1/*
2 * Copyright (c) 2015 Cisco and/or its affiliates.
3 * Licensed under the Apache License, Version 2.0 (the "License");
4 * you may not use this file except in compliance with the License.
5 * You may obtain a copy of the License at:
6 *
7 *     http://www.apache.org/licenses/LICENSE-2.0
8 *
9 * Unless required by applicable law or agreed to in writing, software
10 * distributed under the License is distributed on an "AS IS" BASIS,
11 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12 * See the License for the specific language governing permissions and
13 * limitations under the License.
14 */
15/*
16 * vnet/buffer.h: vnet buffer flags
17 *
18 * Copyright (c) 2008 Eliot Dresselhaus
19 *
20 * Permission is hereby granted, free of charge, to any person obtaining
21 * a copy of this software and associated documentation files (the
22 * "Software"), to deal in the Software without restriction, including
23 * without limitation the rights to use, copy, modify, merge, publish,
24 * distribute, sublicense, and/or sell copies of the Software, and to
25 * permit persons to whom the Software is furnished to do so, subject to
26 * the following conditions:
27 *
28 * The above copyright notice and this permission notice shall be
29 * included in all copies or substantial portions of the Software.
30 *
31 *  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
32 *  EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
33 *  MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
34 *  NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
35 *  LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
36 *  OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
37 *  WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
38 */
39
40#ifndef included_vnet_buffer_h
41#define included_vnet_buffer_h
42
43#include <vlib/vlib.h>
44
45/**
46 * Flags that are set in the high order bits of ((vlib_buffer*)b)->flags
47 *
48 */
49#define foreach_vnet_buffer_flag                        \
50  _( 1, L4_CHECKSUM_COMPUTED, "l4-cksum-computed", 1)	\
51  _( 2, L4_CHECKSUM_CORRECT, "l4-cksum-correct", 1)	\
52  _( 3, VLAN_2_DEEP, "vlan-2-deep", 1)			\
53  _( 4, VLAN_1_DEEP, "vlan-1-deep", 1)			\
54  _( 5, SPAN_CLONE, "span-clone", 1)                    \
55  _( 6, LOOP_COUNTER_VALID, "loop-counter-valid", 0)    \
56  _( 7, LOCALLY_ORIGINATED, "local", 1)                 \
57  _( 8, IS_IP4, "ip4", 1)                               \
58  _( 9, IS_IP6, "ip6", 1)                               \
59  _(10, OFFLOAD_IP_CKSUM, "offload-ip-cksum", 1)        \
60  _(11, OFFLOAD_TCP_CKSUM, "offload-tcp-cksum", 1)      \
61  _(12, OFFLOAD_UDP_CKSUM, "offload-udp-cksum", 1)      \
62  _(13, IS_NATED, "natted", 1)                          \
63  _(14, L2_HDR_OFFSET_VALID, "l2_hdr_offset_valid", 0)  \
64  _(15, L3_HDR_OFFSET_VALID, "l3_hdr_offset_valid", 0)  \
65  _(16, L4_HDR_OFFSET_VALID, "l4_hdr_offset_valid", 0)  \
66  _(17, FLOW_REPORT, "flow-report", 1)                  \
67  _(18, IS_DVR, "dvr", 1)                               \
68  _(19, QOS_DATA_VALID, "qos-data-valid", 0)            \
69  _(20, GSO, "gso", 0)                                  \
70  _(21, AVAIL1, "avail1", 1)                            \
71  _(22, AVAIL2, "avail2", 1)                            \
72  _(23, AVAIL3, "avail3", 1)                            \
73  _(24, AVAIL4, "avail4", 1)                            \
74  _(25, AVAIL5, "avail5", 1)                            \
75  _(26, AVAIL6, "avail6", 1)                            \
76  _(27, AVAIL7, "avail7", 1)
77
78/*
79 * Please allocate the FIRST available bit, redefine
80 * AVAIL 1 ... AVAILn-1, and remove AVAILn. Please maintain the
81 * VNET_BUFFER_FLAGS_ALL_AVAIL definition.
