/* * esp_decrypt.c : IPSec ESP decrypt node * * 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. */ #include #include #include #include #include #include #include #include #include #define foreach_esp_decrypt_next \ _(DROP, "error-drop") \ _(IP4_INPUT, "ip4-input-no-checksum") \ _(IP6_INPUT, "ip6-input") \ _(L2_INPUT, "l2-input") \ _(HANDOFF, "handoff") \ _(PENDING, "pending") #define _(v, s) ESP_DECRYPT_NEXT_##v, typedef enum { foreach_esp_decrypt_next #undef _ ESP_DECRYPT_N_NEXT, } esp_decrypt_next_t; #define foreach_esp_decrypt_post_next \ _(DROP, "error-drop") \ _(IP4_INPUT, "ip4-input-no-checksum") \ _(IP6_INPUT, "ip6-input") \ _(L2_INPUT, "l2-input") #define _(v, s) ESP_DECRYPT_POST_NEXT_##v, typedef enum { foreach_esp_decrypt_post_next #undef _ ESP_DECRYPT_POST_N_NEXT, } esp_decrypt_post_next_t; #define foreach_esp_decrypt_error \ _(RX_PKTS, "ESP pkts received") \ _(RX_POST_PKTS, "ESP-POST pkts received") \ _(DECRYPTION_FAILED, "ESP decryption failed") \ _(INTEG_ERROR, "Integrity check failed") \ _(CRYPTO_ENGINE_ERROR, "crypto engine error (packet dropped)") \ _(REPLAY, "SA replayed packet") \ _(RUNT, "undersized packet") \ _(NO_BUFFERS, "no buffers (packet dropped)") \ _(OVERSIZED_HEADER, "buffer with oversized header (dropped)") \ _(NO_TAIL_SPACE, "no enough buffer tail space (dropped)") \ _(TUN_NO_PROTO, "no tunnel protocol") \ _(UNSUP_PAYLOAD, "unsupported payload") \ typedef enum { #define _(sym,str) ESP_DECRYPT_ERROR_##sym, foreach_esp_decrypt_error #undef _ ESP_DECRYPT_N_ERROR, } esp_decrypt_error_t; static char *esp_decrypt_error_strings[] = { #define _(sym,string) string, foreach_esp_decrypt_error #undef _ }; typedef struct { u32 seq; u32 sa_seq; u32 sa_seq_hi; ipsec_crypto_alg_t crypto_alg; ipsec_integ_alg_t integ_alg; } esp_decrypt_trace_t; /* packet trace format function */ static u8 * format_esp_decrypt_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 *); esp_decrypt_trace_t *t = va_arg (*args, esp_decrypt_trace_t *); s = format (s, "esp: crypto %U integrity %U pkt-seq %d sa-seq %u sa-seq-hi %u", format_ipsec_crypto_alg, t->crypto_alg, format_ipsec_integ_alg, t->integ_alg, t->seq, t->sa_seq, t->sa_seq_hi); return s; } #define ESP_ENCRYPT_PD_F_FD_TRANSPORT (1 << 2) static_always_inline void esp_process_ops (vlib_main_t * vm, vlib_node_runtime_t * node, vnet_crypto_op_t * ops, vlib_buffer_t * b[], u16 * nexts, int e) { vnet_crypto_op_t *op = ops; u32 n_fail, n_ops = vec_len (ops); if (n_ops == 0) return; n_fail = n_ops - vnet_crypto_process_ops (vm, op, n_ops); while (n_fail) { ASSERT (op - ops < n_ops); if (op->status != VNET_CRYPTO_OP_STATUS_COMPLETED) { u32 err, bi = op->user_data; if (op->status == VNET_CRYPTO_OP_STATUS_FAIL_BAD_HMAC) err = e; else err = ESP_DECRYPT_ERROR_CRYPTO_ENGINE_ERROR; b[bi]->error = node->errors[err]; nexts[bi] = ESP_DECRYPT_NEXT_DROP; n_fail--; } op++; } } static_always_inline void esp_process_chained_ops (vlib_main_t * vm, vlib_node_runtime_t * node, vnet_crypto_op_t * ops, vlib_buffer_t * b[], u16 * nexts, vnet_crypto_op_chunk_t * chunks, int e) { vnet_crypto_op_t *op = ops; u32 n_fail, n_ops = vec_len (ops); if (n_ops == 0) return; n_fail = n_ops - vnet_crypto_process_chained_ops (vm, op, chunks, n_ops); while (n_fail) { ASSERT (op - ops < n_ops); if (op->status != VNET_CRYPTO_OP_STATUS_COMPLETED) { u32 err, bi = op->user_data; if (op->status == VNET_CRYPTO_OP_STATUS_FAIL_BAD_HMAC) err = e; else err = ESP_DECRYPT_ERROR_CRYPTO_ENGINE_ERROR; b[bi]->error = node->errors[err]; nexts[bi] = ESP_DECRYPT_NEXT_DROP; n_fail--; } op++; } } always_inline void esp_remove_tail (vlib_main_t * vm, vlib_buffer_t * b, vlib_buffer_t * last, u16 tail) { vlib_buffer_t *before_last = b; if (last->current_length > tail) { last->current_length -= tail; return; } ASSERT (b->flags & VLIB_BUFFER_NEXT_PRESENT); while (b->flags & VLIB_BUFFER_NEXT_PRESENT) { before_last = b; b = vlib_get_buffer (vm, b->next_buffer); } before_last->current_length -= tail - last->current_length; vlib_buffer_free_one (vm, before_last->next_buffer); before_last->flags &= ~VLIB_BUFFER_NEXT_PRESENT; } /* ICV is splitted in last two buffers so move it to the last buffer and return pointer to it */ static_always_inline u8 * esp_move_icv (vlib_main_t * vm, vlib_buffer_t * first, esp_decrypt_packet_data2_t * pd2, u16 icv_sz, u16 * dif) { vlib_buffer_t *before_last, *bp; u16 last_sz = pd2->lb->current_length; u16 first_sz = icv_sz - last_sz; bp = before_last = first; while (bp->flags & VLIB_BUFFER_NEXT_PRESENT) { before_last = bp; bp = vlib_get_buffer (vm, bp->next_buffer); } u8 *lb_curr = vlib_buffer_get_current (pd2->lb); memmove (lb_curr + first_sz, lb_curr, last_sz); clib_memcpy_fast (lb_curr, vlib_buffer_get_tail (before_last) - first_sz, first_sz); before_last->current_length -= first_sz; clib_memset (vlib_buffer_get_tail (before_last), 0, first_sz); if (dif) dif[0] = first_sz; pd2->lb = before_last; pd2->icv_removed = 1; pd2->free_buffer_index = before_last->next_buffer; before_last->flags &= ~VLIB_BUFFER_NEXT_PRESENT; return lb_curr; } static_always_inline i16 esp_insert_esn (vlib_main_t * vm, ipsec_sa_t * sa, esp_decrypt_packet_data2_t * pd2, u32 * data_len, u8 ** digest, u16 * len, vlib_buffer_t * b, u8 * payload) { if (!ipsec_sa_is_set_USE_ESN (sa)) return 0; /* shift ICV by 4 bytes to insert ESN */ u32 seq_hi = clib_host_to_net_u32 (sa->seq_hi); u8 tmp[ESP_MAX_ICV_SIZE], sz = sizeof (sa->seq_hi); if (pd2->icv_removed) { u16 space_left = vlib_buffer_space_left_at_end (vm, pd2->lb); if (space_left >= sz) { clib_memcpy_fast (vlib_buffer_get_tail (pd2->lb), &seq_hi, sz); *data_len += sz; } else return sz; len[0] = b->current_length; } else { clib_memcpy_fast (tmp, payload + len[0], ESP_MAX_ICV_SIZE); clib_memcpy_fast (payload + len[0], &seq_hi, sz); clib_memcpy_fast (payload + len[0] + sz, tmp, ESP_MAX_ICV_SIZE); *data_len += sz; *digest += sz; } return sz; } static_always_inline u8 * esp_move_icv_esn (vlib_main_t * vm, vlib_buffer_t * first, esp_decrypt_packet_data2_t * pd2, u16 icv_sz, ipsec_sa_t * sa, u8 * extra_esn, u32 * len) { u16 dif = 0; u8 *digest = esp_move_icv (vm, first, pd2, icv_sz, &dif); if (dif) *len -= dif; if (ipsec_sa_is_set_USE_ESN (sa)) { u8 sz = sizeof (sa->seq_hi); u32 seq_hi = clib_host_to_net_u32 (sa->seq_hi); u16 space_left = vlib_buffer_space_left_at_end (vm, pd2->lb); if (space_left >= sz) { clib_memcpy_fast (vlib_buffer_get_tail (pd2->lb), &seq_hi, sz); *len += sz; } else { /* no space for ESN at the tail, use the next buffer * (with ICV data) */ ASSERT (pd2->icv_removed); vlib_buffer_t *tmp = vlib_get_buffer (vm, pd2->free_buffer_index); clib_memcpy_fast (vlib_buffer_get_current (tmp) - sz, &seq_hi, sz); extra_esn[0] = 1; } } return digest; } static_always_inline int esp_decrypt_chain_integ (vlib_main_t * vm, ipsec_per_thread_data_t * ptd, esp_decrypt_packet_data2_t * pd2, ipsec_sa_t * sa0, vlib_buffer_t * b, u8 icv_sz, u8 * start_src, u32 start_len, u8 ** digest, u16 * n_ch, u32 * integ_total_len) { vnet_crypto_op_chunk_t *ch; vlib_buffer_t *cb = vlib_get_buffer (vm, b->next_buffer); u16 n_chunks = 1; u32 total_len; vec_add2 (ptd->chunks, ch, 1); total_len = ch->len = start_len; ch->src = start_src; while (1) { vec_add2 (ptd->chunks, ch, 1); n_chunks += 1; ch->src = vlib_buffer_get_current (cb); if (
unix {{
{unix}
  log /var/log/vpp/vpp.log
  full-coredump
  cli-listen /run/vpp/cli.sock
}}

