summaryrefslogtreecommitdiffstats
path: root/vnet
diff options
context:
space:
mode:
authorKlement Sekera <ksekera@cisco.com>2016-09-29 14:43:44 +0200
committerDave Barach <openvpp@barachs.net>2016-12-07 19:54:12 +0000
commit0e3c0de1ed87f3cdf16e26e05e39ea6eebeafb18 (patch)
treedd5f11c08311a95ce56eb8c33d4e8681940bf877 /vnet
parentd171d48edc1672564db8bb920586b8ea220df14c (diff)
BFD: basic asynchronous session up/down
This is a work-in-progress basic BFD session handling. Only asynchronous mode is supported at the moment. Setting the session flags doesn't work. Change-Id: Idba27f721b5c35be5a66a6d202a63d23ff7ecf6f Signed-off-by: Klement Sekera <ksekera@cisco.com>
Diffstat (limited to 'vnet')
-rw-r--r--vnet/Makefile.am16
-rw-r--r--vnet/vnet/api_errno.h4
-rw-r--r--vnet/vnet/bfd/bfd_api.h46
-rw-r--r--vnet/vnet/bfd/bfd_debug.h80
-rw-r--r--vnet/vnet/bfd/bfd_doc.md1
-rw-r--r--vnet/vnet/bfd/bfd_main.c928
-rw-r--r--vnet/vnet/bfd/bfd_main.h218
-rw-r--r--vnet/vnet/bfd/bfd_protocol.c74
-rw-r--r--vnet/vnet/bfd/bfd_protocol.h154
-rw-r--r--vnet/vnet/bfd/bfd_udp.c610
-rw-r--r--vnet/vnet/bfd/bfd_udp.h56
-rw-r--r--vnet/vnet/bfd/dir.dox18
-rw-r--r--vnet/vnet/buffer.h4
-rw-r--r--vnet/vnet/ip/ip4_forward.c12
-rw-r--r--vnet/vnet/ip/udp.h2
15 files changed, 2218 insertions, 5 deletions
diff --git a/vnet/Makefile.am b/vnet/Makefile.am
index eb7f8ef1a01..4e4b2c0acb3 100644
--- a/vnet/Makefile.am
+++ b/vnet/Makefile.am
@@ -350,6 +350,22 @@ nobase_include_HEADERS += \
vnet/ip/udp_packet.h
########################################
+# Bidirectional Forwarding Detection
+########################################
+
+nobase_include_HEADERS += \
+ vnet/bfd/bfd_protocol.h \
+ vnet/bfd/bfd_main.h \
+ vnet/bfd/bfd_api.h \
+ vnet/bfd/bfd_udp.h
+
+libvnet_la_SOURCES += \
+ vnet/bfd/bfd_api.h \
+ vnet/bfd/bfd_udp.c \
+ vnet/bfd/bfd_main.c \
+ vnet/bfd/bfd_protocol.c
+
+########################################
# Layer 3 protocol: IPSec
########################################
if WITH_IPSEC
diff --git a/vnet/vnet/api_errno.h b/vnet/vnet/api_errno.h
index 50d6f731035..65e3e59121e 100644
--- a/vnet/vnet/api_errno.h
+++ b/vnet/vnet/api_errno.h
@@ -90,7 +90,9 @@ _(EXCEEDED_NUMBER_OF_PORTS_CAPACITY, -96, "Operation would exceed capacity of nu
_(INVALID_ADDRESS_FAMILY, -97, "Invalid address family") \
_(INVALID_SUB_SW_IF_INDEX, -98, "Invalid sub-interface sw_if_index") \
_(TABLE_TOO_BIG, -99, "Table too big") \
-_(CANNOT_ENABLE_DISABLE_FEATURE, -100, "Cannot enable/disable feature")
+_(CANNOT_ENABLE_DISABLE_FEATURE, -100, "Cannot enable/disable feature") \
+_(BFD_EEXIST, -101, "Duplicate BFD session") \
+_(BFD_NOENT, -102, "No such BFD session")
typedef enum
{
diff --git a/vnet/vnet/bfd/bfd_api.h b/vnet/vnet/bfd/bfd_api.h
new file mode 100644
index 00000000000..cfcd04f3f50
--- /dev/null
+++ b/vnet/vnet/bfd/bfd_api.h
@@ -0,0 +1,46 @@
+/*
+ * Copyright (c) 2011-2016 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
+ * @brief BFD global declarations
+ */
+#ifndef __included_bfd_api_h__
+#define __included_bfd_api_h__
+
+#include <vnet/api_errno.h>
+#include <vnet/vnet.h>
+#include <vnet/ip/ip6_packet.h>
+#include <vnet/bfd/bfd_udp.h>
+
+vnet_api_error_t bfd_udp_add_session (u32 sw_if_index, u32 desired_min_tx_us,
+ u32 required_min_rx_us, u8 detect_mult,
+ const ip46_address_t * local_addr,
+ const ip46_address_t * peer_addr);
+
+vnet_api_error_t bfd_udp_del_session (u32 sw_if_index,
+ const ip46_address_t * local_addr,
+ const ip46_address_t * peer_addr);
+
+vnet_api_error_t bfd_session_set_flags (u32 bs_index, u8 admin_up_down);
+
+#endif /* __included_bfd_api_h__ */
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */
diff --git a/vnet/vnet/bfd/bfd_debug.h b/vnet/vnet/bfd/bfd_debug.h
new file mode 100644
index 00000000000..c11e6d9f499
--- /dev/null
+++ b/vnet/vnet/bfd/bfd_debug.h
@@ -0,0 +1,80 @@
+/*
+ * Copyright (c) 2011-2016 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
+ * @brief BFD global declarations
+ */
+#ifndef __included_bfd_debug_h__
+#define __included_bfd_debug_h__
+
+/* controls debug prints */
+#define BFD_DEBUG (0)
+
+#if BFD_DEBUG
+#define BFD_DEBUG_FILE_DEF \
+ static const char *__file = NULL; \
+ if (!__file) \
+ { \
+ __file = strrchr (__FILE__, '/'); \
+ if (__file) \
+ { \
+ ++__file; \
+ } \
+ else \
+ { \
+ __file = __FILE__; \
+ } \
+ }
+
+#define BFD_DBG(fmt, ...) \
+ do \
+ { \
+ BFD_DEBUG_FILE_DEF \
+ u8 *_s = NULL; \
+ vlib_main_t *vm = vlib_get_main (); \
+ _s = format (_s, "%6.02f:DBG:%s:%d:%s():" fmt, vlib_time_now (vm), \
+ __file, __LINE__, __func__, ##__VA_ARGS__); \
+ printf ("%s\n", _s); \
+ vec_free (_s); \
+ } \
+ while (0);
+
+#define BFD_ERR(fmt, ...) \
+ do \
+ { \
+ BFD_DEBUG_FILE_DEF \
+ u8 *_s = NULL; \
+ vlib_main_t *vm = vlib_get_main (); \
+ _s = format (_s, "%6.02f:ERR:%s:%d:%s():" fmt, vlib_time_now (vm), \
+ __file, __LINE__, __func__, ##__VA_ARGS__); \
+ printf ("%s\n", _s); \
+ vec_free (_s); \
+ } \
+ while (0);
+
+#else
+#define BFD_DBG(...)
+#define BFD_ERR(...)
+#endif
+
+#endif /* __included_bfd_debug_h__ */
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */
diff --git a/vnet/vnet/bfd/bfd_doc.md b/vnet/vnet/bfd/bfd_doc.md
new file mode 100644
index 00000000000..1333ed77b7e
--- /dev/null
+++ b/vnet/vnet/bfd/bfd_doc.md
@@ -0,0 +1 @@
+TODO
diff --git a/vnet/vnet/bfd/bfd_main.c b/vnet/vnet/bfd/bfd_main.c
new file mode 100644
index 00000000000..a72d6fed9bb
--- /dev/null
+++ b/vnet/vnet/bfd/bfd_main.c
@@ -0,0 +1,928 @@
+/*
+ * Copyright (c) 2011-2016 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
+ * @brief BFD nodes implementation
+ */
+
+#include <vppinfra/random.h>
+#include <vppinfra/error.h>
+#include <vppinfra/hash.h>
+#include <vnet/ethernet/ethernet.h>
+#include <vnet/ethernet/packet.h>
+#include <vnet/bfd/bfd_debug.h>
+#include <vnet/bfd/bfd_protocol.h>
+#include <vnet/bfd/bfd_main.h>
+
+static u64
+bfd_us_to_clocks (bfd_main_t * bm, u64 us)
+{
+ return bm->cpu_cps * ((f64) us / USEC_PER_SECOND);
+}
+
+static vlib_node_registration_t bfd_process_node;
+
+typedef enum
+{
+#define F(t, n) BFD_OUTPUT_##t,
+ foreach_bfd_transport (F)
+#undef F
+ BFD_OUTPUT_N_NEXT,
+} bfd_output_next_t;
+
+static u32 bfd_next_index_by_transport[] = {
+#define F(t, n) [BFD_TRANSPORT_##t] = BFD_OUTPUT_##t,
+ foreach_bfd_transport (F)
+#undef F
+};
+
+/*
+ * We actually send all bfd pkts to the "error" node after scanning
+ * them, so the graph node has only one next-index. The "error-drop"
+ * node automatically bumps our per-node packet counters for us.
