; + } + + vlib_validate_buffer_enqueue_x1(vm, node, next_index, + to_next, n_left_to_next, + ci0, next0); + + /* + * After the enqueue it is possible that we over-flow the + * frame of the to-next node. When this happens we need to + * 'put' that full frame to the node and get a fresh empty + * one. Note that these are macros with side effects that + * change to_next & n_left_to_next + */ + if (PREDICT_FALSE(0 == n_left_to_next)) + { + vlib_put_next_frame (vm, node, next_index, + n_left_to_next); + vlib_get_next_frame (vm, node, next_index, + to_next, n_left_to_next); + } + } + } + else + { + /* + * no clones/replications required. drop this packet + */ + next0 = BIER_LOOKUP_NEXT_DROP; + to_next[0] = bi0; + to_next += 1; + n_left_to_next -= 1; + + if (PREDICT_FALSE(b0->flags & VLIB_BUFFER_IS_TRACED)) + { + bier_lookup_trace_t *tr; + + tr = vlib_add_trace (vm, node, b0, sizeof (*tr)); + + tr->bt_index = bti0; + tr->bfm_index = ~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); + } + + vlib_node_increment_counter (vm, bier_lookup_node.index, + BIER_LOOKUP_ERROR_NONE, + from_frame->n_vectors); + return (from_frame->n_vectors); +} + +static u8 * +format_bier_lookup_trace (u8 * s, va_list * args) +{ + CLIB_UNUSED (vlib_main_t * vm) = va_arg (*args, vlib_main_t *); + CLIB_UNUSED (vlib_node_t * node) = va_arg (*args, vlib_node_t *); + bier_lookup_trace_t * t = va_arg (*args, bier_lookup_trace_t *); + + s = format (s, "BIER: next [%d], tbl:%d BFM:%d", + t->next_index, + t->bt_index, + t->bfm_index); + return s; +} + +VLIB_REGISTER_NODE (bier_lookup_node) = { + .function = bier_lookup, + .name = "bier-lookup", + /* Takes a vector of packets. */ + .vector_size = sizeof (u32), + + .n_errors = BIER_LOOKUP_N_ERROR, + .error_strings = bier_lookup_error_strings, + + .format_trace = format_bier_lookup_trace, + .n_next_nodes = BIER_LOOKUP_N_NEXT, + .next_nodes = { + [BIER_LOOKUP_NEXT_DROP] = "bier-drop", + [BIER_LOOKUP_NEXT_OUTPUT] = "bier-output", + }, +}; + +clib_error_t * +bier_lookup_module_init (vlib_main_t * vm) +{ + bier_lookup_main_t *blm = &bier_lookup_main; + u32 thread_index; + + vec_validate (blm->blm_clones, vlib_num_workers()); + vec_validate (blm->blm_fmasks, vlib_num_workers()); + + for (thread_index = 0; + thread_index <= vlib_num_workers(); + thread_index++) + { + /* + * 4096 is the most we will ever need to support + * a Bit-Mask length of 4096 + */ + vec_validate(blm->blm_fmasks[thread_index], 4095); + vec_validate(blm->blm_clones[thread_index], 4095); + } + + return 0; +} + +VLIB_INIT_FUNCTION (bier_lookup_module_init); |