Common Trace Format

1Common Trace Format (CTF) Specification (v1.8.1)
2
3Mathieu Desnoyers, EfficiOS Inc.
4
5The goal of the present document is to specify a trace format that suits the
6needs of the embedded, telecom, high-performance and kernel communities. It is
7based on the Common Trace Format Requirements (v1.4) document. It is designed to
8allow traces to be natively generated by the Linux kernel, Linux user-space
9applications written in C/C++, and hardware components. One major element of
10CTF is the Trace Stream Description Language (TSDL) which flexibility
11enables description of various binary trace stream layouts.
12
13The latest version of this document can be found at:
14
15  git tree:   git://git.efficios.com/ctf.git
16  gitweb:     http://git.efficios.com/?p=ctf.git
17
18A reference implementation of a library to read and write this trace format is
19being implemented within the BabelTrace project, a converter between trace
20formats. The development tree is available at:
21
22  git tree:   git://git.efficios.com/babeltrace.git
23  gitweb:     http://git.efficios.com/?p=babeltrace.git
24
25The CE Workgroup of the Linux Foundation, Ericsson, and EfficiOS have
26sponsored this work.
27
28
29Table of Contents
30
311. Preliminary definitions
322. High-level representation of a trace
333. Event stream
344. Types
35   4.1 Basic types
36       4.1.1 Type inheritance
37       4.1.2 Alignment
38       4.1.3 Byte order
39       4.1.4 Size
40       4.1.5 Integers
41       4.1.6 GNU/C bitfields
42       4.1.7 Floating point
43       4.1.8 Enumerations
444.2 Compound types
45    4.2.1 Structures
46    4.2.2 Variants (Discriminated/Tagged Unions)
47    4.2.3 Arrays
48    4.2.4 Sequences
49    4.2.5 Strings
505. Event Packet Header
51   5.1 Event Packet Header Description
52   5.2 Event Packet Context Description
536. Event Structure
54   6.1 Event Header
55       6.1.1 Type 1 - Few event IDs
56       6.1.2 Type 2 - Many event IDs
57   6.2 Event Context
58   6.3 Event Payload
59       6.3.1 Padding
60       6.3.2 Alignment
617. Trace Stream Description Language (TSDL)
62   7.1 Meta-data
63   7.2 Declaration vs Definition
64   7.3 TSDL Scopes
65       7.3.1 Lexical Scope
66       7.3.2 Static and Dynamic Scopes
67   7.4 TSDL Examples
688. Clocks
69
70
711. Preliminary definitions
72
73  - Event Trace: An ordered sequence of events.
74  - Event Stream: An ordered sequence of events, containing a subset of the
75                  trace event types.
76  - Event Packet: A sequence of physically contiguous events within an event
77                  stream.
78  - Event: This is the basic entry in a trace. (aka: a trace record).
79    - An event identifier (ID) relates to the class (a type) of event within
80      an event stream.
81        e.g. event: irq_entry.
82    - An event (or event record) relates to a specific instance of an event
83      class.
84        e.g. event: irq_entry, at time X, on CPU Y
85  - Source Architecture: Architecture writing the trace.
86  - Reader Architecture: Architecture reading the trace.
87
88
892. High-level representation of a trace
90
91A trace is divided into multiple event streams. Each event stream contains a
92subset of the trace event types.
93
94The final output of the trace, after its generation and optional transport over
95the network, is expected to be either on permanent or temporary storage in a
96virtual file system. Because each event stream is appended to while a trace is
97being recorded, each is associated with a distinct set of files for
98output. Therefore, a stored trace can be represented as a directory
99containing zero, one or more files per stream.
100
101Meta-data description associated with the trace contains information on
102trace event types expressed in the Trace Stream Description Language
103(TSDL). This language describes:
104
105- Trace version.
106- Types available.
107- Per-trace event header description.
108- Per-stream event header description.
109- Per-stream event context description.
110- Per-event
111  - Event type to stream mapping.
112  - Event type to name mapping.
113  - Event type to ID mapping.
114  - Event context description.
115  - Event fields description.
116
117
1183. Event stream
119
120An event stream can be divided into contiguous event packets of variable
121size. These subdivisions have a variable size. An event packet can
122contain a certain amount of padding at the end. The stream header is
123repeated at the beginning of each event packet. The rationale for the
124event stream design choices is explained in Appendix B. Stream Header
125Rationale.
126
127The event stream header will therefore be referred to as the "event packet
128header" throughout the rest of this document.
129
130
1314. Types
132
133Types are organized as type classes. Each type class belong to either of two
134kind of types: basic types or compound types.