+ */
+typedef enum
+{
+ BFD_INPUT_NEXT_NORMAL,
+ BFD_INPUT_N_NEXT,
+} bfd_input_next_t;
+
+static void bfd_on_state_change (bfd_main_t * bm, bfd_session_t * bs,
+ u64 now);
+
+static void
+bfd_set_defaults (bfd_main_t * bm, bfd_session_t * bs)
+{
+ bs->local_state = BFD_STATE_down;
+ bs->local_diag = BFD_DIAG_CODE_no_diag;
+ bs->remote_state = BFD_STATE_down;
+ bs->local_demand = 0;
+ bs->remote_discr = 0;
+ bs->desired_min_tx_us = BFD_DEFAULT_DESIRED_MIN_TX_US;
+ bs->desired_min_tx_clocks = bfd_us_to_clocks (bm, bs->desired_min_tx_us);
+ bs->remote_min_rx_us = 1;
+ bs->remote_demand = 0;
+}
+
+static void
+bfd_set_diag (bfd_session_t * bs, bfd_diag_code_e code)
+{
+ if (bs->local_diag != code)
+ {
+ BFD_DBG ("set local_diag, bs_idx=%d: '%d:%s'", bs->bs_idx, code,
+ bfd_diag_code_string (code));
+ bs->local_diag = code;
+ }
+}
+
+static void
+bfd_set_state (bfd_main_t * bm, bfd_session_t * bs, bfd_state_e new_state)
+{
+ if (bs->local_state != new_state)
+ {
+ BFD_DBG ("Change state, bs_idx=%d: %s->%s", bs->bs_idx,
+ bfd_state_string (bs->local_state),
+ bfd_state_string (new_state));
+ bs->local_state = new_state;
+ bfd_on_state_change (bm, bs, clib_cpu_time_now ());
+ }
+}
+
+static void
+bfd_recalc_tx_interval (bfd_main_t * bm, bfd_session_t * bs)
+{
+ if (!bs->local_demand)
+ {
+ bs->transmit_interval_clocks =
+ clib_max (bs->desired_min_tx_clocks, bs->remote_min_rx_clocks);
+ }
+ else
+ {
+ /* TODO */
+ }
+ BFD_DBG ("Recalculated transmit interval %lu clocks/%.2fs",
+ bs->transmit_interval_clocks,
+ bs->transmit_interval_clocks / bm->cpu_cps);
+}
+
+static void
+bfd_calc_next_tx (bfd_main_t * bm, bfd_session_t * bs, u64 now)
+{
+ if (!bs->local_demand)
+ {
+ if (bs->local_detect_mult > 1)
+ {
+ /* common case - 75-100% of transmit interval */
+ bs->tx_timeout_clocks = now +
+ (1 - .25 * (random_f64 (&bm->random_seed))) *
+ bs->transmit_interval_clocks;
+ }
+ else
+ {
+ /* special case - 75-90% of transmit interval */
+ bs->tx_timeout_clocks =
+ now +
+ (.9 - .15 * (random_f64 (&bm->random_seed))) *
+ bs->transmit_interval_clocks;
+ }
+ }
+ else
+ {
+ /* TODO */
+ }
+ if (bs->tx_timeout_clocks)
+ {
+ BFD_DBG ("Next transmit in %lu clocks/%.02fs@%lu",
+ bs->tx_timeout_clocks - now,
+ (bs->tx_timeout_clocks - now) / bm->cpu_cps,
+ bs->tx_timeout_clocks);
+ }
+}
+
+static void
+bfd_recalc_detection_time (bfd_main_t * bm, bfd_session_t * bs)
+{
+ if (!bs->local_demand)
+ {
+ bs->detection_time_clocks =
+ bs->remote_detect_mult *
+ bfd_us_to_clocks (bm, clib_max (bs->required_min_rx_us,
+ bs->remote_desired_min_tx_us));
+ }
+ else
+ {
+ bs->detection_time_clocks =
+ bs->local_detect_mult *
+ bfd_us_to_clocks (bm,
+ clib_max (bs->desired_min_tx_us,
+ bs->remote_min_rx_us));
+ }
+ BFD_DBG ("Recalculated detection time %lu clocks/%.2fs",
+ bs->detection_time_clocks,
+ bs->detection_time_clocks / bm->cpu_cps);
+}
+
+static void
+bfd_set_timer (bfd_main_t * bm, bfd_session_t * bs, u64 now)
+{
+ u64 next = 0;
+ u64 rx_timeout = 0;
+ if (BFD_STATE_up == bs->local_state)
+ {
+ rx_timeout = bs->last_rx_clocks + bs->detection_time_clocks;
+ }
+ if (bs->tx_timeout_clocks && rx_timeout)
+ {
+ next = clib_min (bs->tx_timeout_clocks, rx_timeout);
+ }
+ else if (bs->tx_timeout_clocks)
+ {
+ next = bs->tx_timeout_clocks;
+ }
+ else if (rx_timeout)
+ {
+ next = rx_timeout;
+ }
+ BFD_DBG ("bs_idx=%u, tx_timeout=%lu, rx_timeout=%lu, next=%s", bs->bs_idx,
+ bs->tx_timeout_clocks, rx_timeout,
+ next == bs->tx_timeout_clocks ? "tx" : "rx");
+ if (next && (next < bs->wheel_time_clocks || !bs->wheel_time_clocks))
+ {
+ if (bs->wheel_time_clocks)
+ {
+ timing_wheel_delete (&bm->wheel, bs->bs_idx);
+ }
+ bs->wheel_time_clocks = next;
+ BFD_DBG ("timing_wheel_insert(%p, %lu (%ld clocks/%.2fs in the "
+ "future), %u);",
+ &bm->wheel, bs->wheel_time_clocks,
+ (i64) bs->wheel_time_clocks - clib_cpu_time_now (),
+ (i64) (bs->wheel_time_clocks - clib_cpu_time_now ()) /
+ bm->cpu_cps, bs->bs_idx);
+ timing_wheel_insert (&bm->wheel, bs->wheel_time_clocks, bs->bs_idx);
+ }
+}
+
+static void
+bfd_set_desired_min_tx (bfd_main_t * bm, bfd_session_t * bs, u64 now,
+ u32 desired_min_tx_us)
+{
+ bs->desired_min_tx_us = desired_min_tx_us;
+ bs->desired_min_tx_clocks = bfd_us_to_clocks (bm, bs->desired_min_tx_us);
+ BFD_DBG ("Set desired min tx to %uus/%lu clocks/%.2fs",
+ bs->desired_min_tx_us, bs->desired_min_tx_clocks,
+ bs->desired_min_tx_clocks / bm->cpu_cps);
+ bfd_recalc_detection_time (bm, bs);
+ bfd_recalc_tx_interval (bm, bs);
+ bfd_calc_next_tx (bm, bs, now);
+ bfd_set_timer (bm, bs, now);
+}
+
+void
+bfd_session_start (bfd_main_t * bm, bfd_session_t * bs)
+{
+ BFD_DBG ("%U", format_bfd_session, bs);
+ bfd_recalc_tx_interval (bm, bs);
+ vlib_process_signal_event (bm->vlib_main, bm->bfd_process_node_index,
+ BFD_EVENT_NEW_SESSION, bs->bs_idx);
+}
+
+vnet_api_error_t
+bfd_del_session (uword bs_idx)
+{
+ const bfd_main_t *bm = &bfd_main;
+ if (!pool_is_free_index (bm->sessions, bs_idx))
+ {
+ bfd_session_t *bs = pool_elt_at_index (bm->sessions, bs_idx);
+ pool_put (bm->sessions, bs);
+ return 0;
+ }
+ else
+ {
+ BFD_ERR ("no such session");
+ return VNET_API_ERROR_BFD_NOENT;
+ }
+ return 0;
+}
+
+const char *
+bfd_diag_code_string (bfd_diag_code_e diag)
+{
+#define F(n, t, s) \
+ case BFD_DIAG_CODE_NAME (t): \
+ return s;
+ switch (diag)
+ {
+ foreach_bfd_diag_code (F)}
+ return "UNKNOWN";
+#undef F
+}
+
+const char *
+bfd_state_string (bfd_state_e state)
+{
+#define F(n, t, s) \
+ case BFD_STATE_NAME (t): \
+ return s;
+ switch (state)
+ {
+ foreach_bfd_state (F)}
+ return "UNKNOWN";
+#undef F
+}
+
+vnet_api_error_t
+bfd_session_set_flags (u32 bs_idx, u8 admin_up_down)
+{
+ bfd_main_t *bm = &bfd_main;
+ if (pool_is_free_index (bm->sessions, bs_idx))
+ {
+ BFD_ERR ("invalid bs_idx=%u", bs_idx);
+ return VNET_API_ERROR_BFD_NOENT;
+ }
+ bfd_session_t *bs = pool_elt_at_index (bm->sessions, bs_idx);
+ if (admin_up_down)
+ {
+ bfd_set_state (bm, bs, BFD_STATE_down);
+ }
+ else
+ {
+ bfd_set_diag (bs, BFD_DIAG_CODE_neighbor_sig_down);
+ bfd_set_state (bm, bs, BFD_STATE_admin_down);
+ }
+ return 0;
+}
+
+u8 *
+bfd_input_format_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 *);
+ const bfd_input_trace_t *t = va_arg (*args, bfd_input_trace_t *);
+ const bfd_pkt_t *pkt = (bfd_pkt_t *) t->data;
+ if (t->len > STRUCT_SIZE_OF (bfd_pkt_t, head))
+ {
+ s = format (s, "BFD v%u, diag=%u(%s), state=%u(%s),\n"
+ " flags=(P:%u, F:%u, C:%u, A:%u, D:%u, M:%u), detect_mult=%u, "
+ "length=%u\n",
+ bfd_pkt_get_version (pkt), bfd_pkt_get_diag_code (pkt),
+ bfd_diag_code_string (bfd_pkt_get_diag_code (pkt)),
+ bfd_pkt_get_state (pkt),
+ bfd_state_string (bfd_pkt_get_state (pkt)),
+ bfd_pkt_get_poll (pkt), bfd_pkt_get_final (pkt),
+ bfd_pkt_get_control_plane_independent (pkt),
+ bfd_pkt_get_auth_present (pkt), bfd_pkt_get_demand (pkt),
+ bfd_pkt_get_multipoint (pkt), pkt->head.detect_mult,
+ pkt->head.length);
+ if (t->len >= sizeof (bfd_pkt_t)
+ && pkt->head.length >= sizeof (bfd_pkt_t))
+ {
+ s = format (s, " my discriminator: %u\n", pkt->my_disc);
+ s = format (s, " your discriminator: %u\n", pkt->your_disc);
+ s = format (s, " desired min tx interval: %u\n",
+ clib_net_to_host_u32 (pkt->des_min_tx));
+ s = format (s, " required min rx interval: %u\n",
+ clib_net_to_host_u32 (pkt->req_min_rx));
+ s = format (s, " required min echo rx interval: %u\n",
+ clib_net_to_host_u32 (pkt->req_min_echo_rx));
+ }
+ }
+
+ return s;
+}
+
+static void
+bfd_on_state_change (bfd_main_t * bm, bfd_session_t * bs, u64 now)
+{
+ BFD_DBG ("State changed: %U", format_bfd_session, bs);
+ bfd_event (bm, bs);
+ switch (bs->local_state)
+ {
+ case BFD_STATE_admin_down:
+ bfd_set_desired_min_tx (bm, bs, now,
+ clib_max (bs->config_desired_min_tx_us,
+ BFD_DEFAULT_DESIRED_MIN_TX_US));
+ break;
+ case BFD_STATE_down:
+ bfd_set_desired_min_tx (bm, bs, now,
+ clib_max (bs->config_desired_min_tx_us,
+ BFD_DEFAULT_DESIRED_MIN_TX_US));
+ break;
+ case BFD_STATE_init:
+ bfd_set_desired_min_tx (bm, bs, now,
+ clib_max (bs->config_desired_min_tx_us,
+ BFD_DEFAULT_DESIRED_MIN_TX_US));
+ break;
+ case BFD_STATE_up:
+ bfd_set_desired_min_tx (bm, bs, now, bs->config_desired_min_tx_us);
+ break;
+ }
+}
+
+static void
+bfd_add_transport_layer (vlib_main_t * vm, vlib_buffer_t * b,
+ bfd_session_t * bs)
+{
+ switch (bs->transport)
+ {
+ case BFD_TRANSPORT_UDP4:
+ /* fallthrough */
+ case BFD_TRANSPORT_UDP6:
+ BFD_DBG ("Transport bfd via udp, bs_idx=%u", bs->bs_idx);
+ bfd_add_udp_transport (vm, b, &bs->udp);
+ break;
+ }
+}
+
+static vlib_buffer_t *
+bfd_create_frame (vlib_main_t * vm, vlib_node_runtime_t * rt,
+ bfd_session_t * bs)
+{
+ u32 bi;
+ if (vlib_buffer_alloc (vm, &bi, 1) != 1)
+ {
+ clib_warning ("buffer allocation failure");
+ return NULL;
+ }
+
+ vlib_buffer_t *b = vlib_get_buffer (vm, bi);
+ ASSERT (b->current_data == 0);
+
+ u32 *to_next;
+ u32 n_left_to_next;
+
+ vlib_get_next_frame (vm, rt, bfd_next_index_by_transport[bs->transport],
+ to_next, n_left_to_next);
+
+ to_next[0] = bi;
+ n_left_to_next -= 1;
+
+ vlib_put_next_frame (vm, rt, bfd_next_index_by_transport[bs->transport],
+ n_left_to_next);
+ return b;
+}
+
+static void
+bfd_init_control_frame (vlib_buffer_t * b, bfd_session_t * bs)
+{
+ bfd_pkt_t *pkt = vlib_buffer_get_current (b);
+ const u32 bfd_length = 24;
+ memset (pkt, 0, sizeof (*pkt));
+
+ bfd_pkt_set_version (pkt, 1);
+ bfd_pkt_set_diag_code (pkt, bs->local_diag);
+ bfd_pkt_set_state (pkt, bs->local_state);
+ if (bs->local_demand && BFD_STATE_up == bs->local_state &&
+ BFD_STATE_up == bs->remote_state)
+ {
+ bfd_pkt_set_demand (pkt);
+ }
+ pkt->head.detect_mult = bs->local_detect_mult;
+ pkt->head.length = clib_host_to_net_u32 (bfd_length);
+ pkt->my_disc = bs->local_discr;
+ pkt->your_disc = bs->remote_discr;
+ pkt->des_min_tx = clib_host_to_net_u32 (bs->desired_min_tx_us);
+ pkt->req_min_rx = clib_host_to_net_u32 (bs->required_min_rx_us);
+ pkt->req_min_echo_rx = clib_host_to_net_u32 (0); /* FIXME */
+ b->current_length = bfd_length;
+}
+
+static void
+bfd_send_periodic (vlib_main_t * vm, vlib_node_runtime_t * rt,
+ bfd_main_t * bm, bfd_session_t * bs, u64 now)
+{
+ if (!bs->remote_min_rx_us)
+ {
+ BFD_DBG
+ ("bfd.RemoteMinRxInterval is zero, not sending periodic control "
+ "frame");
+ return;
+ }
+ /* FIXME
+ A system MUST NOT periodically transmit BFD Control packets if Demand
+ mode is active on the remote system (bfd.RemoteDemandMode is 1,
+ bfd.SessionState is Up, and bfd.RemoteSessionState is Up) and a Poll
+ Sequence is not being transmitted.