82 */
83
84#define VNET_BUFFER_FLAGS_ALL_AVAIL                                     \
85  (VNET_BUFFER_F_AVAIL1 | VNET_BUFFER_F_AVAIL2 | VNET_BUFFER_F_AVAIL3 | \
86   VNET_BUFFER_F_AVAIL4 | VNET_BUFFER_F_AVAIL5 | VNET_BUFFER_F_AVAIL6 | \
87   VNET_BUFFER_F_AVAIL7)
88
89#define VNET_BUFFER_FLAGS_VLAN_BITS \
90  (VNET_BUFFER_F_VLAN_1_DEEP | VNET_BUFFER_F_VLAN_2_DEEP)
91
92enum
93{
94#define _(bit, name, s, v) VNET_BUFFER_F_##name  = (1 << LOG2_VLIB_BUFFER_FLAG_USER(bit)),
95  foreach_vnet_buffer_flag
96#undef _
97};
98
99enum
100{
101#define _(bit, name, s, v) VNET_BUFFER_F_LOG2_##name  = LOG2_VLIB_BUFFER_FLAG_USER(bit),
102  foreach_vnet_buffer_flag
103#undef _
104};
105
106/* Make sure that the vnet and vlib bits are disjoint */
107STATIC_ASSERT (((VNET_BUFFER_FLAGS_ALL_AVAIL & VLIB_BUFFER_FLAGS_ALL) == 0),
108	       "VLIB / VNET buffer flags overlap");
109
110#define foreach_buffer_opaque_union_subtype     \
111_(ip)                                           \
112_(l2)                                           \
113_(l2t)                                          \
114_(l2_classify)                                  \
115_(policer)                                      \
116_(ipsec)					\
117_(map)						\
118_(map_t)					\
119_(ip_frag)					\
120_(mpls)					        \
121_(tcp)
122
123/*
124 * vnet stack buffer opaque array overlay structure.
125 * The vnet_buffer_opaque_t *must* be the same size as the
126 * vlib_buffer_t "opaque" structure member, 32 bytes.
127 *
128 * When adding a union type, please add a stanza to
129 * foreach_buffer_opaque_union_subtype (directly above).
130 * Code in vnet_interface_init(...) verifies the size
131 * of the union, and will announce any deviations in an
132 * impossible-to-miss manner.
133 */
134typedef struct
135{
136  u32 sw_if_index[VLIB_N_RX_TX];
137  i16 l2_hdr_offset;
138  i16 l3_hdr_offset;
139  i16 l4_hdr_offset;
140  u8 feature_arc_index;
141  u8 dont_waste_me;
142
143  union
144  {
145    /* IP4/6 buffer opaque. */
146    struct
147    {
148      /* Adjacency from destination IP address lookup [VLIB_TX].
149         Adjacency from source IP address lookup [VLIB_RX].
150         This gets set to ~0 until source lookup is performed. */
151      u32 adj_index[VLIB_N_RX_TX];
152
153      union
154      {
155	struct
156	{
157	  /* Flow hash value for this packet computed from IP src/dst address
158	     protocol and ports. */
159	  u32 flow_hash;
160
161	  union
162	  {
163	    /* next protocol */
164	    u32 save_protocol;
165
166	    /* Hint for transport protocols */
167	    u32 fib_index;
168	  };
169
170	  /* Rewrite length */
171	  u8 save_rewrite_length;
172
173	  /* MFIB RPF ID */
174	  u32 rpf_id;
175	};
176
177	/* ICMP */
178	struct
179	{
180	  u8 type;
181	  u8 code;
182	  u32 data;
183	} icmp;
184
185	/* reassembly */
186	union
187	{
188	  /* group input/output to simplify the code, this way
189	   * we can handoff while keeping input variables intact */
190	  struct
191	  {
192	    /* input variables */
193	    struct
194	    {
195	      u32 next_index;	/* index of next node - used by custom apps */
196	      u32 error_next_index;	/* index of next node if error - used by custom apps */
197	    };
198	    /* handoff variables */
199	    struct
200	    {
201	      u16 owner_thread_index;
202	    };
203	  };
204	  /* output variables */
205	  struct
206	  {
207	    union
208	    {
209	      /* shallow virtual reassembly output variables */
210	      struct
211	      {
212		u16 l4_src_port;	/* tcp/udp/icmp src port */
213		u16 l4_dst_port;	/* tcp/udp/icmp dst port */
214		u32 tcp_ack_number;
215		u8 save_rewrite_length;
216		u8 ip_proto;	/* protocol in ip header */
217		u8 icmp_type_or_tcp_flags;
218		u8 is_non_first_fragment;
219		u32 tcp_seq_number;
220	      };
221	      /* full reassembly output variables */
222	      struct
223	      {
224		u16 estimated_mtu;	/* estimated MTU calculated during reassembly */
225	      };
226	    };
227	  };
228	  /* internal variables used during reassembly */
229	  struct
230	  {
231	    u16 fragment_first;
232	    u16 fragment_last;
233	    u16 range_first;
234	    u16 range_last;
235	    u32 next_range_bi;
236	    u16 ip6_frag_hdr_offset;
237	  };
238	} reass;
239      };
240    } ip;
241
242    /*
243     * MPLS:
244     * data copied from the MPLS header that was popped from the packet
245     * during the look-up.