135
1364.1 Basic types
137
138A basic type is a scalar type, as described in this section. It includes
139integers, GNU/C bitfields, enumerations, and floating point values.
140
1414.1.1 Type inheritance
142
143Type specifications can be inherited to allow deriving types from a
144type class. For example, see the uint32_t named type derived from the "integer"
145type class below ("Integers" section). Types have a precise binary
146representation in the trace. A type class has methods to read and write these
147types, but must be derived into a type to be usable in an event field.
148
1494.1.2 Alignment
150
151We define "byte-packed" types as aligned on the byte size, namely 8-bit.
152We define "bit-packed" types as following on the next bit, as defined by the
153"Integers" section.
154
155Each basic type must specify its alignment, in bits. Examples of
156possible alignments are: bit-packed (align = 1), byte-packed (align =
1578), or word-aligned (e.g. align = 32 or align = 64). The choice depends
158on the architecture preference and compactness vs performance trade-offs
159of the implementation.  Architectures providing fast unaligned write
160byte-packed basic types to save space, aligning each type on byte
161boundaries (8-bit). Architectures with slow unaligned writes align types
162on specific alignment values. If no specific alignment is declared for a
163type, it is assumed to be bit-packed for integers with size not multiple
164of 8 bits and for gcc bitfields. All other basic types are byte-packed
165by default. It is however recommended to always specify the alignment
166explicitly. Alignment values must be power of two. Compound types are
167aligned as specified in their individual specification.
168
169TSDL meta-data attribute representation of a specific alignment:
170
171  align = value;                                /* value in bits */
172
1734.1.3 Byte order
174
175By default, the native endianness of the source architecture the trace is used.
176Byte order can be overridden for a basic type by specifying a "byte_order"
177attribute. Typical use-case is to specify the network byte order (big endian:
178"be") to save data captured from the network into the trace without conversion.
179If not specified, the byte order is native.
180
181TSDL meta-data representation:
182
183  byte_order = native OR network OR be OR le;	/* network and be are aliases */
184
1854.1.4 Size
186
187Type size, in bits, for integers and floats is that returned by "sizeof()" in C
188multiplied by CHAR_BIT.
189We require the size of "char" and "unsigned char" types (CHAR_BIT) to be fixed
190to 8 bits for cross-endianness compatibility.
191
192TSDL meta-data representation:
193
194  size = value;    (value is in bits)
195
1964.1.5 Integers
197
198Signed integers are represented in two-complement. Integer alignment,
199size, signedness and byte ordering are defined in the TSDL meta-data.
200Integers aligned on byte size (8-bit) and with length multiple of byte
201size (8-bit) correspond to the C99 standard integers. In addition,
202integers with alignment and/or size that are _not_ a multiple of the
203byte size are permitted; these correspond to the C99 standard bitfields,
204with the added specification that the CTF integer bitfields have a fixed
205binary representation. A MIT-licensed reference implementation of the
206CTF portable bitfields is available at:
207
208  http://git.efficios.com/?p=babeltrace.git;a=blob;f=include/babeltrace/bitfield.h
209
210Binary representation of integers:
211
212- On little and big endian:
213  - Within a byte, high bits correspond to an integer high bits, and low bits
214    correspond to low bits.
215- On little endian:
216  - Integer across multiple bytes are placed from the less significant to the
217    most significant.
218  - Consecutive integers are placed from lower bits to higher bits (even within
219    a byte).
220- On big endian:
221  - Integer across multiple bytes are placed from the most significant to the
222    less significant.
223  - Consecutive integers are placed from higher bits to lower bits (even within
224    a byte).
225
226This binary representation is derived from the bitfield implementation in GCC
227for little and big endian. However, contrary to what GCC does, integers can
228cross units boundaries (no padding is required). Padding can be explicitly
229added (see 4.1.6 GNU/C bitfields) to follow the GCC layout if needed.
230
231TSDL meta-data representation:
232
233  integer {
234    signed = true OR false;                     /* default false */
235    byte_order = native OR network OR be OR le; /* default native */
236    size = value;                               /* value in bits, no default */
237    align = value;                              /* value in bits */
238    /* based used for pretty-printing output, default: decimal. */
239    base = decimal OR dec OR OR d OR i OR u OR 10 OR hexadecimal OR hex OR x OR X OR p OR 16
240           OR octal OR oct OR o OR 8 OR binary OR b OR 2;
241    /* character encoding, default: none */
242    encoding = none or UTF8 or ASCII;
243  }
244
245Example of type inheritance (creation of a uint32_t named type):
246
247typealias integer {
248  size = 32;
249  signed = false;
250  align = 32;
251} := uint32_t;
252
253Definition of a named 5-bit signed bitfield:
254
255typealias integer {
256  size = 5;
257  signed = true;
258  align = 1;
259} := int5_t;
260
261The character encoding field can be used to specify that the integer
262must be printed as a text character when read. e.g.:
263
264typealias integer {
265  size = 8;
266  align = 8;
267  signed = false;
268  encoding = UTF8;
269} := utf_char;
270
271
2724.1.6 GNU/C bitfields
273
274The GNU/C bitfields follow closely the integer representation, with a
275particularity on alignment: if a bitfield cannot fit in the current unit, the
276unit is padded and the bitfield starts at the following unit. The unit size is
277defined by the size of the type "unit_type".
