aboutsummaryrefslogtreecommitdiffstats
path: root/src/vnet/flow/flow.h
blob: 04305edf92ad20726168707a5690656f40da636a (plain) 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299

/*
 * 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.
 */

#ifndef included_vnet_flow_flow_h
#define included_vnet_flow_flow_h

#include <vppinfra/clib.h>
#include <vppinfra/pcap.h>
#include <vnet/l3_types.h>
#include <vnet/ip/ip4_packet.h>
#include <vnet/ip/ip6_packet.h>
#include <vnet/ethernet/packet.h>

#define foreach_flow_type \
  /* l2 flow*/ \
  _(ETHERNET, ethernet, "ethernet") \
  /* l3 IP flow */ \
  _(IP4, ip4, "ipv4") \
  _(IP6, ip6, "ipv6") \
  /* IP tunnel flow */ \
  _(IP4_L2TPV3OIP, ip4_l2tpv3oip, "ipv4-l2tpv3oip") \
  _(IP4_IPSEC_ESP, ip4_ipsec_esp, "ipv4-ipsec-esp") \
  _(IP4_IPSEC_AH, ip4_ipsec_ah, "ipv4-ipsec-ah") \
  /* l4 flow*/ \
  _(IP4_N_TUPLE, ip4_n_tuple, "ipv4-n-tuple") \
  _(IP6_N_TUPLE, ip6_n_tuple, "ipv6-n-tuple") \
  _(IP4_N_TUPLE_TAGGED, ip4_n_tuple_tagged, "ipv4-n-tuple-tagged") \
  _(IP6_N_TUPLE_TAGGED, ip6_n_tuple_tagged, "ipv6-n-tuple-tagged") \
  /* L4 tunnel flow*/ \
  _(IP4_VXLAN, ip4_vxlan, "ipv4-vxlan") \
  _(IP6_VXLAN, ip6_vxlan, "ipv6-vxlan") \
  _(IP4_GTPC, ip4_gtpc, "ipv4-gtpc") \
  _(IP4_GTPU, ip4_gtpu, "ipv4-gtpu")

#define foreach_flow_entry_ethernet \
  _fe(ethernet_header_t, eth_hdr)

#define foreach_flow_entry_ip4 \
  _fe(ip4_address_and_mask_t, src_addr) \
  _fe(ip4_address_and_mask_t, dst_addr) \
  _fe(ip_prot_and_mask_t, protocol)

#define foreach_flow_entry_ip6 \
  _fe(ip6_address_and_mask_t, src_addr) \
  _fe(ip6_address_and_mask_t, dst_addr) \
  _fe(ip_prot_and_mask_t, protocol)

#define foreach_flow_entry_ip4_l2tpv3oip \
  foreach_flow_entry_ip4 \
  _fe(u32, session_id)

#define foreach_flow_entry_ip4_ipsec_esp \
  foreach_flow_entry_ip4 \
  _fe(u32, spi)

#define foreach_flow_entry_ip4_ipsec_ah \
  foreach_flow_entry_ip4 \
  _fe(u32, spi)

#define foreach_flow_entry_ip4_n_tuple \
  foreach_flow_entry_ip4 \
  _fe(ip_port_and_mask_t, src_port) \
  _fe(ip_port_and_mask_t, dst_port)

#define foreach_flow_entry_ip6_n_tuple \
  foreach_flow_entry_ip6 \
  _fe(ip_port_and_mask_t, src_port) \
  _fe(ip_port_and_mask_t, dst_port)

#define foreach_flow_entry_ip4_n_tuple_tagged \
  foreach_flow_entry_ip4 \
  _fe(ip_port_and_mask_t, src_port) \
  _fe(ip_port_and_mask_t, dst_port)

#define foreach_flow_entry_ip6_n_tuple_tagged \
  foreach_flow_entry_ip6 \
  _fe(ip_port_and_mask_t, src_port) \
  _fe(ip_port_and_mask_t, dst_port)

#define foreach_flow_entry_ip4_vxlan \
  foreach_flow_entry_ip4_n_tuple \
  _fe(u16, vni)