+ */
+ if (now >= bs->tx_timeout_clocks)
+ {
+ BFD_DBG ("Send periodic control frame for bs_idx=%lu", bs->bs_idx);
+ vlib_buffer_t *b = bfd_create_frame (vm, rt, bs);
+ if (!b)
+ {
+ return;
+ }
+ bfd_init_control_frame (b, bs);
+ bfd_add_transport_layer (vm, b, bs);
+ bfd_calc_next_tx (bm, bs, now);
+ }
+ else
+ {
+ BFD_DBG ("No need to send control frame now");
+ }
+ bfd_set_timer (bm, bs, now);
+}
+
+void
+bfd_send_final (vlib_main_t * vm, vlib_buffer_t * b, bfd_session_t * bs)
+{
+ BFD_DBG ("Send final control frame for bs_idx=%lu", bs->bs_idx);
+ bfd_init_control_frame (b, bs);
+ bfd_pkt_set_final (vlib_buffer_get_current (b));
+ bfd_add_transport_layer (vm, b, bs);
+}
+
+static void
+bfd_check_rx_timeout (bfd_main_t * bm, bfd_session_t * bs, u64 now)
+{
+ if (bs->last_rx_clocks + bs->detection_time_clocks < now)
+ {
+ BFD_DBG ("Rx timeout, session goes down");
+ bfd_set_diag (bs, BFD_DIAG_CODE_det_time_exp);
+ bfd_set_state (bm, bs, BFD_STATE_down);
+ }
+}
+
+void
+bfd_on_timeout (vlib_main_t * vm, vlib_node_runtime_t * rt, bfd_main_t * bm,
+ bfd_session_t * bs, u64 now)
+{
+ BFD_DBG ("Timeout for bs_idx=%lu", bs->bs_idx);
+ switch (bs->local_state)
+ {
+ case BFD_STATE_admin_down:
+ BFD_ERR ("Unexpected timeout when in %s state",
+ bfd_state_string (bs->local_state));
+ abort ();
+ break;
+ case BFD_STATE_down:
+ bfd_send_periodic (vm, rt, bm, bs, now);
+ break;
+ case BFD_STATE_init:
+ BFD_ERR ("Unexpected timeout when in %s state",
+ bfd_state_string (bs->local_state));
+ abort ();
+ break;
+ case BFD_STATE_up:
+ bfd_check_rx_timeout (bm, bs, now);
+ bfd_send_periodic (vm, rt, bm, bs, now);
+ break;
+ }
+}
+
+/*
+ * bfd input routine
+ */
+bfd_error_t
+bfd_input (vlib_main_t * vm, vlib_buffer_t * b0, u32 bi0)
+{
+ // bfd_main_t *bm = &bfd_main;
+ bfd_error_t e;
+
+ /* find our interface */
+ bfd_session_t *s = NULL;
+ // bfd_get_intf (lm, vnet_buffer (b0)->sw_if_index[VLIB_RX]);
+
+ if (!s)
+ {
+ /* bfd disabled on this interface, we're done */
+ return BFD_ERROR_DISABLED;
+ }
+
+ /* Actually scan the packet */
+ e = BFD_ERROR_NONE; // bfd_packet_scan (lm, n, vlib_buffer_get_current (b0));
+
+ return e;
+}
+
+/*
+ * bfd process node function
+ */
+static uword
+bfd_process (vlib_main_t * vm, vlib_node_runtime_t * rt, vlib_frame_t * f)
+{
+ bfd_main_t *bm = &bfd_main;
+ u32 *expired = 0;
+ uword event_type, *event_data = 0;
+
+ /* So we can send events to the bfd process */
+ bm->bfd_process_node_index = bfd_process_node.index;
+
+ while (1)
+ {
+ u64 now = clib_cpu_time_now ();
+ u64 next_expire = timing_wheel_next_expiring_elt_time (&bm->wheel);
+ BFD_DBG ("timing_wheel_next_expiring_elt_time(%p) returns %lu",
+ &bm->wheel, next_expire);
+ if ((i64) next_expire < 0)
+ {
+ BFD_DBG ("wait for event without timeout");
+ (void) vlib_process_wait_for_event (vm);
+ }
+ else
+ {
+ f64 timeout = ((i64) next_expire - (i64) now) / bm->cpu_cps;
+ BFD_DBG ("wait for event with timeout %.02f", timeout);
+ ASSERT (timeout > 0);
+ (void) vlib_process_wait_for_event_or_clock (vm, timeout);
+ }
+ event_type = vlib_process_get_events (vm, &event_data);
+ now = clib_cpu_time_now ();
+ switch (event_type)
+ {
+ case ~0: /* no events => timeout */
+ /* nothing to do here */
+ break;
+ case BFD_EVENT_RESCHEDULE:
+ /* nothing to do here - reschedule is done automatically after
+ * each event or timeout */
+ break;
+ case BFD_EVENT_NEW_SESSION:
+ do
+ {
+ bfd_session_t *bs =
+ pool_elt_at_index (bm->sessions, *event_data);
+ bfd_send_periodic (vm, rt, bm, bs, now);
+ }
+ while (0);
+ break;
+ default:
+ clib_warning ("BUG: event type 0x%wx", event_type);
+ break;
+ }
+ BFD_DBG ("advancing wheel, now is %lu", now);
+ BFD_DBG ("timing_wheel_advance (%p, %lu, %p, 0);", &bm->wheel, now,
+ expired);
+ expired = timing_wheel_advance (&bm->wheel, now, expired, 0);
+ BFD_DBG ("Expired %d elements", vec_len (expired));
+ u32 *p = NULL;
+ vec_foreach (p, expired)
+ {
+ const u32 bs_idx = *p;
+ if (!pool_is_free_index (bm->sessions, bs_idx))
+ {
+ bfd_session_t *bs = pool_elt_at_index (bm->sessions, bs_idx);
+ bs->wheel_time_clocks = 0; /* no longer scheduled */
+ bfd_on_timeout (vm, rt, bm, bs, now);
+ }
+ }
+ if (expired)
+ {
+ _vec_len (expired) = 0;
+ }
+ if (event_data)
+ {
+ _vec_len (event_data) = 0;
+ }
+ }
+
+ return 0;
+}
+
+/*
+ * bfd process node declaration
+ */
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE (bfd_process_node, static) = {
+ .function = bfd_process,
+ .type = VLIB_NODE_TYPE_PROCESS,
+ .name = "bfd-process",
+ .n_next_nodes = BFD_OUTPUT_N_NEXT,
+ .next_nodes =
+ {
+#define F(t, n) [BFD_OUTPUT_##t] = n,
+ foreach_bfd_transport (F)
+#undef F
+ },
+};
+/* *INDENT-ON* */
+
+static clib_error_t *
+bfd_sw_interface_up_down (vnet_main_t * vnm, u32 sw_if_index, u32 flags)
+{
+ // bfd_main_t *bm = &bfd_main;
+ // vnet_hw_interface_t *hi = vnet_get_sup_hw_interface (vnm, sw_if_index);
+ if (!(flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP))
+ {
+ /* TODO */
+ }
+ return 0;
+}
+
+VNET_SW_INTERFACE_ADMIN_UP_DOWN_FUNCTION (bfd_sw_interface_up_down);
+
+static clib_error_t *
+bfd_hw_interface_up_down (vnet_main_t * vnm, u32 hw_if_index, u32 flags)
+{
+ // bfd_main_t *bm = &bfd_main;
+ if (flags & VNET_HW_INTERFACE_FLAG_LINK_UP)
+ {
+ /* TODO */
+ }
+ return 0;
+}
+
+VNET_HW_INTERFACE_LINK_UP_DOWN_FUNCTION (bfd_hw_interface_up_down);
+
+/*
+ * setup function
+ */
+static clib_error_t *
+bfd_main_init (vlib_main_t * vm)
+{
+ bfd_main_t *bm = &bfd_main;
+ bm->random_seed = random_default_seed ();
+ bm->vlib_main = vm;
+ bm->vnet_main = vnet_get_main ();
+ memset (&bm->wheel, 0, sizeof (bm->wheel));
+ bm->cpu_cps = 2590000000; // vm->clib_time.clocks_per_second;
+ BFD_DBG ("cps is %.2f", bm->cpu_cps);
+ const u64 now = clib_cpu_time_now ();
+ timing_wheel_init (&bm->wheel, now, bm->cpu_cps);
+
+ return 0;
+}
+
+VLIB_INIT_FUNCTION (bfd_main_init);
+
+bfd_session_t *
+bfd_get_session (bfd_main_t * bm, bfd_transport_t t)
+{
+ bfd_session_t *result;
+ pool_get (bm->sessions, result);
+ result->bs_idx = result - bm->sessions;
+ result->transport = t;
+ result->local_discr = random_u32 (&bm->random_seed);
+ bfd_set_defaults (bm, result);
+ hash_set (bm->session_by_disc, result->local_discr, result->bs_idx);
+ return result;
+}
+
+void
+bfd_put_session (bfd_main_t * bm, bfd_session_t * bs)
+{
+ hash_unset (bm->session_by_disc, bs->local_discr);
+ pool_put (bm->sessions, bs);
+}
+
+bfd_session_t *
+bfd_find_session_by_idx (bfd_main_t * bm, uword bs_idx)
+{
+ if (!pool_is_free_index (bm->sessions, bs_idx))
+ {
+ return pool_elt_at_index (bm->sessions, bs_idx);
+ }
+ return NULL;
+}
+
+bfd_session_t *
+bfd_find_session_by_disc (bfd_main_t * bm, u32 disc)
+{
+ uword *p = hash_get (bfd_main.session_by_disc, disc);
+ if (p)
+ {
+ return pool_elt_at_index (bfd_main.sessions, *p);
+ }
+ return NULL;
+}
+
+/**
+ * @brief verify bfd packet - common checks
+ *
+ * @param pkt
+ *
+ * @return 1 if bfd packet is valid
+ */
+int
+bfd_verify_pkt_common (const bfd_pkt_t * pkt)
+{
+ if (1 != bfd_pkt_get_version (pkt))
+ {
+ BFD_ERR ("BFD verification failed - unexpected version: '%d'",
+ bfd_pkt_get_version (pkt));
