path: root/src/vnet/udp/udp_input.c

/*
 * Copyright (c) 2016-2019 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 <vlibmemory/api.h>
#include <vlib/vlib.h>

#include <vppinfra/hash.h>
#include <vppinfra/error.h>
#include <vppinfra/elog.h>

#include <vnet/vnet.h>
#include <vnet/ip/ip.h>
#include <vnet/udp/udp.h>
#include <vnet/udp/udp_packet.h>
#include <vnet/session/session.h>

static char *udp_error_strings[] = {
#define udp_error(n,s) s,
#include "udp_error.def"
#undef udp_error
};

typedef struct
{
  u32 connection;
  u32 disposition;
  u32 thread_index;
} udp_input_trace_t;

/* packet trace format function */
static u8 *
format_udp_input_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 *);
  udp_input_trace_t *t = va_arg (*args, udp_input_trace_t *);

  s = format (s, "UDP_INPUT: connection %d, disposition %d, thread %d",
	      t->connection, t->disposition, t->thread_index);
  return s;
}

#define foreach_udp_input_next			\
  _ (DROP, "error-drop")

typedef enum
{
#define _(s, n) UDP_INPUT_NEXT_##s,
  foreach_udp_input_next
#undef _
    UDP_INPUT_N_NEXT,
} udp_input_next_t;

always_inline void
udp_input_inc_counter (vlib_main_t * vm, u8 is_ip4, u8 evt, u8 val)
{
  if (is_ip4)
    vlib_node_increment_counter (vm, udp4_input_node.index, evt, val);
  else
    vlib_node_increment_counter (vm, udp6_input_node.index, evt, val);
}

#define udp_store_err_counters(vm, is_ip4, cnts)			\
{									\
  int i;								\
  for (i = 0; i < UDP_N_ERROR; i++)					\
    if (cnts[i])							\
      udp_input_inc_counter(vm, is_ip4, i, cnts[i]);			\
}

#define udp_inc_err_counter(cnts, err, val)				\
{									\
  cnts[err] += val;							\
}

static void
udp_trace_buffer (vlib_main_t * vm, vlib_node_runtime_t * node,
		  vlib_buffer_t * b, session_t * s, u16 error0)
{
  udp_input_trace_t *t;

  if (PREDICT_TRUE (!(b->flags & VLIB_BUFFER_IS_TRACED)))
    return;

  t = vlib_add_trace (vm, node, b, sizeof (*t));
  t->connection = s ? s->connection_index : ~0;
  t->disposition = error0;
  t->thread_index = s ? s->thread_index : vm->thread_index;
}

static udp_connection_t *
udp_connection_accept (udp_connection_t * listener, session_dgram_hdr_t * hdr,
		       u32 thread_index)
{
  udp_connection_t *uc;

  uc = udp_connection_alloc (thread_index);
  ip_copy (&uc->c_lcl_ip, &hdr->lcl_ip, hdr->is_ip4);
  ip_copy (&uc->c_rmt_ip, &hdr->rmt_ip, hdr->is_ip4);
  uc->c_lcl_port = hdr->lcl_port;
  uc->c_rmt_port = hdr->rmt_port;
  uc->c_is_ip4 = hdr->is_ip4;
  uc->c_fib_index = listener->c_fib_index;
  uc->mss = listener->mss;
  uc->flags |= UDP_CONN_F_CONNECTED;

  if (session_dgram_accept (&uc->connection, listener->c_s_index,
			    listener->c_thread_index))
    {
      udp_connection_free (uc);
      return 0;
    }
  udp_connection_share_port (clib_net_to_host_u16
			     (uc->c_lcl_port), uc->c_is_ip4);
  return uc;
}

static void
udp_connection_enqueue (udp_connection_t * uc0, session_t * s0,
			session_dgram_hdr_t * hdr0, u32 thread_index,
			vlib_buffer_t * b, u8 queue_event, u32 * error0)
{
  int wrote0;

  clib_spinlock_lock (&uc0->rx_lock);

  if (svm_fifo_max_enqueue_prod (s0->rx_fifo)
      < hdr0->data_length + sizeof (session_dgram_hdr_t))
    {
      *error0 = UDP_ERROR_FIFO_FULL;
      goto unlock_rx_lock;
    }