#define foreach_flow_entry_ip6_vxlan \
  foreach_flow_entry_ip6_n_tuple \
  _fe(u16, vni)

#define foreach_flow_entry_ip4_gtpc \
  foreach_flow_entry_ip4_n_tuple \
  _fe(u32, teid)

#define foreach_flow_entry_ip4_gtpu \
  foreach_flow_entry_ip4_n_tuple \
  _fe(u32, teid)

#define foreach_flow_action \
  _(0, COUNT, "count") \
  _(1, MARK, "mark") \
  _(2, BUFFER_ADVANCE, "buffer-advance") \
  _(3, REDIRECT_TO_NODE, "redirect-to-node") \
  _(4, REDIRECT_TO_QUEUE, "redirect-to-queue") \
  _(5, RSS, "rss") \
  _(6, DROP, "drop")

typedef enum
{
#define _(v,n,s)  VNET_FLOW_ACTION_##n = (1 << v),
  foreach_flow_action
#undef _
} vnet_flow_action_t;

#define foreach_flow_error \
  _( -1, NOT_SUPPORTED, "not supported")			\
  _( -2, ALREADY_DONE, "already done")				\
  _( -3, ALREADY_EXISTS, "already exists")			\

/*
 * Copyright (c) 2018-2019 Cisco and/or its affiliates.
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this
 * 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 <vcl/vcl_private.h>

static pthread_key_t vcl_worker_stop_key;

vcl_mq_evt_conn_t *
vcl_mq_evt_conn_alloc (vcl_worker_t * wrk)
{
  vcl_mq_evt_conn_t *mqc;
  pool_get (wrk->mq_evt_conns, mqc);
  memset (mqc, 0, sizeof (*mqc));
  return mqc;
}

u32
vcl_mq_evt_conn_index (vcl_worker_t * wrk, vcl_mq_evt_conn_t * mqc)
{
  return (mqc - wrk->mq_evt_conns);
}

vcl_mq_evt_conn_t *
vcl_mq_evt_conn_get (vcl_worker_t * wrk, u32 mq_conn_idx)
{
  return pool_elt_at_index (wrk->mq_evt_conns, mq_conn_idx);
}

int
vcl_mq_epoll_add_evfd (vcl_worker_t * wrk, svm_msg_q_t * mq)
{
  struct epoll_event e = { 0 };
  vcl_mq_evt_conn_t *mqc;
  u32 mqc_index;
  int mq_fd;

  mq_fd = svm_msg_q_get_eventfd (mq);

  if (wrk->mqs_epfd < 0 || mq_fd == -1)
    return -1;

  mqc = vcl_mq_evt_conn_alloc (wrk);
  mqc_index = vcl_mq_evt_conn_index (wrk, mqc);
  mqc->mq_fd = mq_fd;
  mqc->mq = mq;

  e.events = EPOLLIN;
  e.data.u32 = mqc_index;
  if (epoll_ctl (wrk->mqs_epfd, EPOLL_CTL_ADD, mq_fd, &e) < 0)
    {
      VDBG (0, "failed to add mq eventfd to mq epoll fd");
      return -1;
    }

  return mqc_index;
}

int
vcl_mq_epoll_del_evfd (vcl_worker_t * wrk, u32 mqc_index)
{
  vcl_mq_evt_conn_t *mqc;

  if (wrk->mqs_epfd || mqc_index == ~0)
    return -1;

  mqc = vcl_mq_evt_conn_get (wrk, mqc_index);
  if (epoll_ctl (wrk->mqs_epfd, EPOLL_CTL_DEL, mqc->mq_fd, 0) < 0)
    {
      VDBG (0, "failed to del mq eventfd to mq epoll fd");
      return -1;
    }
  return 0;
}

static vcl_worker_t *
vcl_worker_alloc (void)
{
  vcl_worker_t *wrk;
  pool_get (vcm->workers, wrk);
  memset (wrk, 0, sizeof (*wrk));
  wrk->wrk_index = wrk - vcm->workers;
  wrk->forked_child = ~0;
  return wrk;
}