+ return 0;
+ }
+ if (pkt->head.length < sizeof (bfd_pkt_t) ||
+ (bfd_pkt_get_auth_present (pkt) &&
+ pkt->head.length < sizeof (bfd_pkt_with_auth_t)))
+ {
+ BFD_ERR ("BFD verification failed - unexpected length: '%d' (auth "
+ "present: %d)",
+ pkt->head.length, bfd_pkt_get_auth_present (pkt));
+ return 0;
+ }
+ if (!pkt->head.detect_mult)
+ {
+ BFD_ERR ("BFD verification failed - unexpected detect-mult: '%d'",
+ pkt->head.detect_mult);
+ return 0;
+ }
+ if (bfd_pkt_get_multipoint (pkt))
+ {
+ BFD_ERR ("BFD verification failed - unexpected multipoint: '%d'",
+ bfd_pkt_get_multipoint (pkt));
+ return 0;
+ }
+ if (!pkt->my_disc)
+ {
+ BFD_ERR ("BFD verification failed - unexpected my-disc: '%d'",
+ pkt->my_disc);
+ return 0;
+ }
+ if (!pkt->your_disc)
+ {
+ const u8 pkt_state = bfd_pkt_get_state (pkt);
+ if (pkt_state != BFD_STATE_down && pkt_state != BFD_STATE_admin_down)
+ {
+ BFD_ERR ("BFD verification failed - unexpected state: '%s' "
+ "(your-disc is zero)", bfd_state_string (pkt_state));
+ return 0;
+ }
+ }
+ return 1;
+}
+
+/**
+ * @brief verify bfd packet - authentication
+ *
+ * @param pkt
+ *
+ * @return 1 if bfd packet is valid
+ */
+int
+bfd_verify_pkt_session (const bfd_pkt_t * pkt, u16 pkt_size,
+ const bfd_session_t * bs)
+{
+ const bfd_pkt_with_auth_t *with_auth = (bfd_pkt_with_auth_t *) pkt;
+ if (!bfd_pkt_get_auth_present (pkt))
+ {
+ if (pkt_size > sizeof (*pkt))
+ {
+ BFD_ERR ("BFD verification failed - unexpected packet size '%d' "
+ "(auth not present)", pkt_size);
+ return 0;
+ }
+ }
+ else
+ {
+ if (!with_auth->auth.type)
+ {
+ BFD_ERR ("BFD verification failed - unexpected auth type: '%d'",
+ with_auth->auth.type);
+ return 0;
+ }
+ /* TODO FIXME - implement the actual verification */
+ }
+ return 1;
+}
+
+void
+bfd_consume_pkt (bfd_main_t * bm, const bfd_pkt_t * pkt, u32 bs_idx)
+{
+ bfd_session_t *bs = bfd_find_session_by_idx (bm, bs_idx);
+ if (!bs)
+ {
+ return;
+ }
+ BFD_DBG ("Scanning bfd packet, bs_idx=%d", bs->bs_idx);
+ bs->remote_discr = pkt->my_disc;
+ bs->remote_state = bfd_pkt_get_state (pkt);
+ bs->remote_demand = bfd_pkt_get_demand (pkt);
+ bs->remote_min_rx_us = clib_net_to_host_u32 (pkt->req_min_rx);
+ bs->remote_min_rx_clocks = bfd_us_to_clocks (bm, bs->remote_min_rx_us);
+ BFD_DBG ("Set remote min rx to %lu clocks/%.2fs", bs->remote_min_rx_clocks,
+ bs->remote_min_rx_clocks / bm->cpu_cps);
+ bs->remote_desired_min_tx_us = clib_net_to_host_u32 (pkt->des_min_tx);
+ bs->remote_detect_mult = pkt->head.detect_mult;
+ bfd_recalc_detection_time (bm, bs);
+ bs->last_rx_clocks = clib_cpu_time_now ();
+ /* FIXME
+ If the Required Min Echo RX Interval field is zero, the
+ transmission of Echo packets, if any, MUST cease.
+
+ If a Poll Sequence is being transmitted by the local system and
+ the Final (F) bit in the received packet is set, the Poll Sequence
+ MUST be terminated.
+ */
+ /* FIXME 6.8.2 */
+ /* FIXME 6.8.4 */
+ if (BFD_STATE_admin_down == bs->local_state)
+ return;
+ if (BFD_STATE_admin_down == bs->remote_state)
+ {
+ bfd_set_diag (bs, BFD_DIAG_CODE_neighbor_sig_down);
+ bfd_set_state (bm, bs, BFD_STATE_down);
+ }
+ else if (BFD_STATE_down == bs->local_state)
+ {
+ if (BFD_STATE_down == bs->remote_state)
+ {
+ bfd_set_state (bm, bs, BFD_STATE_init);
+ }
+ else if (BFD_STATE_init == bs->remote_state)
+ {
+ bfd_set_state (bm, bs, BFD_STATE_up);
+ }
+ }
+ else if (BFD_STATE_init == bs->local_state)
+ {
+ if (BFD_STATE_up == bs->remote_state ||
+ BFD_STATE_init == bs->remote_state)
+ {
+ bfd_set_state (bm, bs, BFD_STATE_up);
+ }
+ }
+ else /* BFD_STATE_up == bs->local_state */
+ {
+ if (BFD_STATE_down == bs->remote_state)
+ {
+ bfd_set_diag (bs, BFD_DIAG_CODE_neighbor_sig_down);
+ bfd_set_state (bm, bs, BFD_STATE_down);
+ }
+ }
+}
+
+u8 *
+format_bfd_session (u8 * s, va_list * args)
+{
+ const bfd_session_t *bs = va_arg (*args, bfd_session_t *);
+ return format (s, "BFD(%u): bfd.SessionState=%s, "
+ "bfd.RemoteSessionState=%s, "
+ "bfd.LocalDiscr=%u, "
+ "bfd.RemoteDiscr=%u, "
+ "bfd.LocalDiag=%s, "
+ "bfd.DesiredMinTxInterval=%u, "
+ "bfd.RequiredMinRxInterval=%u, "
+ "bfd.RemoteMinRxInterval=%u, "
+ "bfd.DemandMode=%s, "
+ "bfd.RemoteDemandMode=%s, "
+ "bfd.DetectMult=%u, ",
+ bs->bs_idx, bfd_state_string (bs->local_state),
+ bfd_state_string (bs->remote_state), bs->local_discr,
+ bs->remote_discr, bfd_diag_code_string (bs->local_diag),
+ bs->desired_min_tx_us, bs->required_min_rx_us,
+ bs->remote_min_rx_us, (bs->local_demand ? "yes" : "no"),
+ (bs->remote_demand ? "yes" : "no"), bs->local_detect_mult);
+}
+
+bfd_main_t bfd_main;
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */
diff --git a/vnet/vnet/bfd/bfd_main.h b/vnet/vnet/bfd/bfd_main.h
new file mode 100644
index 00000000000..727903bd286
--- /dev/null
+++ b/vnet/vnet/bfd/bfd_main.h
@@ -0,0 +1,218 @@
+/*
+ * Copyright (c) 2011-2016 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
+ * @brief BFD global declarations
+ */
+#ifndef __included_bfd_main_h__
+#define __included_bfd_main_h__
+
+#include <vppinfra/timing_wheel.h>
+#include <vnet/vnet.h>
+#include <vnet/bfd/bfd_protocol.h>
+#include <vnet/bfd/bfd_udp.h>
+
+#define foreach_bfd_transport(F) \
+ F (UDP4, "ip4-rewrite") \
+ F (UDP6, "ip6-rewrite")
+
+typedef enum
+{
+#define F(t, n) BFD_TRANSPORT_##t,
+ foreach_bfd_transport (F)
+#undef F
+} bfd_transport_t;
+
+#define foreach_bfd_mode(F) \
+ F (asynchronous) \
+ F (demand)
+
+typedef enum
+{
+#define F(x) BFD_MODE_##x,
+ foreach_bfd_mode (F)
+#undef F
+} bfd_mode_e;
+
+typedef struct
+{
+ /* index in bfd_main.sessions pool */
+ uword bs_idx;
+
+ /* session state */
+ bfd_state_e local_state;
+
+ /* local diagnostics */
+ bfd_diag_code_e local_diag;
+
+ /* remote session state */
+ bfd_state_e remote_state;
+
+ /* local discriminator */
+ u32 local_discr;
+
+ /* remote discriminator */
+ u32 remote_discr;
+
+ /* configured desired min tx interval (microseconds) */
+ u32 config_desired_min_tx_us;
+
+ /* desired min tx interval (microseconds) */
+ u32 desired_min_tx_us;
+
+ /* desired min tx interval (clocks) */
+ u64 desired_min_tx_clocks;
+
+ /* required min rx interval */
+ u32 required_min_rx_us;
+
+ /* remote min rx interval (microseconds) */
+ u32 remote_min_rx_us;
+
+ /* remote min rx interval (clocks) */
+ u64 remote_min_rx_clocks;
+
+ /* remote desired min tx interval */
+ u32 remote_desired_min_tx_us;
+
+ /* 1 if in demand mode, 0 otherwise */
+ u8 local_demand;
+
+ /* 1 if remote system sets demand mode, 0 otherwise */
+ u8 remote_demand;
+
+ u8 local_detect_mult;
+ u8 remote_detect_mult;
+
+ /* set to value of timer in timing wheel, 0 if not set */
+ u64 wheel_time_clocks;
+
+ /* transmit interval */
+ u64 transmit_interval_clocks;
+
+ /* next time at which to transmit a packet */
+ u64 tx_timeout_clocks;
+
+ /* timestamp of last packet received */
+ u64 last_rx_clocks;
+
+ /* detection time */
+ u64 detection_time_clocks;
+
+ /* transport type for this session */
+ bfd_transport_t transport;
+
+ union
+ {
+ bfd_udp_session_t udp;
+ };
+} bfd_session_t;
+
+typedef struct
+{
+ u32 client_index;
+ u32 client_pid;
+} event_subscriber_t;
+
+typedef struct
+{
+ /* pool of bfd sessions context data */