api-trace {{
  on
}}

cpu {{
{cpu}
  # scheduler-policy fifo
  # scheduler-priority 50

        ## In the VPP there is one main thread and optionally the user can create worker(s)
	## The main thread and worker thread(s) can be pinned to CPU core(s) manually or automatically

	## Manual pinning of thread(s) to CPU core(s)

	## Set logical CPU core where main thread runs
	# main-core 1

	## Set logical CPU core(s) where worker threads are running
	# corelist-workers 2-3,18-19

	## Automatic pinning of thread(s) to CPU core(s)

	## Sets number of CPU core(s) to be skipped (1 ... N-1)
	## Skipped CPU core(s) are not used for pinning main thread and working thread(s).
	## The main thread is automatically pinned to the first available CPU core and worker(s)
	## are pinned to next free CPU core(s) after core assigned to main thread
	# skip-cores 4

	## Specify a number of workers to be created
	## Workers are pinned to N consecutive CPU cores while skipping "skip-cores" CPU core(s)
	## and main thread's CPU core
	# workers 2

	## Set scheduling policy and priority of main and worker threads

	## Scheduling policy options are: other (SCHED_OTHER), batch (SCHED_BATCH)
	## idle (SCHED_IDLE), fifo (SCHED_FIFO), rr (SCHED_RR)
	# scheduler-policy fifo

	## Scheduling priority is used only for "real-time policies (fifo and rr),
	## and has to be in the range of priorities supported for a particular policy
	# scheduler-priority 50
}}

dpdk {{
{devices}

        ## Change default settings for all interfaces
	# dev default {{
		## Number of receive queues, enables RSS
		## Default is 1
		# num-rx-queues 3

		## Number of transmit queues, Default is equal
		## to number of worker threads or 1 if no workers treads
		# num-tx-queues 3

		## Number of descriptors in transmit and receive rings
		## increasing or reducing number can impact performance
		## Default is 1024 for both rx and tx
		# num-rx-desc 512
		# num-tx-desc 512

		## VLAN strip offload mode for interface
		## Default is off
		# vlan-strip-offload on
	# }}

	## Whitelist specific interface by specifying PCI address
	# dev 0000:02:00.0

	## Whitelist specific interface by specifying PCI address and in
	## addition specify custom parameters for this interface
	# dev 0000:02:00.1 {{
	#	num-rx-queues 2
	# }}

	## Specify bonded interface and its slaves via PCI addresses
	## 
        ## Bonded interface in XOR load balance mode (mode 2) with L3 and L4 headers 
	# vdev eth_bond0,mode=2,slave=0000:02:00.0,slave=0000:03:00.0,xmit_policy=l34
	# vdev eth_bond1,mode=2,slave=0000:02:00.1,slave=0000:03:00.1,xmit_policy=l34
	##
	## Bonded interface in Active-Back up mode (mode 1)
	# vdev eth_bond0,mode=1,slave=0000:02:00.0,slave=0000:03:00.0
	# vdev eth_bond1,mode=1,slave=0000:02:00.1,slave=0000:03:00.1

	## Change UIO driver used by VPP, Options are: igb_uio, vfio-pci
	## and uio_pci_generic (default)
	# uio-driver vfio-pci

	## Disable multi-segment buffers, improves performance but
	## disables Jumbo MTU support
	# no-multi-seg

	## Increase number of buffers allocated, needed only in scenarios with
	## large number of interfaces and worker threads. Value is per CPU socket.
	## Default is 16384
	# num-mbufs 128000

	## Change hugepages allocation per-socket, needed only if there is need for
	## larger number of mbufs. Default is 256M on each detected CPU socket
	# socket-mem 2048,2048
}}

# Adjusting the plugin path depending on where the VPP plugins are:
#plugins
#{{
#	path /home/bms/vpp/build-root/install-vpp-native/vpp/lib/vpp_plugins
#}}

# Alternate syntax to choose plugin path
#plugin_path /home/bms/vpp/build-root/install-vpp-native/vpp/lib/vpp_plugins

#node-variants {{
#	defaults {{
#		avx512 100
#	}}
#	ip4-inacl {{
#	avx2 100
#		avx512 50
#	}}
#}}