246     */
247    struct
248    {
249      /* do not overlay w/ ip.adj_index[0,1] nor flow hash */
250      u32 pad[VLIB_N_RX_TX + 1];
251      u8 ttl;
252      u8 exp;
253      u8 first;
254      u8 pyld_proto:3;		/* dpo_proto_t */
255      u8 rsvd:5;
256      /* Rewrite length */
257      u8 save_rewrite_length;
258      /* Save the mpls header length including all label stack */
259      u8 mpls_hdr_length;
260      /*
261       * BIER - the number of bytes in the header.
262       *  the len field in the header is not authoritative. It's the
263       * value in the table that counts.
264       */
265      struct
266      {
267	u8 n_bytes;
268      } bier;
269    } mpls;
270
271    /* l2 bridging path, only valid there */
272    struct opaque_l2
273    {
274      u32 feature_bitmap;
275      u16 bd_index;		/* bridge-domain index */
276      u16 l2fib_sn;		/* l2fib bd/int seq_num */
277      u8 l2_len;		/* ethernet header length */
278      u8 shg;			/* split-horizon group */
279      u8 bd_age;		/* aging enabled */
280    } l2;
281
282    /* l2tpv3 softwire encap, only valid there */
283    struct
284    {
285      u32 pad[4];		/* do not overlay w/ ip.adj_index[0,1] */
286      u8 next_index;
287      u32 session_index;
288    } l2t;
289
290    /* L2 classify */
291    struct
292    {
293      struct opaque_l2 pad;
294      union
295      {
296	u32 table_index;
297	u32 opaque_index;
298      };
299      u64 hash;
300    } l2_classify;
301
302    /* vnet policer */
303    struct
304    {
305      u32 pad[8 - VLIB_N_RX_TX - 1];	/* to end of opaque */
306      u32 index;
307    } policer;
308
309    /* interface output features */
310    struct
311    {
312      u32 sad_index;
313      u32 protect_index;
314    } ipsec;
315
316    /* MAP */
317    struct
318    {
319      u16 mtu;
320    } map;
321
322    /* MAP-T */
323    struct
324    {
325      u32 map_domain_index;
326      struct
327      {
328	u32 saddr, daddr;
329	u16 frag_offset;	//Fragmentation header offset
330	u16 l4_offset;		//L4 header overall offset
331	u8 l4_protocol;		//The final protocol number
332      } v6;			//Used by ip6_map_t only
333      u16 checksum_offset;	//L4 checksum overall offset
334      u16 mtu;			//Exit MTU
335    } map_t;
336
337    /* IP Fragmentation */
338    struct
339    {
340      u32 pad[2];		/* do not overlay w/ ip.adj_index[0,1] */
341      u16 mtu;
342      u8 next_index;
343      u8 flags;			//See ip_frag.h
344    } ip_frag;
345
346    /* COP - configurable junk filter(s) */
347    struct
348    {
349      /* Current configuration index. */
350      u32 current_config_index;
351    } cop;
352
353    /* LISP */
354    struct
355    {
356      /* overlay address family */
357      u16 overlay_afi;
358    } lisp;
359
360    /* TCP */
361    struct
362    {
363      u32 connection_index;
364      union
365      {
366	u32 seq_number;
367	u32 next_node_opaque;
368      };
369      u32 seq_end;
370      u32 ack_number;
371      u16 hdr_offset;		/**< offset relative to ip hdr */
372      u16 data_offset;		/**< offset relative to ip hdr */
373      u16 data_len;		/**< data len */
374      u8 flags;
375    } tcp;
376
377    /* SNAT */
378    struct
379    {
380      u32 flags;
381    } snat;
382
383    u32 unused[6];
384  };
385} vnet_buffer_opaque_t;
386
387#define VNET_REWRITE_TOTAL_BYTES (VLIB_BUFFER_PRE_DATA_SIZE)
388
389STATIC_ASSERT (STRUCT_SIZE_OF (vnet_buffer_opaque_t, ip.save_rewrite_length)
390	       == STRUCT_SIZE_OF (vnet_buffer_opaque_t,
391				  ip.reass.save_rewrite_length)
392	       && STRUCT_SIZE_OF (vnet_buffer_opaque_t,
