From 492d7790ff26c569bee81617c662363652891140 Mon Sep 17 00:00:00 2001 From: Gabriel Oginski Date: Wed, 10 Nov 2021 07:59:56 +0000 Subject: wireguard: add async mode for encryption packets Originally wireguard doesn't support async mode for encryption packets. This patch add async mode for encryption in wireguard and also adds support chacha20-poly1305 algorithm in cryptodev for async handler. In addition it contains new command line to activate async mode for wireguard: set wireguard async mode on|off and also add new command to check active mode for wireguard: show wireguard mode Type: improvement Signed-off-by: Gabriel Oginski Change-Id: I141d48b42ee8dbff0112b8542ab5205268089da6 --- src/plugins/wireguard/wireguard_output_tun.c | 452 +++++++++++++++++++++++++-- 1 file changed, 427 insertions(+), 25 deletions(-) (limited to 'src/plugins/wireguard/wireguard_output_tun.c') diff --git a/src/plugins/wireguard/wireguard_output_tun.c b/src/plugins/wireguard/wireguard_output_tun.c index 2feb0570a31..f7ae4d174c0 100644 --- a/src/plugins/wireguard/wireguard_output_tun.c +++ b/src/plugins/wireguard/wireguard_output_tun.c @@ -21,11 +21,12 @@ #include #include -#define foreach_wg_output_error \ - _(NONE, "No error") \ - _(PEER, "Peer error") \ - _(KEYPAIR, "Keypair error") \ - _(TOO_BIG, "packet too big") \ +#define foreach_wg_output_error \ + _ (NONE, "No error") \ + _ (PEER, "Peer error") \ + _ (KEYPAIR, "Keypair error") \ + _ (TOO_BIG, "packet too big") \ + _ (CRYPTO_ENGINE_ERROR, "crypto engine error (packet dropped)") typedef enum { @@ -56,6 +57,12 @@ typedef struct u8 is_ip4; } wg_output_tun_trace_t; +typedef struct +{ + index_t peer; + u32 next_index; +} wg_output_tun_post_trace_t; + u8 * format_ip4_udp_header (u8 * s, va_list * args) { @@ -93,6 +100,47 @@ format_wg_output_tun_trace (u8 * s, va_list * args) return s; } +/* post node - packet trace format function */ +static u8 * +format_wg_output_tun_post_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 *); + + wg_output_tun_post_trace_t *t = va_arg (*args, wg_output_tun_post_trace_t *); + + s = format (s, "peer: %d\n", t->peer); + s = format (s, " wg-post: next node index %u", t->next_index); + return s; +} + +static_always_inline void +wg_prepare_sync_enc_op (vlib_main_t *vm, vnet_crypto_op_t **crypto_ops, + u8 *src, u32 src_len, u8 *dst, u8 *aad, u32 aad_len, + u64 nonce, vnet_crypto_key_index_t key_index, u32 bi, + u8 *iv) +{ + vnet_crypto_op_t _op, *op = &_op; + u8 src_[] = {}; + + clib_memset (iv, 0, 4); + clib_memcpy (iv + 4, &nonce, sizeof (nonce)); + + vec_add2_aligned (crypto_ops[0], op, 1, CLIB_CACHE_LINE_BYTES); + vnet_crypto_op_init (op, VNET_CRYPTO_OP_CHACHA20_POLY1305_ENC); + + op->tag_len = NOISE_AUTHTAG_LEN; + op->tag = dst + src_len; + op->src = !src ? src_ : src; + op->len = src_len; + op->dst = dst; + op->key_index = key_index; + op->aad = aad; + op->aad_len = aad_len; + op->iv = iv; + op->user_data = bi; +} + static_always_inline void wg_output_process_ops (vlib_main_t *vm, vlib_node_runtime_t *node, vnet_crypto_op_t *ops, vlib_buffer_t *b[], u16 *nexts, @@ -121,10 +169,148 @@ wg_output_process_ops (vlib_main_t *vm, vlib_node_runtime_t *node, } } +static_always_inline void +wg_output_tun_add_to_frame (vlib_main_t *vm, vnet_crypto_async_frame_t *f, + u32 key_index, u32 crypto_len, + i16 crypto_start_offset, u32 buffer_index, + u16 next_node, u8 *iv, u8 *tag, u8 flags) +{ + vnet_crypto_async_frame_elt_t *fe; + u16 index; + + ASSERT (f->n_elts < VNET_CRYPTO_FRAME_SIZE); + + index = f->n_elts; + fe = &f->elts[index]; + f->n_elts++; + fe->key_index = key_index; + fe->crypto_total_length = crypto_len; + fe->crypto_start_offset = crypto_start_offset; + fe->iv = iv; + fe->tag = tag; + fe->flags = flags; + f->buffer_indices[index] = buffer_index; + f->next_node_index[index] = next_node; +} + +static_always_inline enum noise_state_crypt +wq_output_tun_process (vlib_main_t *vm, vnet_crypto_op_t **crypto_ops, + noise_remote_t *r, uint32_t *r_idx, uint64_t *nonce, + uint8_t *src, size_t srclen, uint8_t *dst, u32 bi, + u8 *iv, f64 time) +{ + noise_keypair_t *kp; + enum noise_state_crypt ret = SC_FAILED; + + if ((kp = r->r_current) == NULL) + goto error; + + /* We confirm that our values are within our tolerances. We want: + * - a valid keypair + * - our keypair to be less than REJECT_AFTER_TIME seconds old + * - our receive counter to be less than REJECT_AFTER_MESSAGES + * - our send counter to be less than REJECT_AFTER_MESSAGES + */ + if (!kp->kp_valid || + wg_birthdate_has_expired_opt (kp->kp_birthdate, REJECT_AFTER_TIME, + time) || + kp->kp_ctr.c_recv >= REJECT_AFTER_MESSAGES || + ((*nonce = noise_counter_send (&kp->kp_ctr)) > REJECT_AFTER_MESSAGES)) + goto error; + + /* We encrypt into the same buffer, so the caller must ensure that buf + * has NOISE_AUTHTAG_LEN bytes to store the MAC. The nonce and index + * are passed back out to the caller through the provided data pointer. */ + *r_idx = kp->kp_remote_index; + + wg_prepare_sync_enc_op (vm, crypto_ops, src, srclen, dst, NULL, 0, *nonce, + kp->kp_send_index, bi, iv); + + /* If our values are still within tolerances, but we are approaching + * the tolerances, we notify the caller with ESTALE that they should + * establish a new keypair. The current keypair can continue to be used + * until the tolerances are hit. We notify if: + * - our send counter is valid and not less than REKEY_AFTER_MESSAGES + * - we're the initiator and our keypair is older than + * REKEY_AFTER_TIME seconds */ + ret = SC_KEEP_KEY_FRESH; + if ((kp->kp_valid && *nonce >= REKEY_AFTER_MESSAGES) || + (kp->kp_is_initiator && wg_birthdate_has_expired_opt ( + kp->kp_birthdate, REKEY_AFTER_TIME, time))) + goto error; + + ret = SC_OK; +error: + return ret; +} + +static_always_inline enum noise_state_crypt +wg_add_to_async_frame (vlib_main_t *vm, wg_per_thread_data_t *ptd, + vnet_crypto_async_frame_t *async_frame, + vlib_buffer_t *b, u8 *payload, u32 payload_len, u32 bi, + u16 next, u16 async_next, noise_remote_t *r, + uint32_t *r_idx, uint64_t *nonce, u8 *iv, f64 time) +{ + wg_post_data_t *post = wg_post_data (b); + u8 flag = 0; + noise_keypair_t *kp; + + post->next_index = next; + + /* crypto */ + enum noise_state_crypt ret = SC_FAILED; + + if ((kp = r->r_current) == NULL) + goto error; + + /* We confirm that our values are within our tolerances. We want: + * - a valid keypair + * - our keypair to be less than REJECT_AFTER_TIME seconds old + * - our receive counter to be less than REJECT_AFTER_MESSAGES + * - our send counter to be less than REJECT_AFTER_MESSAGES + */ + if (!kp->kp_valid || + wg_birthdate_has_expired_opt (kp->kp_birthdate, REJECT_AFTER_TIME, + time) || + kp->kp_ctr.c_recv >= REJECT_AFTER_MESSAGES || + ((*nonce = noise_counter_send (&kp->kp_ctr)) > REJECT_AFTER_MESSAGES)) + goto error; + + /* We encrypt into the same buffer, so the caller must ensure that buf + * has NOISE_AUTHTAG_LEN bytes to store