  /* If session is owned by another thread and rx event needed,
   * enqueue event now while we still have the peeker lock */
  if (s0->thread_index != thread_index)
    {
      wrote0 = session_enqueue_dgram_connection (s0, hdr0, b,
						 TRANSPORT_PROTO_UDP,
						 /* queue event */ 0);
      if (queue_event && !svm_fifo_has_event (s0->rx_fifo))
	session_enqueue_notify (s0);
    }
  else
    {
      wrote0 = session_enqueue_dgram_connection (s0, hdr0, b,
						 TRANSPORT_PROTO_UDP,
						 queue_event);
    }
  ASSERT (wrote0 > 0);

unlock_rx_lock:

  clib_spinlock_unlock (&uc0->rx_lock);
}

always_inline session_t *
udp_parse_and_lookup_buffer (vlib_buffer_t * b, session_dgram_hdr_t * hdr,
			     u8 is_ip4)
{
  udp_header_t *udp;
  u32 fib_index;
  session_t *s;

  /* udp_local hands us a pointer to the udp data */
  udp = (udp_header_t *) (vlib_buffer_get_current (b) - sizeof (*udp));
  fib_index = vnet_buffer (b)->ip.fib_index;

  hdr->data_offset = 0;
  hdr->lcl_port = udp->dst_port;
  hdr->rmt_port = udp->src_port;
  hdr->is_ip4 = is_ip4;

  if (is_ip4)
    {
      ip4_header_t *ip4;

      /* TODO: must fix once udp_local does ip options correctly */
      ip4 = (ip4_header_t *) (((u8 *) udp) - sizeof (*ip4));
      ip_set (&hdr->lcl_ip, &ip4->dst_address, 1);
      ip_set (&hdr->rmt_ip, &ip4->src_address, 1);
      hdr->data_length = clib_net_to_host_u16 (ip4->length);
      hdr->data_length -= sizeof (ip4_header_t) + sizeof (udp_header_t);
      s = session_lookup_safe4 (fib_index, &ip4->dst_address,
				&ip4->src_address, udp->dst_port,
				udp->src_port, TRANSPORT_PROTO_UDP);
    }
  else
    {
      ip6_header_t *ip60;

      ip60 = (ip6_header_t *) (((u8 *) udp) - sizeof (*ip60));
      ip_set (&hdr->lcl_ip, &ip60->dst_address, 0);
      ip_set (&hdr->rmt_ip, &ip60->src_address, 0);
      hdr->data_length = clib_net_to_host_u16 (ip60->payload_length);
      hdr->data_length -= sizeof (
/*
 * 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/adj/adj_nbr.h>
#include <vnet/adj/adj_internal.h>
#include <vnet/ethernet/arp_packet.h>
#include <vnet/fib/fib_walk.h>

/*
 * Vector Hash tables of neighbour (traditional) adjacencies
 *  Key: interface(for the vector index), address (and its proto),
 *       link-type/ether-type.
 */
static uword **adj_nbr_tables[FIB_PROTOCOL_IP_MAX];

typedef struct adj_nbr_key_t_
{
    ip46_address_t ank_ip;
    u64 ank_linkt;
} adj_nbr_key_t;

#define ADJ_NBR_SET_KEY(_key, _lt, _nh)         \
{						\
    ip46_address_copy(&(_key).ank_ip, (_nh));   \
    _key.ank_linkt = (_lt);			\
}

#define ADJ_NBR_ITF_OK(_proto, _itf)			\
    (((_itf) < vec_len(adj_nbr_tables[_proto])) &&	\
     (NULL != adj_nbr_tables[_proto][(_itf)]))

#define ADJ_NBR_ASSERT_NH_PROTO(nh_proto, err)          \
  do {                                                  \
      ASSERT (nh_proto < FIB_PROTOCOL_IP_MAX);          \
      const fib_protocol_t nh_proto__ = (nh_proto);     \
      if (nh_proto__ >= FIB_PROTOCOL_IP_MAX)            \
        {                                               \
          clib_warning ("BUG: protocol %d > %d\n",      \
                        (int)nh_proto__,                \
                        FIB_PROTOCOL_IP_MAX);           \
          return err;                                   \
        }                                               \
  } while (0)

static void
adj_nbr_insert (fib_protocol_t nh_proto,
		vnet_link_t link_type,
		const ip46_address_t *nh_addr,
		u32 sw_if_index,
		adj_index_t adj_index)
{
    adj_nbr_key_t kv;

    ADJ_NBR_ASSERT_NH_PROTO (nh_proto,);

    if (sw_if_index >= vec_len(adj_nbr_tables[nh_proto]))
    {
	vec_validate(adj_nbr_tables[nh_proto], sw_if_index);
    }
    if (NULL == adj_nbr_tables[nh_proto][sw_if_index])
    {
	adj_nbr_tables[nh_proto][sw_if_index] =
	    hash_create_mem(0, sizeof(adj_nbr_key_t), sizeof(adj_index_t));
    }