+ bfd_session_t *sessions;
+
+ /* timing wheel for scheduling timeouts */
+ timing_wheel_t wheel;
+
+ /* hashmap - bfd session by discriminator */
+ u32 *session_by_disc;
+
+ /* background process node index */
+ u32 bfd_process_node_index;
+
+ /* convenience variables */
+ vlib_main_t *vlib_main;
+ vnet_main_t *vnet_main;
+
+ /* cpu clocks per second */
+ f64 cpu_cps;
+
+ /* for generating random numbers */
+ u32 random_seed;
+
+ /* pool of event subscribers */
+ //event_subscriber_t *subscribers;
+
+} bfd_main_t;
+
+extern bfd_main_t bfd_main;
+
+/* Packet counters */
+#define foreach_bfd_error(F) \
+ F (NONE, "good bfd packets (processed)") \
+ F (BAD, "invalid bfd packets") \
+ F (DISABLED, "bfd packets received on disabled interfaces")
+
+typedef enum
+{
+#define F(sym, str) BFD_ERROR_##sym,
+ foreach_bfd_error (F)
+#undef F
+ BFD_N_ERROR,
+} bfd_error_t;
+
+/* bfd packet trace capture */
+typedef struct
+{
+ u32 len;
+ u8 data[400];
+} bfd_input_trace_t;
+
+enum
+{
+ BFD_EVENT_RESCHEDULE = 1,
+ BFD_EVENT_NEW_SESSION,
+} bfd_process_event_e;
+
+bfd_error_t bfd_input (vlib_main_t * vm, vlib_buffer_t * b0, u32 bi0);
+u8 *bfd_input_format_trace (u8 * s, va_list * args);
+
+bfd_session_t *bfd_get_session (bfd_main_t * bm, bfd_transport_t t);
+void bfd_put_session (bfd_main_t * bm, bfd_session_t * bs);
+bfd_session_t *bfd_find_session_by_idx (bfd_main_t * bm, uword bs_idx);
+bfd_session_t *bfd_find_session_by_disc (bfd_main_t * bm, u32 disc);
+void bfd_session_start (bfd_main_t * bm, bfd_session_t * bs);
+void bfd_consume_pkt (bfd_main_t * bm, const bfd_pkt_t * bfd, u32 bs_idx);
+int bfd_verify_pkt_common (const bfd_pkt_t * pkt);
+int bfd_verify_pkt_session (const bfd_pkt_t * pkt, u16 pkt_size,
+ const bfd_session_t * bs);
+void bfd_event (bfd_main_t * bm, bfd_session_t * bs);
+void bfd_send_final (vlib_main_t * vm, vlib_buffer_t * b, bfd_session_t * bs);
+u8 *format_bfd_session (u8 * s, va_list * args);
+
+
+#define USEC_PER_MS 1000LL
+#define USEC_PER_SECOND (1000 * USEC_PER_MS)
+
+/* default, slow transmission interval for BFD packets, per spec at least 1s */
+#define BFD_DEFAULT_DESIRED_MIN_TX_US USEC_PER_SECOND
+
+#endif /* __included_bfd_main_h__ */
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */
diff --git a/vnet/vnet/bfd/bfd_protocol.c b/vnet/vnet/bfd/bfd_protocol.c
new file mode 100644
index 00000000000..ede9536f3cf
--- /dev/null
+++ b/vnet/vnet/bfd/bfd_protocol.c
@@ -0,0 +1,74 @@
+#include <vnet/bfd/bfd_protocol.h>
+
+u8 bfd_pkt_get_version (const bfd_pkt_t *pkt)
+{
+ return pkt->head.vers_diag >> 5;
+}
+
+void bfd_pkt_set_version (bfd_pkt_t *pkt, int version)
+{
+ pkt->head.vers_diag =
+ (version << 5) | (pkt->head.vers_diag & ((1 << 5) - 1));
+}
+
+u8 bfd_pkt_get_diag_code (const bfd_pkt_t *pkt)
+{
+ return pkt->head.vers_diag & ((1 << 5) - 1);
+}
+
+void bfd_pkt_set_diag_code (bfd_pkt_t *pkt, int value)
+{
+ pkt->head.vers_diag =
+ (pkt->head.vers_diag & ~((1 << 5) - 1)) | (value & ((1 << 5) - 1));
+}
+
+u8 bfd_pkt_get_state (const bfd_pkt_t *pkt)
+{
+ return pkt->head.sta_flags >> 6;
+}
+
+void bfd_pkt_set_state (bfd_pkt_t *pkt, int value)
+{
+ pkt->head.sta_flags = (value << 6) | (pkt->head.sta_flags & ((1 << 6) - 1));
+}
+
+u8 bfd_pkt_get_poll (const bfd_pkt_t *pkt)
+{
+ return (pkt->head.sta_flags >> 5) & 1;
+}
+
+void bfd_pkt_set_final (bfd_pkt_t *pkt) { pkt->head.sta_flags |= 1 << 5; }
+
+u8 bfd_pkt_get_final (const bfd_pkt_t *pkt)
+{
+ return (pkt->head.sta_flags >> 4) & 1;
+}
+
+void bfd_pkt_set_poll (bfd_pkt_t *pkt);
+u8 bfd_pkt_get_control_plane_independent (const bfd_pkt_t *pkt)
+{
+ return (pkt->head.sta_flags >> 3) & 1;
+}
+
+void bfd_pkt_set_control_plane_independent (bfd_pkt_t *pkt);
+
+u8 bfd_pkt_get_auth_present (const bfd_pkt_t *pkt)
+{
+ return (pkt->head.sta_flags >> 2) & 1;
+}
+
+void bfd_pkt_set_auth_present (bfd_pkt_t *pkt);
+
+u8 bfd_pkt_get_demand (const bfd_pkt_t *pkt)
+{
+ return (pkt->head.sta_flags >> 1) & 1;
+}
+
+void bfd_pkt_set_demand (bfd_pkt_t *pkt) { pkt->head.sta_flags |= 1 << 1; }
+
+u8 bfd_pkt_get_multipoint (const bfd_pkt_t *pkt)
+{
+ return pkt->head.sta_flags & 1;
+}
+
+void bfd_pkt_set_multipoint (bfd_pkt_t *pkt);
diff --git a/vnet/vnet/bfd/bfd_protocol.h b/vnet/vnet/bfd/bfd_protocol.h
new file mode 100644
index 00000000000..cf751b3b89a
--- /dev/null
+++ b/vnet/vnet/bfd/bfd_protocol.h
@@ -0,0 +1,154 @@
+/*
+ * Copyright (c) 2011-2016 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_bfd_protocol_h__
+#define __included_bfd_protocol_h__
+/**
+ * @file
+ * @brief BFD protocol declarations
+ */
+
+#include <vppinfra/types.h>
+#include <vppinfra/clib.h>
+
+/* *INDENT-OFF* */
+typedef CLIB_PACKED (struct {
+ /*
+ An optional Authentication Section MAY be present:
+
+ 0 1 2 3
+ 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ | Auth Type | Auth Len | Authentication Data... |
+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ */
+ u8 type;
+ u8 len;
+ u8 data[0];
+}) bfd_auth_t;
+/* *INDENT-ON* */
+
+/* *INDENT-OFF* */
+typedef CLIB_PACKED (struct {
+ /*
+ The Mandatory Section of a BFD Control packet has the following
+ format:
+
+ 0 1 2 3
+ 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ |Vers | Diag |Sta|P|F|C|A|D|M| Detect Mult | Length |
+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ | My Discriminator |
+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ | Your Discriminator |
+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ | Desired Min TX Interval |
+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ | Required Min RX Interval |
+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ | Required Min Echo RX Interval |
+ +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
+ */
+ struct
+ {
+ u8 vers_diag;
+ u8 sta_flags;
+ u8 detect_mult;
+ u8 length;
+ } head;
+ u32 my_disc;
+ u32 your_disc;
+ u32 des_min_tx;
+ u32 req_min_rx;
+ u32 req_min_echo_rx;
+}) bfd_pkt_t;
+/* *INDENT-ON* */
+
+/* *INDENT-OFF* */
+typedef CLIB_PACKED (struct {
+ bfd_pkt_t pkt;
+ bfd_auth_t auth;
+}) bfd_pkt_with_auth_t;
+/* *INDENT-ON* */
+
+u8 bfd_pkt_get_version (const bfd_pkt_t * pkt);
+void bfd_pkt_set_version (bfd_pkt_t * pkt, int version);
+u8 bfd_pkt_get_diag_code (const bfd_pkt_t * pkt);
+void bfd_pkt_set_diag_code (bfd_pkt_t * pkt, int value);
+u8 bfd_pkt_get_state (const bfd_pkt_t * pkt);
+void bfd_pkt_set_state (bfd_pkt_t * pkt, int value);
+u8 bfd_pkt_get_poll (const bfd_pkt_t * pkt);
+void bfd_pkt_set_final (bfd_pkt_t * pkt);
+u8 bfd_pkt_get_final (const bfd_pkt_t * pkt);
+void bfd_pkt_set_poll (bfd_pkt_t * pkt);
+u8 bfd_pkt_get_control_plane_independent (const bfd_pkt_t * pkt);
+void bfd_pkt_set_control_plane_independent (bfd_pkt_t * pkt);
+u8 bfd_pkt_get_auth_present (const bfd_pkt_t * pkt);
+void bfd_pkt_set_auth_present (bfd_pkt_t * pkt);
+u8 bfd_pkt_get_demand (const bfd_pkt_t * pkt);
+void bfd_pkt_set_demand (bfd_pkt_t * pkt);
+u8 bfd_pkt_get_multipoint (const bfd_pkt_t * pkt);
+void bfd_pkt_set_multipoint (bfd_pkt_t * pkt);
+
+/* BFD diagnostic codes */
+#define foreach_bfd_diag_code(F) \
+ F (0, no_diag, "No Diagnostic") \
+ F (1, det_time_exp, "Control Detection Time Expired") \