static void
vcl_worker_free (vcl_worker_t * wrk)
{
  pool_put (vcm->workers, wrk);
}

int
vcl_api_app_worker_add (void)
{
  if (vcm->cfg.vpp_app_socket_api)
(wrk); } void vcl_worker_cleanup (vcl_worker_t * wrk, u8 notify_vpp) { clib_spinlock_lock (&vcm->workers_lock); if (notify_vpp) vcl_api_app_worker_del (wrk); if (wrk->mqs_epfd > 0) close (wrk->mqs_epfd); hash_free (wrk->session_index_by_vpp_handles); vec_free (wrk->mq_events); vec_free (wrk->mq_msg_vector); vcl_worker_free (wrk); clib_spinlock_unlock (&vcm->workers_lock); } static void vcl_worker_cleanup_cb (void *arg) { vcl_worker_t *wrk = vcl_worker_get_current (); u32 wrk_index = wrk->wrk_index; vcl_worker_cleanup (wrk, 1 /* notify vpp */ ); vcl_set_worker_index (~0); VDBG (0, "cleaned up worker %u", wrk_index); } vcl_worker_t * vcl_worker_alloc_and_init () { vcl_worker_t *wrk; /* This was initialized already */ if (vcl_get_worker_index () != ~0) return 0; /* Use separate heap map entry for worker */ clib_mem_set_thread_index (); if (pool_elts (vcm->workers) == vcm->cfg.max_workers) { VDBG (0, "max-workers %u limit reached", vcm->cfg.max_workers); return 0; } clib_spinlock_lock (&vcm->workers_lock); wrk = vcl_worker_alloc (); vcl_set_worker_index (wrk->wrk_index); wrk->thread_id = pthread_self (); wrk->current_pid = getpid (); wrk->mqs_epfd = -1; if (vcm->cfg.use_mq_eventfd) { wrk->vcl_needs_real_epoll = 1; wrk->mqs_epfd = epoll_create (1); wrk->vcl_needs_real_epoll = 0; if (wrk->mqs_epfd < 0) { clib_unix_warning ("epoll_create() returned"); goto done; } } wrk->session_index_by_vpp_handles = hash_create (0, sizeof (uword)); clib_time_init (&wrk->clib_time); vec_validate (wrk->mq_events, 64); vec_validate (wrk->mq_msg_vector, 128); vec_reset_length (wrk->mq_msg_vector); vec_validate (wrk->unhandled_evts_vector, 128); vec_reset_length (wrk->unhandled_evts_vector); clib_spinlock_unlock (&vcm->workers_lock); done: return wrk; } int vcl_worker_register_with_vpp (void) { vcl_worker_t *wrk = vcl_worker_get_current (); clib_spinlock_lock (&vcm->workers_lock); if (vcl_api_app_worker_add ()) { VDBG (0, "failed to add worker to vpp"); clib_spinlock_unlock (&vcm->workers_lock); return -1; } if (pthread_key_create (&vcl_worker_stop_key, vcl_worker_cleanup_cb)) VDBG (0, "failed to add pthread cleanup function"); if (pthread_setspecific (vcl_worker_stop_key, &wrk->thread_id)) VDBG (0, "failed to setup key value"); clib_spinlock_unlock (&vcm->workers_lock); VDBG (0, "added worker %u", wrk->wrk_index); return 0; } svm_msg_q_t * vcl_worker_ctrl_mq (vcl_worker_t * wrk) { return wrk->ctrl_mq; } int vcl_session_read_ready (vcl_session_t * s) { if (PREDICT_FALSE (s->flags & VCL_SESSION_F_IS_VEP)) { VDBG (0, "ERROR: session %u: cannot read from an epoll session!", s->session_index); return VPPCOM_EBADFD; } if (vcl_session_is_open (s)) { if (vcl_session_is_ct (s)) return svm_fifo_max_dequeue_cons (s->ct_rx_fifo); if (s->is_dgram) { session_dgram_pre_hdr_t ph; u32 max_deq; max_deq = svm_fifo_max_dequeue_cons (s->rx_fifo); if (max_deq <= SESSION_CONN_HDR_LEN) return 0; if (svm_fifo_peek (s->rx_fifo, 0, sizeof (ph), (u8 *) & ph) < 0) return 0; if (ph.data_length + SESSION_CONN_HDR_LEN > max_deq) return 0; return ph.data_length; } return svm_fifo_max_dequeue_cons (s->rx_fifo); } else if (s->session_state == VCL_STATE_LISTEN) { return clib_fifo_elts (s->accept_evts_fifo); } else { return (s->session_state == VCL_STATE_DISCONNECT) ? VPPCOM_ECONNRESET : VPPCOM_ENOTCONN; } } int vcl_session_write_ready (vcl_session_t * s) { if (PREDICT_FALSE (s->flags & VCL_SESSION_F_IS_VEP)) { VDBG (0, "session %u [0x%llx]: cannot write to an epoll session!", s->session_index, s->vpp_handle); return VPPCOM_EBADFD; } if (vcl_session_is_open (s)) { if (vcl_session_is_ct (s)) return svm_fifo_max_enqueue_prod (s->ct_tx_fifo); if (s->is_dgram) { u32 max_enq = svm_fifo_max_enqueue_prod (s->tx_fifo); if (max_enq <= sizeof (session_dgram_hdr_t)) return 0; return max_enq - sizeof (session_dgram_hdr_t); } return svm_fifo_max_enqueue_prod (s->tx_fifo); } else if (s->session_state == VCL_STATE_LISTEN) { if (s->tx_fifo) return svm_fifo_max_enqueue_prod (s->tx_fifo); else return VPPCOM_EBADFD; } else { return (s->session_state == VCL_STATE_DISCONNECT) ? VPPCOM_ECONNRESET : VPPCOM_ENOTCONN; } } int vcl_segment_attach (u64 segment_handle, char *name, ssvm_segment_type_t type, int fd) { fifo_segment_create_args_t _a, *a = &_a; int rv; memset (a, 0, sizeof (*a)); a->segment_name = name; a->segment_type = type; if (type == SSVM_SEGMENT_MEMFD) a->memfd_fd = fd; clib_rwlock_writer_lock (&vcm->segment_table_lock); if ((rv = fifo_segment_attach (&vcm->segment_main, a))) { clib_warning ("svm_fifo_segment_attach ('%s') failed", name); return rv; } hash_set (vcm->segment_table, segment_handle, a->new_segment_indices[0]); clib_rwlock_writer_unlock (&vcm->segment_table_lock); vec_reset_length (a->new_segment_indices); return 0; } u32 vcl_segment_table_lookup (u64 segment_handle) { uword *seg_indexp; clib_rwlock_reader_lock (&vcm->segment_table_lock); seg_indexp = hash_get (vcm->segment_table, segment_handle); clib_rwlock_reader_unlock (&vcm->segment_table_lock); if (!seg_indexp) return VCL_INVALID_SEGMENT_INDEX; return ((u32) * seg_indexp); } void vcl_segment_detach (u64 segment_handle) { fifo_segment_main_t *sm = &vcm->segment_main; fifo_segment_t *segment; u32 segment_index; segment_index = vcl_segment_table_lookup (segment_handle); if (segment_index == (u32) ~ 0) return; clib_rwlock_writer_lock (&vcm->segment_table_lock); segment = fifo_segment_get_segment (sm, segment_index); fifo_segment_delete (sm, segment); hash_unset (vcm->segment_table, segment_handle); clib_rwlock_writer_unlock (&vcm->segment_table_lock); VDBG (0, "detached segment %u handle %u", segment_index, segment_handle); } int vcl_segment_attach_session (uword segment_handle, uword rxf_offset, uword txf_offset, uword mq_offset, u8 is_ct, vcl_session_t *s) { u32 fs_index, eqs_index; svm_fifo_t *rxf, *txf; fifo_segment_t *fs; u64 eqs_handle; fs_index = vcl_segment_table_lookup (segment_handle); if (fs_index == VCL_INVALID_SEGMENT_INDEX) { VDBG (0, "ERROR: segment for session %u is not mounted!", s->session_index); return -1; } if (!is_ct && mq_offset != (uword) ~0) { eqs_handle = vcl_vpp_worker_segment_handle (0); eqs_index = vcl_segment_table_lookup (eqs_handle); ASSERT (eqs_index != VCL_INVALID_SEGMENT_INDEX); } clib_rwlock_reader_lock (&vcm->segment_table_lock); fs = fifo_segment_get_segment (&vcm->segment_main, fs_index); rxf = fifo_segment_alloc_fifo_w_offset (fs, rxf_offset); txf = fifo_segment_alloc_fifo_w_offset (fs, txf_offset); rxf->segment_index = fs_index; txf->segment_index = fs_index; if (!is_ct && mq_offset != (uword) ~0) { fs = fifo_segment_get_segment (&vcm->segment_main, eqs_index); s->vpp_evt_q = fifo_segment_msg_q_attach (fs, mq_offset, rxf->shr->slice_index); } clib_rwlock_reader_unlock (&vcm->segment_table_lock); if (!is_ct) { rxf->shr->client_session_index = s->session_index; txf->shr->client_session_index = s->session_index; rxf->client_thread_index = vcl_get_worker_index (); txf->client_thread_index = vcl_get_worker_index (); s->rx_fifo = rxf; s->tx_fifo = txf; } else { s->ct_rx_fifo = rxf; s->ct_tx_fifo = txf; } return 0; } void vcl_session_detach_fifos (vcl_session_t *s) { fifo_segment_t *fs; if (!s->rx_fifo) return; clib_rwlock_reader_lock (&vcm->segment_table_lock); fs = fifo_segment_get_segment_if_valid (&vcm->segment_main, s->rx_fifo->segment_index); if (!fs) goto done; fifo_segment_free_client_fifo (fs, s->rx_fifo); fifo_segment_free_client_fifo (fs, s->tx_fifo); if (s->ct_rx_fifo) { fs = fifo_segment_get_segment (&vcm->segment_main, s->ct_rx_fifo->segment_index); fifo_segment_free_client_fifo (fs, s->ct_rx_fifo); fifo_segment_free_client_fifo (fs, s->ct_tx_fifo); } done: clib_rwlock_reader_unlock (&vcm->segment_table_lock); } int vcl_segment_attach_mq (uword segment_handle, uword mq_offset, u32 mq_index, svm_msg_q_t **mq) { fifo_segment_t *fs; u32 fs_index; fs_index = vcl_segment_table_lookup (segment_handle); if (fs_index == VCL_INVALID_SEGMENT_INDEX) { VDBG (0, "ERROR: mq segment %lx for is not attached!", segment_handle); return -1; } clib_rwlock_reader_lock (&vcm->segment_table_lock); fs = fifo_segment_get_segment (&vcm->segment_main, fs_index); *mq = fifo_segment_msg_q_attach (fs, mq_offset, mq_index); clib_rwlock_reader_unlock (&vcm->segment_table_lock); return 0; } int vcl_segment_discover_mqs (uword segment_handle, int *fds, u32 n_fds) { fifo_segment_t *fs; u32 fs_index; fs_index = vcl_segment_table_lookup (segment_handle); if (fs_index == VCL_INVALID_SEGMENT_INDEX) { VDBG (0, "ERROR: mq segment %lx for is not attached!", segment_handle); return -1; } clib_rwlock_reader_lock (&vcm->segment_table_lock); fs = fifo_segment_get_segment (&vcm->segment_main, fs_index); fifo_segment_msg_qs_discover (fs, fds, n_fds); clib_rwlock_reader_unlock (&vcm->segment_table_lock); return 0; } /* * fd.io coding-style-patch-verification: ON * * Local Variables: * eval: (c-set-style "gnu") * End: */