    ADJ_NBR_SET_KEY(kv, link_type, nh_addr);

    hash_set_mem_alloc (&adj_nbr_tables[nh_proto][sw_if_index],
                        &kv, adj_index);
}

void
adj_nbr_remove (adj_index_t ai,
                fib_protocol_t nh_proto,
		vnet_link_t link_type,
		const ip46_address_t *nh_addr,
		u32 sw_if_index)
{
    adj_nbr_key_t kv;

    ADJ_NBR_ASSERT_NH_PROTO (nh_proto,);

    if (!ADJ_NBR_ITF_OK(nh_proto, sw_if_index))
	return;

    ADJ_NBR_SET_KEY(kv, link_type, nh_addr);

    hash_unset_mem_free(&adj_nbr_tables[nh_proto][sw_if_index], &kv);

    if (0 == hash_elts(adj_nbr_tables[nh_proto][sw_if_index]))
    {
        hash_free(adj_nbr_tables[nh_proto][sw_if_index]);
    }
}

adj_index_t
adj_nbr_find (fib_protocol_t nh_proto,
	      vnet_link_t link_type,
	      const ip46_address_t *nh_addr,
	      u32 sw_if_index)
{
    adj_nbr_key_t kv;
    uword *p;

    ADJ_NBR_ASSERT_NH_PROTO (nh_proto, ADJ_INDEX_INVALID);

    ADJ_NBR_SET_KEY(kv, link_type, nh_addr);

    if (!ADJ_NBR_ITF_OK(nh_proto, sw_if_index))
	return (ADJ_INDEX_INVALID);

    p = hash_get_mem(adj_nbr_tables[nh_proto][sw_if_index], &kv);

    if (p)
    {
	return (p[0]);
    }
    return (ADJ_INDEX_INVALID);
}

static inline u32
adj_get_nd_node (fib_protocol_t proto)
{
    switch (proto) {
    case FIB_PROTOCOL_IP4:
	return (ip4_arp_node.index);
    case FIB_PROTOCOL_IP6:
	return (ip6_discover_neighbor_node.index);
    case FIB_PROTOCOL_MPLS:
	break;
    }
    ASSERT(0);
    return (ip4_arp_node.index);
}

/**
 * @brief Check and set feature flags if o/p interface has any o/p features.
 */
static void
adj_nbr_evaluate_feature (adj_index_t ai)
{
    ip_adjacency_t *adj;
    vnet_feature_main_t *fm = &feature_main;
    i16 feature_count;
    u8 arc_index;
    u32 sw_if_index;

    adj = adj_get(ai);

    switch (adj->ia_link)
    {
    case VNET_LINK_IP4:
        arc_index = ip4_main.lookup_main.output_feature_arc_index;
        break;
    case VNET_LINK_IP6:
        arc_index = ip6_main.lookup_main.output_feature_arc_index;
        break;
    case VNET_LINK_MPLS:
        arc_index = mpls_main.output_feature_arc_index;
        break;
    default:
        return;
    }

    sw_if_index = adj->rewrite_header.sw_if_index;
    if (vec_len(fm->feature_count_by_sw_if_index[arc_index]) > sw_if_index)
    {
        feature_count = fm->feature_count_by_sw_if_index[arc_index][sw_if_index];
        if (feature_count > 0)
        {
            vnet_feature_config_main_t *cm;

            adj->rewrite_header.flags |= VNET_REWRITE_HAS_FEATURES;
            cm = &fm->feature_config_mains[arc_index];

            adj->ia_cfg_index = vec_elt (cm->config_index_by_sw_if_index,
                                         sw_if_index);
        }
    }
    return;
}

static ip_adjacency_t*
adj_nbr_alloc (fib_protocol_t nh_proto,
	       vnet_link_t link_type,
	       const ip46_address_t *nh_addr,
	       u32 sw_if_index)
{
    ip_adjacency_t *adj;

    adj = adj_alloc(nh_proto);

    adj_nbr_insert(nh_proto, link_type, nh_addr,
		   sw_if_index,
		   adj_get_index(adj));

    /*
     * since we just added the ADJ we have no rewrite string for it,
     * so its for ARP
     */
    adj->lookup_next_index = IP_LOOKUP_NEXT_ARP;
    adj->sub_type.nbr.next_hop = *nh_addr;
    adj->ia_link = link_type;
    adj->ia_nh_proto = nh_proto;
    adj->rewrite_header.sw_if_index = sw_if_index;
    vnet_rewrite_update_mtu(vnet_get_main(), adj->ia_link,
                            &adj->rewrite_header);

    adj_nbr_evaluate_feature (adj_get_index(adj));
    return (adj);
}

void
adj_nbr_set_mtu (adj_index_t adj_index, u16 mtu)
{
    ip_adjacency_t *adj;

    ASSERT(ADJ_INDEX_INVALID != adj_index);

    adj = adj_get(adj_index);

    if (0