+ F (2, echo_failed, "Echo Function Failed") \
+ F (3, neighbor_sig_down, "Neighbor Signaled Session Down") \
+ F (4, fwd_plain_reset, "Forwarding Plane Reset") \
+ F (5, path_down, "Path Down") \
+ F (6, concat_path_down, "Concatenated Path Down") \
+ F (7, admin_down, "Administratively Down") \
+ F (8, reverse_concat_path_down, "Reverse Concatenated Path Down")
+
+#define BFD_DIAG_CODE_NAME(t) BFD_DIAG_CODE_##t
+
+typedef enum
+{
+#define F(n, t, s) BFD_DIAG_CODE_NAME (t) = n,
+ foreach_bfd_diag_code (F)
+#undef F
+} bfd_diag_code_e;
+
+const char *bfd_diag_code_string (bfd_diag_code_e diag);
+
+/* BFD state values */
+#define foreach_bfd_state(F) \
+ F (0, admin_down, "AdminDown") \
+ F (1, down, "Down") \
+ F (2, init, "Init") \
+ F (3, up, "Up")
+
+#define BFD_STATE_NAME(t) BFD_STATE_##t
+
+typedef enum
+{
+#define F(n, t, s) BFD_STATE_NAME (t) = n,
+ foreach_bfd_state (F)
+#undef F
+} bfd_state_e;
+
+const char *bfd_state_string (bfd_state_e state);
+
+#endif /* __included_bfd_protocol_h__ */
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */
diff --git a/vnet/vnet/bfd/bfd_udp.c b/vnet/vnet/bfd/bfd_udp.c
new file mode 100644
index 00000000000..ded3342559c
--- /dev/null
+++ b/vnet/vnet/bfd/bfd_udp.c
@@ -0,0 +1,610 @@
+#include <vppinfra/types.h>
+#include <vlibmemory/api.h>
+#include <vlib/vlib.h>
+#include <vlib/buffer.h>
+#include <vnet/ip/format.h>
+#include <vnet/ethernet/packet.h>
+#include <vnet/ip/udp_packet.h>
+#include <vnet/ip/lookup.h>
+#include <vnet/ip/icmp46_packet.h>
+#include <vnet/ip/ip4.h>
+#include <vnet/ip/ip6.h>
+#include <vnet/ip/udp.h>
+#include <vnet/ip/ip6_packet.h>
+#include <vnet/adj/adj.h>
+#include <vnet/adj/adj_nbr.h>
+#include <vnet/bfd/bfd_debug.h>
+#include <vnet/bfd/bfd_udp.h>
+#include <vnet/bfd/bfd_main.h>
+#include <vnet/bfd/bfd_api.h>
+
+typedef struct
+{
+ bfd_main_t *bfd_main;
+ /* hashmap - bfd session index by bfd key - used for CLI/API lookup, where
+ * discriminator is unknown */
+ mhash_t bfd_session_idx_by_bfd_key;
+} bfd_udp_main_t;
+
+bfd_udp_main_t bfd_udp_main;
+
+void bfd_udp_transport_to_buffer (vlib_main_t *vm, vlib_buffer_t *b,
+ bfd_udp_session_t *bus)
+{
+ udp_header_t *udp;
+ u16 udp_length, ip_length;
+ bfd_udp_key_t *key = &bus->key;
+
+ b->flags |= VNET_BUFFER_LOCALLY_ORIGINATED;
+ if (ip46_address_is_ip4 (&key->local_addr))
+ {
+ ip4_header_t *ip4;
+ const size_t data_size = sizeof (*ip4) + sizeof (*udp);
+ vlib_buffer_advance (b, -data_size);
+ ip4 = vlib_buffer_get_current (b);
+ udp = (udp_header_t *)(ip4 + 1);
+ memset (ip4, 0, data_size);
+ ip4->ip_version_and_header_length = 0x45;
+ ip4->ttl = 255;
+ ip4->protocol = IP_PROTOCOL_UDP;
+ ip4->src_address.as_u32 = key->local_addr.ip4.as_u32;
+ ip4->dst_address.as_u32 = key->peer_addr.ip4.as_u32;
+
+ udp->src_port = clib_host_to_net_u16 (50000); /* FIXME */
+ udp->dst_port = clib_host_to_net_u16 (UDP_DST_PORT_bfd4);
+
+ /* fix ip length, checksum and udp length */
+ ip_length = vlib_buffer_length_in_chain (vm, b);
+
+ ip4->length = clib_host_to_net_u16 (ip_length);
+ ip4->checksum = ip4_header_checksum (ip4);
+
+ udp_length = ip_length - (sizeof (*ip4));
+ udp->length = clib_host_to_net_u16 (udp_length);
+ }
+ else
+ {
+ BFD_ERR ("not implemented");
+ abort ();
+ }
+}
+
+void bfd_add_udp_transport (vlib_main_t *vm, vlib_buffer_t *b,
+ bfd_udp_session_t *bus)
+{
+ vnet_buffer (b)->ip.adj_index[VLIB_RX] = bus->adj_index;
+ vnet_buffer (b)->ip.adj_index[VLIB_TX] = bus->adj_index;
+ bfd_udp_transport_to_buffer (vm, b, bus);
+}
+
+static bfd_session_t *bfd_lookup_session (bfd_udp_main_t *bum,
+ const bfd_udp_key_t *key)
+{
+ uword *p = mhash_get (&bum->bfd_session_idx_by_bfd_key, key);
+ if (p)
+ {
+ return bfd_find_session_by_idx (bum->bfd_main, *p);
+ }
+ return 0;
+}
+
+static vnet_api_error_t
+bfd_udp_add_session_internal (bfd_udp_main_t *bum, u32 sw_if_index,
+ u32 desired_min_tx_us, u32 required_min_rx_us,
+ u8 detect_mult, const ip46_address_t *local_addr,
+ const ip46_address_t *peer_addr)
+{
+ vnet_sw_interface_t *sw_if =
+ vnet_get_sw_interface (vnet_get_main (), sw_if_index);
+ /* get a pool entry and if we end up not needing it, give it back */
+ bfd_transport_t t = BFD_TRANSPORT_UDP4;
+ if (!ip46_address_is_ip4 (local_addr))
+ {
+ t = BFD_TRANSPORT_UDP6;
+ }
+ bfd_session_t *bs = bfd_get_session (bum->bfd_main, t);
+ bfd_udp_session_t *bus = &bs->udp;
+ memset (bus, 0, sizeof (*bus));
+ bfd_udp_key_t *key = &bus->key;
+ key->sw_if_index = sw_if->sw_if_index;
+ key->local_addr.as_u64[0] = local_addr->as_u64[0];
+ key->local_addr.as_u64[1] = local_addr->as_u64[1];
+ key->peer_addr.as_u64[0] = peer_addr->as_u64[0];
+ key->peer_addr.as_u64[1] = peer_addr->as_u64[1];
+ const bfd_session_t *tmp = bfd_lookup_session (bum, key);
+ if (tmp)
+ {
+ BFD_ERR ("duplicate bfd-udp session, existing bs_idx=%d", tmp->bs_idx);
+ bfd_put_session (bum->bfd_main, bs);
+ return VNET_API_ERROR_BFD_EEXIST;
+ }
+ key->sw_if_index = sw_if->sw_if_index;
+ mhash_set (&bum->bfd_session_idx_by_bfd_key, key, bs->bs_idx, NULL);
+ BFD_DBG ("session created, bs_idx=%u, sw_if_index=%d, local=%U, peer=%U",
+ bs->bs_idx, key->sw_if_index, format_ip46_address, &key->local_addr,
+ IP46_TYPE_ANY, format_ip46_address, &key->peer_addr, IP46_TYPE_ANY);
+ if (BFD_TRANSPORT_UDP4 == t)
+ {
+ bus->adj_index = adj_nbr_add_or_lock (FIB_PROTOCOL_IP4, VNET_LINK_IP4,
+ &key->peer_addr, key->sw_if_index);
+ BFD_DBG ("adj_nbr_add_or_lock(FIB_PROTOCOL_IP4, VNET_LINK_IP4, %U, %d) "
+ "returns %d",
+ format_ip46_address, &key->peer_addr, IP46_TYPE_ANY,
+ key->sw_if_index, bus->adj_index);
+ }
+ else
+ {
+ bus->adj_index = adj_nbr_add_or_lock (FIB_PROTOCOL_IP6, VNET_LINK_IP6,
+ &key->peer_addr, key->sw_if_index);
+ BFD_DBG ("adj_nbr_add_or_lock(FIB_PROTOCOL_IP6, VNET_LINK_IP6, %U, %d) "
+ "returns %d",
+ format_ip46_address, &key->peer_addr, IP46_TYPE_ANY,
+ key->sw_if_index, bus->adj_index);
+ }
+ bs->config_desired_min_tx_us = desired_min_tx_us;
+ bs->required_min_rx_us = required_min_rx_us;
+ bs->local_detect_mult = detect_mult;
+ bfd_session_start (bum->bfd_main, bs);
+ return 0;
+}
+
+static vnet_api_error_t
+bfd_udp_validate_api_input (u32 sw_if_index, const ip46_address_t *local_addr,
+ const ip46_address_t *peer_addr)
+{
+ vnet_sw_interface_t *sw_if =
+ vnet_get_sw_interface (vnet_get_main (), sw_if_index);
+ u8 local_ip_valid = 0;
+ ip_interface_address_t *ia = NULL;
+ if (!sw_if)
+ {
+ BFD_ERR ("got NULL sw_if");
+ return VNET_API_ERROR_INVALID_SW_IF_INDEX;
+ }
+ if (ip46_address_is_ip4 (local_addr))
+ {
+ if (!ip46_address_is_ip4 (peer_addr))
+ {
+ BFD_ERR ("IP family mismatch");
+ return VNET_API_ERROR_INVALID_ARGUMENT;
+ }
+ ip4_main_t *im = &ip4_main;
+
+ /* *INDENT-OFF* */
+ foreach_ip_interface_address (
+ &im->lookup_main, ia, sw_if_index, 0 /* honor unnumbered */, ({
+ ip4_address_t *x =
+ ip_interface_address_get_address (&im->lookup_main, ia);
+ if (x->as_u32 == local_addr->ip4.as_u32)
+ {
+ /* valid address for this interface */
+ local_ip_valid = 1;
+ break;
+ }
+ }));
+ /* *INDENT-ON* */
+ }
+ else
+ {
+ if (ip46_address_is_ip4 (peer_addr))
+ {
+ BFD_ERR ("IP family mismatch");