393				  ip.reass.save_rewrite_length) ==
394	       STRUCT_SIZE_OF (vnet_buffer_opaque_t, mpls.save_rewrite_length)
395	       && STRUCT_SIZE_OF (vnet_buffer_opaque_t,
396				  mpls.save_rewrite_length) == 1
397	       && VNET_REWRITE_TOTAL_BYTES < UINT8_MAX,
398	       "save_rewrite_length member must be able to hold the max value of rewrite length");
399
400STATIC_ASSERT (STRUCT_OFFSET_OF (vnet_buffer_opaque_t, ip.save_rewrite_length)
401	       == STRUCT_OFFSET_OF (vnet_buffer_opaque_t,
402				    ip.reass.save_rewrite_length)
403	       && STRUCT_OFFSET_OF (vnet_buffer_opaque_t,
404				    mpls.save_rewrite_length) ==
405	       STRUCT_OFFSET_OF (vnet_buffer_opaque_t,
406				 ip.reass.save_rewrite_length),
407	       "save_rewrite_length must be aligned so that reass doesn't overwrite it");
408
409/*
410 * The opaque field of the vlib_buffer_t is interpreted as a
411 * vnet_buffer_opaque_t. Hence it should be big enough to accommodate one.
412 */
413STATIC_ASSERT (sizeof (vnet_buffer_opaque_t) <=
414	       STRUCT_SIZE_OF (vlib_buffer_t, opaque),
415	       "VNET buffer meta-data too large for vlib_buffer");
416
417#define vnet_buffer(b) ((vnet_buffer_opaque_t *) (b)->opaque)
418
419/* Full cache line (64 bytes) of additional space */
420typedef struct
421{
422  /**
423   * QoS marking data that needs to persist from the recording nodes
424   * (nominally in the ingress path) to the marking node (in the
425   * egress path)
426   */
427  struct
428  {
429    u8 bits;
430    u8 source;
431  } qos;
432
433  u8 loop_counter;
434  u8 __unused[1];
435
436  /* Group Based Policy */
437  struct
438  {
439    u8 __unused;
440    u8 flags;
441    u16 sclass;
442  } gbp;
443
444  /**
445   * The L4 payload size set on input on GSO enabled interfaces
446   * when we receive a GSO packet (a chain of buffers with the first one
447   * having GSO bit set), and needs to persist all the way to the interface-output,
448   * in case the egress interface is not GSO-enabled - then we need to perform
449   * the segmentation, and use this value to cut the payload appropriately.
450   */
451  u16 gso_size;
452  /* size of L4 prototol header */
453  u16 gso_l4_hdr_sz;
454
455  /* The union below has a u64 alignment, so this space is unused */
456  u32 __unused2[1];
457
458  struct
459  {
460    u32 arc_next;
461    u32 ed_out2in_nat_session_index;
462  } nat;
463
464  union
465  {
466    struct
467    {
468#if VLIB_BUFFER_TRACE_TRAJECTORY > 0
469      /* buffer trajectory tracing */
470      u16 *trajectory_trace;
471#endif
472    };
473    struct
474    {
475      u64 pad[1];
476      u64 pg_replay_timestamp;
477    };
478    u32 unused[8];
479  };
480} vnet_buffer_opaque2_t;
481
482#define vnet_buffer2(b) ((vnet_buffer_opaque2_t *) (b)->opaque2)
483
484/*
485 * The opaque2 field of the vlib_buffer_t is interpreted as a
486 * vnet_buffer_opaque2_t. Hence it should be big enough to accommodate one.
487 */
488STATIC_ASSERT (sizeof (vnet_buffer_opaque2_t) <=
489	       STRUCT_SIZE_OF (vlib_buffer_t, opaque2),
490	       "VNET buffer opaque2 meta-data too large for vlib_buffer");
491
492#define gso_mtu_sz(b) (vnet_buffer2(b)->gso_size + \
493                       vnet_buffer2(b)->gso_l4_hdr_sz + \
494                       vnet_buffer(b)->l4_hdr_offset - \
495                       vnet_buffer (b)->l3_hdr_offset)
496
497
498format_function_t format_vnet_buffer;
499
500#endif /* included_vnet_buffer_h */
501
502/*
503 * fd.io coding-style-patch-verification: ON
504 *
505 * Local Variables:
506 * eval: (c-set-style "gnu")
507 * End:
508 */
509