the MAC. The nonce and index + * are passed back out to the caller through the provided data pointer. */ + *r_idx = kp->kp_remote_index; + + clib_memset (iv, 0, 4); + clib_memcpy (iv + 4, nonce, sizeof (nonce)); + + /* this always succeeds because we know the frame is not full */ + wg_output_tun_add_to_frame (vm, async_frame, kp->kp_send_index, payload_len, + payload - b->data, bi, async_next, iv, + payload + payload_len, flag); + + /* If our values are still within tolerances, but we are approaching + * the tolerances, we notify the caller with ESTALE that they should + * establish a new keypair. The current keypair can continue to be used + * until the tolerances are hit. We notify if: + * - our send counter is valid and not less than REKEY_AFTER_MESSAGES + * - we're the initiator and our keypair is older than + * REKEY_AFTER_TIME seconds */ + ret = SC_KEEP_KEY_FRESH; + if ((kp->kp_valid && *nonce >= REKEY_AFTER_MESSAGES) || + (kp->kp_is_initiator && wg_birthdate_has_expired_opt ( + kp->kp_birthdate, REKEY_AFTER_TIME, time))) + goto error; + + ret = SC_OK; +error: + return ret; +} + /* is_ip4 - inner header flag */ always_inline uword wg_output_tun_inline (vlib_main_t *vm, vlib_node_runtime_t *node, - vlib_frame_t *frame, u8 is_ip4) + vlib_frame_t *frame, u8 is_ip4, u16 async_next_node) { wg_main_t *wmp = &wg_main; wg_per_thread_data_t *ptd = @@ -140,10 +326,18 @@ wg_output_tun_inline (vlib_main_t *vm, vlib_node_runtime_t *node, vlib_buffer_t *sync_bufs[VLIB_FRAME_SIZE]; u32 thread_index = vm->thread_index; u16 n_sync = 0; - u16 drop_next = WG_OUTPUT_NEXT_ERROR; + const u16 drop_next = WG_OUTPUT_NEXT_ERROR; + const u8 is_async = wg_op_mode_is_set_ASYNC (); + vnet_crypto_async_frame_t *async_frame = NULL; + u16 n_async = 0; + u16 noop_nexts[VLIB_FRAME_SIZE], *noop_next = noop_nexts, n_noop = 0; + u16 err = !0; + u32 sync_bi[VLIB_FRAME_SIZE]; + u32 noop_bi[VLIB_FRAME_SIZE]; vlib_get_buffers (vm, from, bufs, n_left_from); vec_reset_length (ptd->crypto_ops); + vec_reset_length (ptd->async_frames); wg_peer_t *peer = NULL; u32 adj_index = 0; @@ -169,7 +363,8 @@ wg_output_tun_inline (vlib_main_t *vm, vlib_node_runtime_t *node, LOAD); } - next[0] = WG_OUTPUT_NEXT_ERROR; + noop_next[0] = WG_OUTPUT_NEXT_ERROR; + err = WG_OUTPUT_NEXT_ERROR; adj_index = vnet_buffer (b[0])->ip.adj_index[VLIB_TX]; @@ -193,9 +388,10 @@ wg_output_tun_inline (vlib_main_t *vm, vlib_node_runtime_t *node, wg_peer_assign_thread (thread_index)); } - if (PREDICT_TRUE (thread_index != peer->output_thread_index)) + if (PREDICT_FALSE (thread_index != peer->output_thread_index)) { - next[0] = WG_OUTPUT_NEXT_HANDOFF; + noop_next[0] = WG_OUTPUT_NEXT_HANDOFF; + err = WG_OUTPUT_NEXT_HANDOFF; goto next; } @@ -245,12 +441,35 @@ wg_output_tun_inline (vlib_main_t *vm, vlib_node_runtime_t *node, last_adj_index = adj_index; } + /* Here we are sure that can send packet to next node */ + next[0] = WG_OUTPUT_NEXT_INTERFACE_OUTPUT; + enum noise_state_crypt state; - state = noise_sync_remote_encrypt ( - vm, crypto_ops, &peer->remote, &message_data_wg->receiver_index, - &message_data_wg->counter, plain_data, plain_data_len, plain_data, - n_sync, iv_data, time); + if (is_async) + { + /* get a frame for this op if we don't yet have one or it's full */ + if (NULL == async_frame || + vnet_crypto_async_frame_is_full (async_frame)) + { + async_frame = vnet_crypto_async_get_frame ( + vm, VNET_CRYPTO_OP_CHACHA20_POLY1305_TAG16_AAD0_ENC); + /* Save the frame to