+ return VNET_API_ERROR_INVALID_ARGUMENT;
+ }
+ ip6_main_t *im = &ip6_main;
+ /* *INDENT-OFF* */
+ foreach_ip_interface_address (
+ &im->lookup_main, ia, sw_if_index, 0 /* honor unnumbered */, ({
+ ip6_address_t *x =
+ ip_interface_address_get_address (&im->lookup_main, ia);
+ if (local_addr->ip6.as_u64[0] == x->as_u64[0] &&
+ local_addr->ip6.as_u64[1] == x->as_u64[1])
+ {
+ /* valid address for this interface */
+ local_ip_valid = 1;
+ break;
+ }
+ }));
+ /* *INDENT-ON* */
+ }
+
+ if (!local_ip_valid)
+ {
+ BFD_ERR ("address not found on interface");
+ return VNET_API_ERROR_ADDRESS_NOT_FOUND_FOR_INTERFACE;
+ }
+
+ return 0;
+}
+
+vnet_api_error_t bfd_udp_add_session (u32 sw_if_index, u32 desired_min_tx_us,
+ u32 required_min_rx_us, u8 detect_mult,
+ const ip46_address_t *local_addr,
+ const ip46_address_t *peer_addr)
+{
+ vnet_api_error_t rv =
+ bfd_udp_validate_api_input (sw_if_index, local_addr, peer_addr);
+ if (rv)
+ {
+ return rv;
+ }
+ if (detect_mult < 1)
+ {
+ BFD_ERR ("detect_mult < 1");
+ return VNET_API_ERROR_INVALID_ARGUMENT;
+ }
+ if (desired_min_tx_us < 1)
+ {
+ BFD_ERR ("desired_min_tx_us < 1");
+ return VNET_API_ERROR_INVALID_ARGUMENT;
+ }
+ return bfd_udp_add_session_internal (&bfd_udp_main, sw_if_index,
+ desired_min_tx_us, required_min_rx_us,
+ detect_mult, local_addr, peer_addr);
+}
+
+vnet_api_error_t bfd_udp_del_session (u32 sw_if_index,
+ const ip46_address_t *local_addr,
+ const ip46_address_t *peer_addr)
+{
+ vnet_api_error_t rv =
+ bfd_udp_validate_api_input (sw_if_index, local_addr, peer_addr);
+ if (rv)
+ {
+ return rv;
+ }
+ bfd_udp_main_t *bum = &bfd_udp_main;
+ vnet_sw_interface_t *sw_if =
+ vnet_get_sw_interface (vnet_get_main (), sw_if_index);
+ bfd_udp_key_t key;
+ memset (&key, 0, sizeof (key));
+ key.sw_if_index = sw_if->sw_if_index;
+ key.local_addr.as_u64[0] = local_addr->as_u64[0];
+ key.local_addr.as_u64[1] = local_addr->as_u64[1];
+ key.peer_addr.as_u64[0] = peer_addr->as_u64[0];
+ key.peer_addr.as_u64[1] = peer_addr->as_u64[1];
+ bfd_session_t *tmp = bfd_lookup_session (bum, &key);
+ if (tmp)
+ {
+ BFD_DBG ("free bfd-udp session, bs_idx=%d", tmp->bs_idx);
+ mhash_unset (&bum->bfd_session_idx_by_bfd_key, &key, NULL);
+ adj_unlock (tmp->udp.adj_index);
+ bfd_put_session (bum->bfd_main, tmp);
+ }
+ else
+ {
+ BFD_ERR ("no such session");
+ return VNET_API_ERROR_BFD_NOENT;
+ }
+ return 0;
+}
+
+typedef enum {
+ BFD_UDP_INPUT_NEXT_NORMAL,
+ BFD_UDP_INPUT_NEXT_REPLY,
+ BFD_UDP_INPUT_N_NEXT,
+} bfd_udp_input_next_t;
+
+/* Packet counters */
+#define foreach_bfd_udp_error(F) \
+ F (NONE, "good bfd packets (processed)") \
+ F (BAD, "invalid bfd packets") \
+ F (DISABLED, "bfd packets received on disabled interfaces")
+
+#define F(sym, string) static char BFD_UDP_ERR_##sym##_STR[] = string;
+foreach_bfd_udp_error (F);
+#undef F
+
+static char *bfd_udp_error_strings[] = {
+#define F(sym, string) BFD_UDP_ERR_##sym##_STR,
+ foreach_bfd_udp_error (F)
+#undef F
+};
+
+typedef enum {
+#define F(sym, str) BFD_UDP_ERROR_##sym,
+ foreach_bfd_udp_error (F)
+#undef F
+ BFD_UDP_N_ERROR,
+} bfd_udp_error_t;
+
+static void bfd_udp4_find_headers (vlib_buffer_t *b, const ip4_header_t **ip4,
+ const udp_header_t **udp)
+{
+ *ip4 = vnet_buffer (b)->ip.header;
+ *udp = (udp_header_t *)((*ip4) + 1);
+}
+
+static bfd_udp_error_t bfd_udp4_verify_transport (const ip4_header_t *ip4,
+ const udp_header_t *udp,
+ const bfd_session_t *bs)
+{
+ const bfd_udp_session_t *bus = &bs->udp;
+ const bfd_udp_key_t *key = &bus->key;
+ if (ip4->src_address.as_u32 != key->peer_addr.ip4.as_u32)
+ {
+ BFD_ERR ("IP src addr mismatch, got %U, expected %U", format_ip4_address,
+ ip4->src_address.as_u32, format_ip4_address,
+ key->peer_addr.ip4.as_u32);
+ return BFD_UDP_ERROR_BAD;
+ }
+ if (ip4->dst_address.as_u32 != key->local_addr.ip4.as_u32)
+ {
+ BFD_ERR ("IP dst addr mismatch, got %U, expected %U", format_ip4_address,
+ ip4->dst_address.as_u32, format_ip4_address,
+ key->local_addr.ip4.as_u32);
+ return BFD_UDP_ERROR_BAD;
+ }
+ const u8 expected_ttl = 255;
+ if (ip4->ttl != expected_ttl)
+ {
+ BFD_ERR ("IP unexpected TTL value %d, expected %d", ip4->ttl,
+ expected_ttl);
+ return BFD_UDP_ERROR_BAD;
+ }
+ if (clib_net_to_host_u16 (udp->src_port) < 49152 ||
+ clib_net_to_host_u16 (udp->src_port) > 65535)
+ {
+ BFD_ERR ("Invalid UDP src port %d, out of range <49152,65535>",
+ udp->src_port);
+ }
+ return BFD_UDP_ERROR_NONE;
+}
+
+typedef struct
+{
+ u32 bs_idx;
+ bfd_pkt_t pkt;
+} bfd_rpc_update_t;
+
+static void bfd_rpc_update_session_cb (const bfd_rpc_update_t *a)
+{
+ bfd_consume_pkt (bfd_udp_main.bfd_main, &a->pkt, a->bs_idx);
+}
+
+static void bfd_rpc_update_session (u32 bs_idx, const bfd_pkt_t *pkt)
+{
+ /* packet length was already verified to be correct by the caller */
+ const u32 data_size = sizeof (bfd_rpc_update_t) -
+ STRUCT_SIZE_OF (bfd_rpc_update_t, pkt) +
+ pkt->head.length;
+ u8 data[data_size];
+ bfd_rpc_update_t *update = (bfd_rpc_update_t *)data;
+ update->bs_idx = bs_idx;
+ clib_memcpy (&update->pkt, pkt, pkt->head.length);
+ vl_api_rpc_call_main_thread (bfd_rpc_update_session_cb, data, data_size);
+}
+
+static bfd_udp_error_t bfd_udp4_scan (vlib_main_t *vm, vlib_node_runtime_t *rt,
+ vlib_buffer_t *b, bfd_session_t **bs_out)
+{
+ const bfd_pkt_t *pkt = vlib_buffer_get_current (b);
+ if (sizeof (*pkt) > b->current_length)
+ {
+ BFD_ERR (
+ "Payload size %d too small to hold bfd packet of minimum size %d",
+ b->current_length, sizeof (*pkt));
+ return BFD_UDP_ERROR_BAD;
+ }
+ const ip4_header_t *ip4;
+ const udp_header_t *udp;
+ bfd_udp4_find_headers (b, &ip4, &udp);
+ if (!ip4 || !udp)
+ {
+ BFD_ERR ("Couldn't find ip4 or udp header");
+ return BFD_UDP_ERROR_BAD;
+ }
+ if (!bfd_verify_pkt_common (pkt))
+ {
+ return BFD_UDP_ERROR_BAD;
+ }
+ bfd_session_t *bs = NULL;
+ if (pkt->your_disc)
+ {
+ BFD_DBG ("Looking up BFD session using discriminator %u",
+ pkt->your_disc);
+ bs = bfd_find_session_by_disc (bfd_udp_main.bfd_main, pkt->your_disc);
+ }
+ else
+ {
+ bfd_udp_key_t key;
+ memset (&key, 0, sizeof (key));
+ key.sw_if_index = vnet_buffer (b)->sw_if_index[VLIB_RX];
+ key.local_addr.ip4.as_u32 = ip4->dst_address.as_u32;
+ key.peer_addr.ip4.as_u32 = ip4->src_address.as_u32;
+ BFD_DBG ("Looking up BFD session using key (sw_if_index=%u, local=%U, "
+ "peer=%U)",
+ key.sw_if_index, format_ip4_address, key.local_addr.ip4.as_u8,
+ format_ip4_address, key.peer_addr.ip4.as_u8);
+ bs = bfd_lookup_session (&bfd_udp_main, &key);
+ }
+ if (!bs)
+ {
+ BFD_ERR ("BFD session lookup failed - no session matches BFD pkt");
+ return BFD_UDP_ERROR_BAD;
+ }
+ BFD_DBG ("BFD session found, bs_idx=%d", bs->bs_idx);
+ if (!bfd_verify_pkt_session (pkt, b->current_length, bs))
+ {
+ return BFD_UDP_ERROR_BAD;
+ }
+ bfd_udp_error_t err;
+ if (BFD_UDP_ERROR_NONE != (err = bfd_udp4_verify_transport (ip4, udp, bs)))
+ {
+ return err;
+ }
+ bfd_rpc_update_session (bs->bs_idx, pkt);
+ *bs_out = bs;
+ return BFD_UDP_ERROR_NONE;
+}
+
+static bfd_udp_error_t bfd_udp6_scan (vlib_main_t *vm, vlib_buffer_t *b)
+{
+ /* TODO */
+ return BFD_UDP_ERROR_BAD;
+}
+
+/*
+ * Process a frame of bfd packets
+ * Expect 1 packet / frame
+ */