{tcp}
of (esp_footer_t) + icv_sz) { /* esp footer is either splitted in two buffers or in the before * last buffer */ vlib_buffer_t *before_last = b, *bp = b; while (bp->flags & VLIB_BUFFER_NEXT_PRESENT) { before_last = bp; bp = vlib_get_buffer (vm, bp->next_buffer); } u8 *bt = vlib_buffer_get_tail (before_last); if (lb->current_length == icv_sz) { esp_footer_t *f = (esp_footer_t *) (bt - sizeof (*f)); pad_length = f->pad_length; next_header = f->next_header; } else { pad_length = (bt - 1)[0]; next_header = ((u8 *) vlib_buffer_get_current (lb))[0]; } } else { esp_footer_t *f = (esp_footer_t *) (lb->data + lb->current_data + lb->current_length - sizeof (esp_footer_t) - icv_sz); pad_length = f->pad_length; next_header = f->next_header; } } else { icv_sz = pd->icv_sz; esp_footer_t *f = (esp_footer_t *) (lb->data + lb->current_data + lb->current_length - sizeof (esp_footer_t) - icv_sz); pad_length = f->pad_length; next_header = f->next_header; } u16 adv = pd->iv_sz + esp_sz; u16 tail = sizeof (esp_footer_t) + pad_length + icv_sz; u16 tail_orig = sizeof (esp_footer_t) + pad_length + pd->icv_sz; b->flags &= ~VLIB_BUFFER_TOTAL_LENGTH_VALID; if ((pd->flags & tun_flags) == 0 && !is_tun) /* transport mode */ { u8 udp_sz = (is_ip6 == 0 && pd->flags & IPSEC_SA_FLAG_UDP_ENCAP) ? sizeof (udp_header_t) : 0; u16 ip_hdr_sz = pd->hdr_sz - udp_sz; u8 *old_ip = b->data + pd->current_data - ip_hdr_sz - udp_sz; u8 *ip = old_ip + adv + udp_sz; if (is_ip6 && ip_hdr_sz > 64) memmove (ip, old_ip, ip_hdr_sz); else clib_memcpy_le64 (ip, old_ip, ip_hdr_sz); b->current_data = pd->current_data + adv - ip_hdr_sz; b->current_length += ip_hdr_sz - adv; esp_remove_tail (vm, b, lb, tail); if (is_ip6) { ip6_header_t *ip6 = (ip6_header_t *) ip; u16 len = clib_net_to_host_u16 (ip6->payload_length); len -= adv + tail_orig; ip6->payload_length = clib_host_to_net_u16 (len); ip6->protocol = next_header; next[0] = ESP_DECRYPT_NEXT_IP6_INPUT; } else { ip4_header_t *ip4 = (ip4_header_t *) ip; ip_csum_t sum = ip4->checksum; u16 len = clib_net_to_host_u16 (ip4->length); len = clib_host_to_net_u16 (len - adv - tail_orig - udp_sz); sum = ip_csum_update (sum, ip4->protocol, next_header, ip4_header_t, protocol); sum = ip_csum_update (sum, ip4->length, len, ip4_header_t, length); ip4->checksum = ip_csum_fold (sum); ip4->protocol = next_header; ip4->length = len; next[0] = ESP_DECRYPT_NEXT_IP4_INPUT; } } else { if (PREDICT_TRUE (next_header == IP_PROTOCOL_IP_IN_IP)) { next[0] = ESP_DECRYPT_NEXT_IP4_INPUT; b->current_data = pd->current_data + adv; b->current_length = pd->current_length - adv; esp_remove_tail (vm, b, lb, tail); } else if (next_header == IP_PROTOCOL_IPV6) { next[0] = ESP_DECRYPT_NEXT_IP6_INPUT; b->current_data = pd->current_data + adv; b->current_length = pd->current_length - adv; esp_remove_tail (vm, b, lb, tail); } else { if (is_tun && next_header == IP_PROTOCOL_GRE) { gre_header_t *gre; b->current_data = pd->current_data + adv; b->current_length = pd->current_length - adv - tail; gre = vlib_buffer_get_current (b); vlib_buffer_advance (b, sizeof (*gre)); switch (clib_net_to_host_u16 (gre->protocol)) { case GRE_PROTOCOL_teb: vnet_update_l2_len (b); next[0] = ESP_DECRYPT_NEXT_L2_INPUT; break; case GRE_PROTOCOL_ip4: next[0] = ESP_DECRYPT_NEXT_IP4_INPUT; break; case GRE_PROTOCOL_ip6: next[0] = ESP_DECRYPT_NEXT_IP6_INPUT; break; default: b->error = node->errors[ESP_DECRYPT_ERROR_UNSUP_PAYLOAD]; next[0] = ESP_DECRYPT_NEXT_DROP; break; } } else { next[0] = ESP_DECRYPT_NEXT_DROP; b->error = node->errors[ESP_DECRYPT_ERROR_UNSUP_PAYLOAD]; return; } } if (is_tun) { if (ipsec_sa_is_set_IS_PROTECT (sa0)) { /* * There are two encap possibilities * 1) the tunnel and ths SA are prodiving encap, i.e. it's * MAC | SA-IP | TUN-IP | ESP | PAYLOAD * implying the SA is in tunnel mode (on a tunnel interface) * 2) only the tunnel provides encap * MAC | TUN-IP | ESP | PAYLOAD * implying the SA is in transport mode. * * For 2) we need only strip the