the list we'll submit at the end */ + vec_add1 (ptd->async_frames, async_frame); + } + state = wg_add_to_async_frame ( + vm, ptd, async_frame, b[0], plain_data, plain_data_len, + from[b - bufs], next[0], async_next_node, &peer->remote, + &message_data_wg->receiver_index, &message_data_wg->counter, + iv_data, time); + } + else + { + state = wq_output_tun_process ( + vm, crypto_ops, &peer->remote, &message_data_wg->receiver_index, + &message_data_wg->counter, plain_data, plain_data_len, plain_data, + n_sync, iv_data, time); + } if (PREDICT_FALSE (state == SC_KEEP_KEY_FRESH)) { @@ -261,11 +480,11 @@ wg_output_tun_inline (vlib_main_t *vm, vlib_node_runtime_t *node, // TODO: Maybe wrong wg_send_handshake_from_mt (peeri, false); wg_peer_update_flags (peeri, WG_PEER_ESTABLISHED, false); + noop_next[0] = WG_OUTPUT_NEXT_ERROR; goto out; } - /* Here we are sure that can send packet to next node */ - next[0] = WG_OUTPUT_NEXT_INTERFACE_OUTPUT; + err = WG_OUTPUT_NEXT_INTERFACE_OUTPUT; if (is_ip4_out) { @@ -304,31 +523,214 @@ wg_output_tun_inline (vlib_main_t *vm, vlib_node_runtime_t *node, } next: - sync_bufs[n_sync] = b[0]; - n_sync += 1; + if (PREDICT_FALSE (err != WG_OUTPUT_NEXT_INTERFACE_OUTPUT)) + { + noop_bi[n_noop] = from[b - bufs]; + n_noop++; + noop_next++; + goto next_left; + } + if (!is_async) + { + sync_bi[n_sync] = from[b - bufs]; + sync_bufs[n_sync] = b[0]; + n_sync += 1; + next += 1; + } + else + { + n_async++; + } + next_left: n_left_from -= 1; - next += 1; b += 1; } - /* wg-output-process-ops */ - wg_output_process_ops (vm, node, ptd->crypto_ops, sync_bufs, nexts, - drop_next); + if (n_sync) + { + /* wg-output-process-ops */ + wg_output_process_ops (vm, node, ptd->crypto_ops, sync_bufs, nexts, + drop_next); + vlib_buffer_enqueue_to_next (vm, node, sync_bi, nexts, n_sync); + } + if (n_async) + { + /* submit all of the open frames */ + vnet_crypto_async_frame_t **async_frame; + + vec_foreach (async_frame, ptd->async_frames) + { + if (PREDICT_FALSE ( + vnet_crypto_async_submit_open_frame (vm, *async_frame) < 0)) + { + u32 n_drop = (*async_frame)->n_elts; + u32 *bi = (*async_frame)->buffer_indices; + u16 index = n_noop; + while (n_drop--) + { + noop_bi[index] = bi[0]; + vlib_buffer_t *b = vlib_get_buffer (vm, bi[0]); + noop_nexts[index] = drop_next; + b->error = node->errors[WG_OUTPUT_ERROR_CRYPTO_ENGINE_ERROR]; + bi++; + index++; + } + n_noop += (*async_frame)->n_elts; + + vnet_crypto_async_reset_frame (*async_frame); + vnet_crypto_async_free_frame (vm, *async_frame); + } + } + } + if (n_noop) + { + vlib_buffer_enqueue_to_next (vm, node, noop_bi, noop_nexts, n_noop); + } - vlib_buffer_enqueue_to_next (vm, node, from, nexts, n_sync); return frame->n_vectors; } +always_inline uword +wg_output_tun_post (vlib_main_t *vm, vlib_node_runtime_t *node, + vlib_frame_t *frame) +{ + vlib_buffer_t *bufs[VLIB_FRAME_SIZE], **b = bufs; + u16 nexts[VLIB_FRAME_SIZE], *next = nexts; + u32 *from = vlib_frame_vector_args (frame); + u32 n_left = frame->n_vectors; + + index_t peeri = ~0; + + vlib_get_buffers (vm, from, b, n_left); + + if (n_left >= 4) + { + vlib_prefetch_buffer_header (b[0], LOAD); + vlib_prefetch_buffer_header (b[1], LOAD); + vlib_prefetch_buffer_header (b[2], LOAD); + vlib_prefetch_buffer_header (b[3], LOAD); + } + + while (n_left > 8) + { + vlib_prefetch_buffer_header (b[4], LOAD); + vlib_prefetch_buffer_header (b[5], LOAD); + vlib_prefetch_buffer_header (b[6], LOAD); + vlib_prefetch_buffer_header (b[7], LOAD); + + next[0] = (wg_post_data (b[0]))->next_index; + next[1] = (wg_post_data (b[1]))->next_index; + next[2] = (wg_post_data (b[2]))->next_index; + next[3] = (wg_post_data (b[3]))->next_index; + + if (PREDICT_FALSE (node->flags & VLIB_NODE_FLAG_TRACE)) + { + if (b[0]->flags & VLIB_BUFFER_IS_TRACED) + { + wg_output_tun_post_trace_t *tr = + vlib_add_trace (vm, node, b[0], sizeof (*tr)); + peeri = wg_peer_get_by_adj_index ( + vnet_buffer (b[0])->ip.adj_index[VLIB_TX]); + tr->peer = peeri; + tr->next_index = next[0]; + } + if (b[1]->flags & VLIB_BUFFER_IS_TRACED) + { + wg_output_tun_post_trace_t *tr = + vlib_add_trace (vm, node, b[1], sizeof (*tr)); + peeri = wg_peer_get_by_adj_index ( + vnet_buffer (b[1])->ip.adj_index[VLIB_TX]); + tr->next_index = next[1]; + } + if (b[2]->flags & VLIB_BUFFER_IS_TRACED) + { + wg_output_tun_post_trace_t *tr = + vlib_add_trace (vm, node, b[2], sizeof (*tr)); + peeri = wg_peer_get_by_adj_index ( + vnet_buffer (b[2])->ip.adj_index[VLIB_TX]); + tr->next_index = next[2]; + } + if (b[3]->flags & VLIB_BUFFER_IS_TRACED) + { + wg_output_tun_post_trace_t *tr = + vlib_add_trace (vm, node, b[3], sizeof (*tr)); + peeri = wg_peer_get_by_adj_index ( + vnet_buffer (b[3])->ip.adj_index[VLIB_TX]); + tr->next_index = next[3]; + } + } + + b += 4; + next += 4; + n_left -= 4; + } + + while (n_left > 0) + { + next[0] = (wg_post_data (b[0]))->next_index; + if (PREDICT_FALSE ((node->flags & VLIB_NODE_FLAG_TRACE) && + (b[0]->flags & VLIB_BUFFER_IS_TRACED))) + { + wg_output_tun_post_trace_t *tr = + vlib_add_trace (vm, node, b[0], sizeof (*tr)); + peeri = wg_peer_get_by_adj_index ( + vnet_buffer (b[0])->ip.adj_index[VLIB_TX]); + tr->next_index = next[0]; + } + + b += 1; + next += 1; + n_left -= 1; + } + + vlib_buffer_enqueue_to_next (vm, node, from, nexts, frame->n_vectors); + return frame->n_vectors; +} + +VLIB_REGISTER_NODE (wg4_output_tun_post_node) = { + .name = "wg4-output-tun-post-node", + .vector_size = sizeof (u32), + .format_trace = format_wg_output_tun_post_trace, + .type = VLIB_NODE_TYPE_INTERNAL, + .sibling_of = "wg4-output-tun", + .n_errors = ARRAY_LEN (wg_output_error_strings), + .error_strings = wg_output_error_strings, +}; + +VLIB_REGISTER_NODE (wg6_output_tun_post_node) = { + .name = "wg6-output-tun-post-node", + .vector_size = sizeof (u32), + .format_trace = format_wg_output_tun_post_trace, + .type = VLIB_NODE_TYPE_INTERNAL, + .sibling_of = "wg6-output-tun", + .n_errors = ARRAY_LEN (wg_output_error_strings), + .error_strings = wg_output_error_strings, +}; + +VLIB_NODE_FN (wg4_output_tun_post_node) +(vlib_main_t *vm, vlib_node_runtime_t *node, vlib_frame_t *from_frame) +{ + return wg_output_tun_post (vm, node, from_frame); +} + +VLIB_NODE_FN (wg6_output_tun_post_node) +(vlib_main_t *vm, vlib_node_runtime_t *node, vlib_frame_t *from_frame) +{ + return wg_output_tun_post (vm, node, from_frame); +} + VLIB_NODE_FN (wg4_output_tun_node) (vlib_main_t *vm, vlib_node_runtime_t *node, vlib_frame_t *frame) { - return wg_output_tun_inline (vm, node, frame, /* is_ip4 */ 1); + return wg_output_tun_inline (vm, node, frame, /* is_ip4 */ 1, + wg_encrypt_async_next.wg4_post_next); } VLIB_NODE_FN (wg6_output_tun_node) (vlib_main_t *vm, vlib_node_runtime_t *node, vlib_frame_t *frame) { - return wg_output_tun_inline (vm, node, frame, /* is_ip4 */ 0); + return wg_output_tun_inline (vm, node, frame, /* is_ip4 */ 0, + wg_encrypt_async_next.wg6_post_next); } /* *INDENT-OFF* */ -- cgit 1.2.3-korg