+static uword bfd_udp_input (vlib_main_t *vm, vlib_node_runtime_t *rt,
+ vlib_frame_t *f, int is_ipv6)
+{
+ u32 n_left_from, *from;
+ bfd_input_trace_t *t0;
+
+ from = vlib_frame_vector_args (f); /* array of buffer indices */
+ n_left_from = f->n_vectors; /* number of buffer indices */
+
+ while (n_left_from > 0)
+ {
+ u32 bi0;
+ vlib_buffer_t *b0;
+ u32 next0, error0;
+
+ bi0 = from[0];
+ b0 = vlib_get_buffer (vm, bi0);
+
+ bfd_session_t *bs = NULL;
+
+ /* If this pkt is traced, snapshot the data */
+ if (b0->flags & VLIB_BUFFER_IS_TRACED)
+ {
+ int len;
+ t0 = vlib_add_trace (vm, rt, b0, sizeof (*t0));
+ len = (b0->current_length < sizeof (t0->data)) ? b0->current_length
+ : sizeof (t0->data);
+ t0->len = len;
+ clib_memcpy (t0->data, vlib_buffer_get_current (b0), len);
+ }
+
+ /* scan this bfd pkt. error0 is the counter index to bmp */
+ if (is_ipv6)
+ {
+ error0 = bfd_udp6_scan (vm, b0);
+ }
+ else
+ {
+ error0 = bfd_udp4_scan (vm, rt, b0, &bs);
+ }
+ b0->error = rt->errors[error0];
+
+ next0 = BFD_UDP_INPUT_NEXT_NORMAL;
+ if (BFD_UDP_ERROR_NONE == error0)
+ {
+ /* if everything went fine, check for poll bit, if present, re-use
+ the buffer and based on (now update) session parameters, send the
+ final packet back */
+ const bfd_pkt_t *pkt = vlib_buffer_get_current (b0);
+ if (bfd_pkt_get_poll (pkt))
+ {
+ bfd_send_final (vm, b0, bs);
+ next0 = BFD_UDP_INPUT_NEXT_REPLY;
+ }
+ }
+ vlib_set_next_frame_buffer (vm, rt, next0, bi0);
+
+ from += 1;
+ n_left_from -= 1;
+ }
+
+ return f->n_vectors;
+}
+
+static uword bfd_udp4_input (vlib_main_t *vm, vlib_node_runtime_t *rt,
+ vlib_frame_t *f)
+{
+ return bfd_udp_input (vm, rt, f, 0);
+}
+
+/*
+ * bfd input graph node declaration
+ */
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE (bfd_udp4_input_node, static) = {
+ .function = bfd_udp4_input,
+ .name = "bfd-udp4-input",
+ .vector_size = sizeof (u32),
+ .type = VLIB_NODE_TYPE_INTERNAL,
+
+ .n_errors = BFD_UDP_N_ERROR,
+ .error_strings = bfd_udp_error_strings,
+
+ .format_trace = bfd_input_format_trace,
+
+ .n_next_nodes = BFD_UDP_INPUT_N_NEXT,
+ .next_nodes =
+ {
+ [BFD_UDP_INPUT_NEXT_NORMAL] = "error-drop",
+ [BFD_UDP_INPUT_NEXT_REPLY] = "ip4-lookup",
+ },
+};
+/* *INDENT-ON* */
+
+static uword bfd_udp6_input (vlib_main_t *vm, vlib_node_runtime_t *rt,
+ vlib_frame_t *f)
+{
+ return bfd_udp_input (vm, rt, f, 1);
+}
+
+/* *INDENT-OFF* */
+VLIB_REGISTER_NODE (bfd_udp6_input_node, static) = {
+ .function = bfd_udp6_input,
+ .name = "bfd-udp6-input",
+ .vector_size = sizeof (u32),
+ .type = VLIB_NODE_TYPE_INTERNAL,
+
+ .n_errors = BFD_UDP_N_ERROR,
+ .error_strings = bfd_udp_error_strings,
+
+ .format_trace = bfd_input_format_trace,
+
+ .n_next_nodes = BFD_UDP_INPUT_N_NEXT,
+ .next_nodes =
+ {
+ [BFD_UDP_INPUT_NEXT_NORMAL] = "error-drop",
+ [BFD_UDP_INPUT_NEXT_REPLY] = "ip6-lookup",
+ },
+};
+/* *INDENT-ON* */
+
+static clib_error_t *bfd_sw_interface_up_down (vnet_main_t *vnm,
+ u32 sw_if_index, u32 flags)
+{
+ // vnet_hw_interface_t *hi = vnet_get_sup_hw_interface (vnm, sw_if_index);
+ if (!(flags & VNET_SW_INTERFACE_FLAG_ADMIN_UP))
+ {
+ /* TODO */
+ }
+ return 0;
+}
+
+VNET_SW_INTERFACE_ADMIN_UP_DOWN_FUNCTION (bfd_sw_interface_up_down);
+
+static clib_error_t *bfd_hw_interface_up_down (vnet_main_t *vnm,
+ u32 hw_if_index, u32 flags)
+{
+ if (flags & VNET_HW_INTERFACE_FLAG_LINK_UP)
+ {
+ /* TODO */
+ }
+ return 0;
+}
+
+VNET_HW_INTERFACE_LINK_UP_DOWN_FUNCTION (bfd_hw_interface_up_down);
+
+/*
+ * setup function
+ */
+static clib_error_t *bfd_udp_init (vlib_main_t *vm)
+{
+ mhash_init (&bfd_udp_main.bfd_session_idx_by_bfd_key, sizeof (uword),
+ sizeof (bfd_udp_key_t));
+ bfd_udp_main.bfd_main = &bfd_main;
+ udp_register_dst_port (vm, UDP_DST_PORT_bfd4, bfd_udp4_input_node.index, 1);
+ udp_register_dst_port (vm, UDP_DST_PORT_bfd6, bfd_udp6_input_node.index, 0);
+ return 0;
+}
+
+VLIB_INIT_FUNCTION (bfd_udp_init);
diff --git a/vnet/vnet/bfd/bfd_udp.h b/vnet/vnet/bfd/bfd_udp.h
new file mode 100644
index 00000000000..51f5327be01
--- /dev/null
+++ b/vnet/vnet/bfd/bfd_udp.h
@@ -0,0 +1,56 @@
+/* * Copyright (c) 2011-2016 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
+ * @brief BFD global declarations
+ */
+
+#ifndef __included_bfd_udp_h__
+#define __included_bfd_udp_h__
+
+#include <vppinfra/clib.h>
+#include <vnet/adj/adj_types.h>
+#include <vnet/ip/ip6_packet.h>
+
+#define BFD_UDP_KEY_BODY
+
+/* *INDENT-OFF* */
+typedef CLIB_PACKED (struct {
+
+ u32 sw_if_index;
+ ip46_address_t local_addr;
+ ip46_address_t peer_addr;
+
+}) bfd_udp_key_t;
+/* *INDENT-ON* */
+
+typedef struct
+{
+ bfd_udp_key_t key;
+
+ adj_index_t adj_index;
+} bfd_udp_session_t;
+
+void bfd_add_udp_transport (vlib_main_t * vm, vlib_buffer_t * b,
+ bfd_udp_session_t * bs);
+
+#endif /* __included_bfd_udp_h__ */
+
+/*
+ * fd.io coding-style-patch-verification: ON
+ *
+ * Local Variables:
+ * eval: (c-set-style "gnu")
+ * End:
+ */
diff --git a/vnet/vnet/bfd/dir.dox b/vnet/vnet/bfd/dir.dox
new file mode 100644
index 00000000000..ed656b52074
--- /dev/null
+++ b/vnet/vnet/bfd/dir.dox
@@ -0,0 +1,18 @@
+/*
+ * 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.
+ */
+/**
+ @dir vnet/vnet/bfd
+ @brief Bidirectional Forwarding Detection (BFD) implementation
+*/
diff --git a/vnet/vnet/buffer.h b/vnet/vnet/buffer.h
index 898c94ee7b9..82806bba986 100644
--- a/vnet/vnet/buffer.h
+++ b/vnet/vnet/buffer.h
@@ -143,7 +143,11 @@ typedef struct
u8 code;
u32 data;
} icmp;
+
+ /* IP header - saved by ip*_local nodes */
+ void *header;
};
+
} ip;
/*
diff --git a/vnet/vnet/ip/ip4_forward.c b/vnet/vnet/ip/ip4_forward.c
index fc7b34963fe..38729c8ecfb 100644
--- a/vnet/vnet/ip/ip4_forward.c
+++ b/vnet/vnet/ip/ip4_forward.c
@@ -1472,9 +1472,12 @@ ip4_local (vlib_main_t * vm,
ip0 = vlib_buffer_get_current (p0);
ip1 = vlib_buffer_get_current (p1);
- fib_index0 = vec_elt (im->fib_index_by_sw_if_index,
- vnet_buffer(p0)->sw_if_index[VLIB_RX]);
- fib_index1 = vec_elt (im->fib_index_by_sw_if_index,
+ vnet_buffer (p0)->ip.header = ip0;
+ vnet_buffer (p1)->ip.header = ip1;
+
+ fib_index0 = vec_elt (im->fib_index_by_sw_if_index,
+ vnet_buffer (p0)->sw_if_index[VLIB_RX]);
+ fib_index1 = vec_elt (im->fib_index_by_sw_if_index,
vnet_buffer(p1)->sw_if_index[VLIB_RX]);
mtrie0 = &ip4_fib_get (fib_index0)->mtrie;
@@ -1679,6 +1682,8 @@ ip4_local (vlib_main_t * vm,
ip0 = vlib_buffer_get_current (p0);
+ vnet_buffer (p0)->ip.header = ip0;
+
fib_index0 = vec_elt (im->fib_index_by_sw_if_index,
vnet_buffer(p0)->sw_if_index[VLIB_RX]);
@@ -3294,4 +3299,3 @@ VLIB_CLI_COMMAND (set_ip_classify_command, static) = {
.function = set_ip_classify_command_fn,
};
/* *INDENT-ON* */
-
diff --git a/vnet/vnet/ip/udp.h b/vnet/vnet/ip/udp.h
index 4de30f1d16f..f8ff777e275 100644
--- a/vnet/vnet/ip/udp.h
+++ b/vnet/vnet/ip/udp.h
@@ -37,6 +37,7 @@ typedef enum {
_ (67, dhcp_to_server) \
_ (68, dhcp_to_client) \
_ (500, ikev2) \
+_ (3784, bfd4) \
_ (4341, lisp_gpe) \
_ (4342, lisp_cp) \
_ (4739, ipfix) \
@@ -49,6 +50,7 @@ _ (6633, vpath_3)
#define foreach_udp6_dst_port \
_ (547, dhcpv6_to_server) \
_ (546, dhcpv6_to_client) \
+_ (3784, bfd6) \
_ (4341, lisp_gpe6) \
_ (4342, lisp_cp6) \
_ (4790, vxlan6_gpe) \