tunnel encap and we're good. * since the tunnel and crypto ecnap (int the tun=protect * object) are the same and we verified above that these match * for 1) we need to strip the SA-IP outer headers, to * reveal the tunnel IP and then check that this matches * the configured tunnel. */ const ipsec_tun_protect_t *itp; if (is_async) itp = ipsec_tun_protect_get (pd->protect_index); else itp = ipsec_tun_protect_get (vnet_buffer (b)-> ipsec.protect_index); if (PREDICT_TRUE (next_header == IP_PROTOCOL_IP_IN_IP)) { const ip4_header_t *ip4; ip4 = vlib_buffer_get_current (b); if (!ip46_address_is_equal_v4 (&itp->itp_tun.src, &ip4->dst_address) || !ip46_address_is_equal_v4 (&itp->itp_tun.dst, &ip4->src_address)) { next[0] = ESP_DECRYPT_NEXT_DROP; b->error = node->errors[ESP_DECRYPT_ERROR_TUN_NO_PROTO]; } } else if (next_header == IP_PROTOCOL_IPV6) { const ip6_header_t *ip6; ip6 = vlib_buffer_get_current (b); if (!ip46_address_is_equal_v6 (&itp->itp_tun.src, &ip6->dst_address) || !ip46_address_is_equal_v6 (&itp->itp_tun.dst, &ip6->src_address)) { next[0] = ESP_DECRYPT_NEXT_DROP; b->error = node->errors[ESP_DECRYPT_ERROR_TUN_NO_PROTO]; } } } } } } /* when submitting a frame is failed, drop all buffers in the frame */ static_always_inline void esp_async_recycle_failed_submit (vnet_crypto_async_frame_t * f, vlib_buffer_t ** b, u16 * next) { u32 n_drop = f->n_elts; while (--n_drop) { (b - n_drop)[0]->error = ESP_DECRYPT_ERROR_CRYPTO_ENGINE_ERROR; (next - n_drop)[0] = ESP_DECRYPT_NEXT_DROP; } vnet_crypto_async_reset_frame (f); } always_inline uword esp_decrypt_inline (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * from_frame, int is_ip6, int is_tun, u16 async_next) { ipsec_main_t *im = &ipsec_main; u32 thread_index = vm->thread_index; u16 len; ipsec_per_thread_data_t *ptd = vec_elt_at_index (im->ptd, thread_index); u32 *from = vlib_frame_vector_args (from_frame); u32 n_left = from_frame->n_vectors; vlib_buffer_t *bufs[VLIB_FRAME_SIZE], **b = bufs; u16 nexts[VLIB_FRAME_SIZE], *next = nexts; esp_decrypt_packet_data_t pkt_data[VLIB_FRAME_SIZE], *pd = pkt_data; esp_decrypt_packet_data2_t pkt_data2[VLIB_FRAME_SIZE], *pd2 = pkt_data2; esp_decrypt_packet_data_t cpd = { }; u32 current_sa_index = ~0, current_sa_bytes = 0, current_sa_pkts = 0; const u8 esp_sz = sizeof (esp_header_t); ipsec_sa_t *sa0 = 0; vnet_crypto_op_t _op, *op = &_op; vnet_crypto_op_t **crypto_ops = &ptd->crypto_ops; vnet_crypto_op_t **integ_ops = &ptd->integ_ops; vnet_crypto_async_frame_t *async_frame = 0; int is_async = im->async_mode; vnet_crypto_async_op_id_t last_async_op = ~0; vlib_get_buffers (vm, from, b, n_left); if (!is_async) { vec_reset_length (ptd->crypto_ops); vec_reset_length (ptd->integ_ops); vec_reset_length (ptd->chained_crypto_ops); vec_reset_length (ptd->chained_integ_ops); } vec_reset_length (ptd->chunks); clib_memset_u16 (nexts, -1, n_left); while (n_left > 0) { u8 *payload; if (n_left > 2) { u8 *p; vlib_prefetch_buffer_header (b[2], LOAD); p = vlib_buffer_get_current (b[1]); CLIB_PREFETCH (p, CLIB_CACHE_LINE_BYTES, LOAD); p -= CLIB_CACHE_LINE_BYTES; CLIB_PREFETCH (p, CLIB_CACHE_LINE_BYTES, LOAD); } u32 n_bufs = vlib_buffer_chain_linearize (vm, b[0]); if (n_bufs == 0) { b[0]->error = node->errors[ESP_DECRYPT_ERROR_NO_BUFFERS]; next[0] = ESP_DECRYPT_NEXT_DROP; goto next; } if (vnet_buffer (b[0])->ipsec.sad_index != current_sa_index) { if (current_sa_pkts) vlib_increment_combined_counter (&ipsec_sa_counters, thread_index, current_sa_index, current_sa_pkts, current_sa_bytes); current_sa_bytes = current_sa_pkts = 0; current_sa_index = vnet_buffer (b[0])->ipsec.sad_index; sa0 = pool_elt_at_index (im->sad, current_sa_index); cpd.icv_sz = sa0->integ_icv_size; cpd.iv_sz = sa0->crypto_iv_size; cpd.flags = sa0->flags; cpd.sa_index = current_sa_index; /* submit frame when op_id is different then the old one */ if (is_async && last_async_op != sa0->crypto_async_dec_op_id) { if (async_frame && async_frame->n_elts) { if (vnet_crypto_async_submit_open_frame (vm, async_frame)) esp_async_recycle_failed_submit (async_frame, b, next); } async_frame = vnet_crypto_async_get_frame (vm, sa0->crypto_async_dec_op_id); last_async_op = sa0->crypto_async_dec_op_id; } } if (PREDICT_FALSE (~0 == sa0->decrypt_thread_index)) { /* this is the first packet to use this SA, claim the SA * for this thread. this could happen simultaneously on * another thread */ clib_atomic_cmp_and_swap (&sa0->decrypt_thread_index, ~0, ipsec_sa_assign_thread (thread_index)); } if (PREDICT_TRUE (thread_index != sa0->decrypt_thread_index)) { next[0] = ESP_DECRYPT_NEXT_HANDOFF; goto next; } /* store packet data for next round for easier prefetch */ pd->sa_data = cpd.sa_data; pd->current_data = b[0]->current_data; pd->hdr_sz = pd->current_data - vnet_buffer (b[0])->l3_hdr_offset; payload = b[0]->data + pd->current_data; pd->seq = clib_host_to_net_u32 (((esp_header_t *) payload)->seq); pd->is_chain = 0; pd2->lb = b[0]; pd2->free_buffer_index = 0; pd2->icv_removed = 0; if (n_bufs > 1) { pd->is_chain = 1; /* find last buffer in the chain */ while (pd2->lb->flags & VLIB_BUFFER_NEXT_PRESENT) pd2->lb = vlib_get_buffer (vm, pd2->lb->next_buffer); crypto_ops = &ptd->chained_crypto_ops; integ_ops = &ptd->chained_integ_ops; } pd->current_length = b[0]->current_length; /* anti-reply check */ if (ipsec_sa_anti_replay_check (sa0, pd->seq)) { b[0]->error = node->errors[ESP_DECRYPT_ERROR_REPLAY]; next[0] = ESP_DECRYPT_NEXT_DROP; goto next; } if (pd->current_length < cpd.icv_sz + esp_sz + cpd.iv_sz) { b[0]->error = node->errors[ESP_DECRYPT_ERROR_RUNT]; next[0] = ESP_DECRYPT_NEXT_DROP; goto next; } len = pd->current_length - cpd.icv_sz; current_sa_pkts += 1; current_sa_bytes += vlib_buffer_length_in_chain (vm, b[0]); if (is_async) { int ret = esp_decrypt_prepare_async_frame (vm, node, ptd, &async_frame, sa0, payload, len, cpd.icv_sz, cpd.iv_sz, pd, pd2, from[b - bufs], b[0], next, async_next); if (PREDICT_FALSE (ret < 0)) { esp_async_recycle_failed_submit (async_frame, b, next); goto next; } } else esp_decrypt_prepare_sync_op (vm, node, ptd, &crypto_ops, &integ_ops, op, sa0, payload, len, cpd.icv_sz, cpd.iv_sz, pd, pd2, b[0], next, b - bufs); /* next */ next: n_left -= 1; next += 1; pd += 1; pd2 += 1; b += 1; } if (PREDICT_TRUE (~0 != current_sa_index)) vlib_increment_combined_counter (&ipsec_sa_counters, thread_index, current_sa_index, current_sa_pkts, current_sa_bytes); if (is_async) { if (async_frame && async_frame->n_elts) { if (vnet_crypto_async_submit_open_frame (vm, async_frame) < 0) esp_async_recycle_failed_submit (async_frame, b, next); } /* no post process in async */ n_left = from_frame->n_vectors; vlib_node_increment_counter (vm, node->node_index, ESP_DECRYPT_ERROR_RX_PKTS, n_left); vlib_buffer_enqueue_to_next (vm, node, from, nexts, n_left); return n_left; } else { esp_process_ops (vm, node, ptd->integ_ops, bufs, nexts, ESP_DECRYPT_ERROR_INTEG_ERROR); esp_process_chained_ops (vm, node, ptd->chained_integ_ops, bufs, nexts, ptd->chunks, ESP_DECRYPT_ERROR_INTEG_ERROR); esp_process_ops (vm, node, ptd->crypto_ops, bufs, nexts, ESP_DECRYPT_ERROR_DECRYPTION_FAILED); esp_process_chained_ops (vm, node, ptd->chained_crypto_ops, bufs, nexts, ptd->chunks, ESP_DECRYPT_ERROR_DECRYPTION_FAILED); } /* Post decryption ronud - adjust packet data start and length and next node */ n_left = from_frame->n_vectors; next = nexts; pd = pkt_data; pd2 = pkt_data2; b = bufs; while (n_left) { if (n_left >= 2) { void *data = b[1]->data + pd[1].current_data; /* buffer metadata */ vlib_prefetch_buffer_header (b[1], LOAD); /* esp_footer_t */ CLIB_PREFETCH (data + pd[1].current_length - pd[1].icv_sz - 2, CLIB_CACHE_LINE_BYTES, LOAD); /* packet headers */ CLIB_PREFETCH (data - CLIB_CACHE_LINE_BYTES, CLIB_CACHE_LINE_BYTES * 2, LOAD); } /* save the sa_index as GRE_teb post_crypto changes L2 opaque */ if (PREDICT_FALSE (b[0]->flags & VLIB_BUFFER_IS_TRACED)) current_sa_index = vnet_buffer (b[0])->ipsec.sad_index; if (next[0] >= ESP_DECRYPT_N_NEXT) esp_decrypt_post_crypto (vm, node, pd, pd2, b[0], next, is_ip6, is_tun, 0); /* trace: */ if (PREDICT_FALSE (b[0]->flags & VLIB_BUFFER_IS_TRACED)) { esp_decrypt_trace_t *tr; tr = vlib_add_trace (vm, node, b[0], sizeof (*tr)); sa0 = pool_elt_at_index (im->sad, current_sa_index); tr->crypto_alg = sa0->crypto_alg; tr->integ_alg = sa0->integ_alg; tr->seq = pd->seq; tr->sa_seq = sa0->last_seq; tr->sa_seq_hi = sa0->seq_hi; } /* next */ n_left -= 1; next += 1; pd += 1; pd2 += 1; b += 1; } n_left = from_frame->n_vectors; vlib_node_increment_counter (vm, node->node_index, ESP_DECRYPT_ERROR_RX_PKTS, n_left); vlib_buffer_enqueue_to_next (vm, node, from, nexts, n_left); return n_left; } always_inline uword esp_decrypt_post_inline (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * from_frame, int is_ip6, int is_tun) { ipsec_main_t *im = &ipsec_main; u32 *from = vlib_frame_vector_args (from_frame); u32 n_left = from_frame->n_vectors; vlib_buffer_t *bufs[VLIB_FRAME_SIZE], **b = bufs; u16 nexts[VLIB_FRAME_SIZE], *next = nexts; vlib_get_buffers (vm, from, b, n_left); while (n_left > 0) { esp_decrypt_packet_data_t *pd = &(esp_post_data (b[0]))->decrypt_data; if (n_left > 2) { vlib_prefetch_buffer_header (b[2], LOAD); vlib_prefetch_buffer_header (b[1], LOAD); } if (!pd->is_chain) esp_decrypt_post_crypto (vm, node, pd, 0, b[0], next, is_ip6, is_tun, 1); else { esp_decrypt_packet_data2_t *pd2 = esp_post_data2 (b[0]); esp_decrypt_post_crypto (vm, node, pd, pd2, b[0], next, is_ip6, is_tun, 1); } /*trace: */ if (PREDICT_FALSE (b[0]->flags & VLIB_BUFFER_IS_TRACED)) { ipsec_sa_t *sa0 = pool_elt_at_index (im->sad, pd->sa_index); esp_decrypt_trace_t *tr; esp_decrypt_packet_data_t *async_pd = &(esp_post_data (b[0]))->decrypt_data; tr = vlib_add_trace (vm, node, b[0], sizeof (*tr)); sa0 = pool_elt_at_index (im->sad, async_pd->sa_index); tr->crypto_alg = sa0->crypto_alg; tr->integ_alg = sa0->integ_alg; tr->seq = pd->seq; tr->sa_seq = sa0->last_seq; tr->sa_seq_hi = sa0->seq_hi; } n_left--; next++; b++; } n_left = from_frame->n_vectors; vlib_node_increment_counter (vm, node->node_index, ESP_DECRYPT_ERROR_RX_POST_PKTS, n_left); vlib_buffer_enqueue_to_next (vm, node, from, nexts, n_left); return n_left; } VLIB_NODE_FN (esp4_decrypt_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * from_frame) { return esp_decrypt_inline (vm, node, from_frame, 0, 0, esp_decrypt_async_next.esp4_post_next); } VLIB_NODE_FN (esp4_decrypt_post_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * from_frame) { return esp_decrypt_post_inline (vm, node, from_frame, 0, 0); } VLIB_NODE_FN (esp4_decrypt_tun_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * from_frame) { return esp_decrypt_inline (vm, node, from_frame, 0, 1, esp_decrypt_async_next.esp4_tun_post_next); } VLIB_NODE_FN (esp4_decrypt_tun_post_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * from_frame) { return esp_decrypt_post_inline (vm, node, from_frame, 0, 1); } VLIB_NODE_FN (esp6_decrypt_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * from_frame) { return esp_decrypt_inline (vm, node, from_frame, 1, 0, esp_decrypt_async_next.esp6_post_next); } VLIB_NODE_FN (esp6_decrypt_post_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * from_frame) { return esp_decrypt_post_inline (vm, node, from_frame, 1, 0); } VLIB_NODE_FN (esp6_decrypt_tun_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * from_frame) { return esp_decrypt_inline (vm, node, from_frame, 1, 1, esp_decrypt_async_next.esp6_tun_post_next); } VLIB_NODE_FN (esp6_decrypt_tun_post_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * from_frame) { return esp_decrypt_post_inline (vm, node, from_frame, 1, 1); } VLIB_NODE_FN (esp_decrypt_pending_node) (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_frame_t * from_frame) { return from_frame->n_vectors; } /* *INDENT-OFF* */ VLIB_REGISTER_NODE (esp_decrypt_pending_node) = { .name = "esp-decrypt-pending", .vector_size = sizeof (u32), .type = VLIB_NODE_TYPE_INTERNAL, .n_next_nodes = 0 }; /* *INDENT-ON* */ /* *INDENT-OFF* */ VLIB_REGISTER_NODE (esp4_decrypt_node) = { .name = "esp4-decrypt", .vector_size = sizeof (u32), .format_trace = format_esp_decrypt_trace, .type = VLIB_NODE_TYPE_INTERNAL, .n_errors = ARRAY_LEN(esp_decrypt_error_strings), .error_strings = esp_decrypt_error_strings, .n_next_nodes = ESP_DECRYPT_N_NEXT, .next_nodes = { [ESP_DECRYPT_NEXT_DROP] = "ip4-drop", [ESP_DECRYPT_NEXT_IP4_INPUT] = "ip4-input-no-checksum", [ESP_DECRYPT_NEXT_IP6_INPUT] = "ip6-input", [ESP_DECRYPT_NEXT_L2_INPUT] = "l2-input", [ESP_DECRYPT_NEXT_HANDOFF] = "esp4-decrypt-handoff", [ESP_DECRYPT_NEXT_PENDING] = "esp-decrypt-pending" }, }; VLIB_REGISTER_NODE (esp4_decrypt_post_node) = { .name = "esp4-decrypt-post", .vector_size = sizeof (u32), .format_trace = format_esp_decrypt_trace, .type = VLIB_NODE_TYPE_INTERNAL, .n_errors = ARRAY_LEN(esp_decrypt_error_strings), .error_strings = esp_decrypt_error_strings, .sibling_of = "esp4-decrypt", }; VLIB_REGISTER_NODE (esp6_decrypt_node) = { .name = "esp6-decrypt", .vector_size = sizeof (u32), .format_trace = format_esp_decrypt_trace, .type = VLIB_NODE_TYPE_INTERNAL, .n_errors = ARRAY_LEN(esp_decrypt_error_strings), .error_strings = esp_decrypt_error_strings, .n_next_nodes = ESP_DECRYPT_N_NEXT, .next_nodes = { [ESP_DECRYPT_NEXT_DROP] = "ip6-drop", [ESP_DECRYPT_NEXT_IP4_INPUT] = "ip4-input-no-checksum", [ESP_DECRYPT_NEXT_IP6_INPUT] = "ip6-input", [ESP_DECRYPT_NEXT_L2_INPUT] = "l2-input", [ESP_DECRYPT_NEXT_HANDOFF]= "esp6-decrypt-handoff", [ESP_DECRYPT_NEXT_PENDING] = "esp-decrypt-pending" }, }; VLIB_REGISTER_NODE (esp6_decrypt_post_node) = { .name = "esp6-decrypt-post", .vector_size = sizeof (u32), .format_trace = format_esp_decrypt_trace, .type = VLIB_NODE_TYPE_INTERNAL, .n_errors = ARRAY_LEN(esp_decrypt_error_strings), .error_strings = esp_decrypt_error_strings, .sibling_of = "esp6-decrypt", }; VLIB_REGISTER_NODE (esp4_decrypt_tun_node) = { .name = "esp4-decrypt-tun", .vector_size = sizeof (u32), .format_trace = format_esp_decrypt_trace, .type = VLIB_NODE_TYPE_INTERNAL, .n_errors = ARRAY_LEN(esp_decrypt_error_strings), .error_strings = esp_decrypt_error_strings, .n_next_nodes = ESP_DECRYPT_N_NEXT, .next_nodes = { [ESP_DECRYPT_NEXT_DROP] = "ip4-drop", [ESP_DECRYPT_NEXT_IP4_INPUT] = "ip4-input-no-checksum", [ESP_DECRYPT_NEXT_IP6_INPUT] = "ip6-input", [ESP_DECRYPT_NEXT_L2_INPUT] = "l2-input", [ESP_DECRYPT_NEXT_HANDOFF] = "esp4-decrypt-tun-handoff", [ESP_DECRYPT_NEXT_PENDING] = "esp-decrypt-pending" }, }; VLIB_REGISTER_NODE (esp4_decrypt_tun_post_node) = { .name = "esp4-decrypt-tun-post", .vector_size = sizeof (u32), .format_trace = format_esp_decrypt_trace, .type = VLIB_NODE_TYPE_INTERNAL, .n_errors = ARRAY_LEN(esp_decrypt_error_strings), .error_strings = esp_decrypt_error_strings, .sibling_of = "esp4-decrypt-tun", }; VLIB_REGISTER_NODE (esp6_decrypt_tun_node) = { .name = "esp6-decrypt-tun", .vector_size = sizeof (u32), .format_trace = format_esp_decrypt_trace, .type = VLIB_NODE_TYPE_INTERNAL, .n_errors = ARRAY_LEN(esp_decrypt_error_strings), .error_strings = esp_decrypt_error_strings, .n_next_nodes = ESP_DECRYPT_N_NEXT, .next_nodes = { [ESP_DECRYPT_NEXT_DROP] = "ip6-drop", [ESP_DECRYPT_NEXT_IP4_INPUT] = "ip4-input-no-checksum", [ESP_DECRYPT_NEXT_IP6_INPUT] = "ip6-input", [ESP_DECRYPT_NEXT_L2_INPUT] = "l2-input", [ESP_DECRYPT_NEXT_HANDOFF]= "esp6-decrypt-tun-handoff", [ESP_DECRYPT_NEXT_PENDING] = "esp-decrypt-pending" }, }; VLIB_REGISTER_NODE (esp6_decrypt_tun_post_node) = { .name = "esp6-decrypt-tun-post", .vector_size = sizeof (u32), .format_trace = format_esp_decrypt_trace, .type = VLIB_NODE_TYPE_INTERNAL, .n_errors = ARRAY_LEN(esp_decrypt_error_strings), .error_strings = esp_decrypt_error_strings, .sibling_of = "esp6-decrypt-tun", }; /* *INDENT-ON* */ /* * fd.io coding-style-patch-verification: ON * * Local Variables: * eval: (c-set-style "gnu") * End: */