278
279TSDL meta-data representation:
280
281  unit_type name:size;
282
283As an example, the following structure declared in C compiled by GCC:
284
285struct example {
286  short a:12;
287  short b:5;
288};
289
290The example structure is aligned on the largest element (short). The second
291bitfield would be aligned on the next unit boundary, because it would not fit in
292the current unit.
293
2944.1.7 Floating point
295
296The floating point values byte ordering is defined in the TSDL meta-data.
297
298Floating point values follow the IEEE 754-2008 standard interchange formats.
299Description of the floating point values include the exponent and mantissa size
300in bits. Some requirements are imposed on the floating point values:
301
302- FLT_RADIX must be 2.
303- mant_dig is the number of digits represented in the mantissa. It is specified
304  by the ISO C99 standard, section 5.2.4, as FLT_MANT_DIG, DBL_MANT_DIG and
305  LDBL_MANT_DIG as defined by <float.h>.
306- exp_dig is the number of digits represented in the exponent. Given that
307  mant_dig is one bit more than its actual size in bits (leading 1 is not
308  needed) and also given that the sign bit always takes one bit, exp_dig can be
309  specified as:
310
311  - sizeof(float) * CHAR_BIT - FLT_MANT_DIG
312  - sizeof(double) * CHAR_BIT - DBL_MANT_DIG
313  - sizeof(long double) * CHAR_BIT - LDBL_MANT_DIG
314
315TSDL meta-data representation:
316
317floating_point {
318  exp_dig = value;
319  mant_dig = value;
320  byte_order = native OR network OR be OR le;
321  align = value;
322}
323
324Example of type inheritance:
325
326typealias floating_point {
327  exp_dig = 8;         /* sizeof(float) * CHAR_BIT - FLT_MANT_DIG */
328  mant_dig = 24;       /* FLT_MANT_DIG */
329  byte_order = native;
330  align = 32;
331} := float;
332
333TODO: define NaN, +inf, -inf behavior.
334
335Bit-packed, byte-packed or larger alignments can be used for floating
336point values, similarly to integers.
337
3384.1.8 Enumerations
339
340Enumerations are a mapping between an integer type and a table of strings. The
341numerical representation of the enumeration follows the integer type specified
342by the meta-data. The enumeration mapping table is detailed in the enumeration
343description within the meta-data. The mapping table maps inclusive value
344ranges (or single values) to strings. Instead of being limited to simple
345"value -> string" mappings, these enumerations map
346"[ start_value ... end_value ] -> string", which map inclusive ranges of
347values to strings.  An enumeration from the C language can be represented in
348this format by having the same start_value and end_value for each element, which
349is in fact a range of size 1. This single-value range is supported without
350repeating the start and end values with the value = string declaration.
351
352enum name : integer_type {
353  somestring          = start_value1 ... end_value1,
354  "other string"      = start_value2 ... end_value2,
355  yet_another_string,	/* will be assigned to end_value2 + 1 */
356  "some other string" = value,
357  ...
358};
359
360If the values are omitted, the enumeration starts at 0 and increment of 1 for
361each entry:
362
363enum name : unsigned int {
364  ZERO,
365  ONE,
366  TWO,
367  TEN = 10,
368  ELEVEN,
369};
370
371Overlapping ranges within a single enumeration are implementation defined.
372
373A nameless enumeration can be declared as a field type or as part of a typedef:
374
375enum : integer_type {
376  ...
377}
378
379Enumerations omitting the container type ": integer_type" use the "int"
380type (for compatibility with C99). The "int" type must be previously
381declared. E.g.:
382
383typealias integer { size = 32; align = 32; signed = true } := int;
384
385enum {
386  ...
387}
388
389
3904.2 Compound types
391
392Compound are aggregation of type declarations. Compound types include
393structures, variant, arrays, sequences, and strings.
394
3954.2.1 Structures
396
397Structures are aligned on the largest alignment required by basic types
398contained within the structure. (This follows the ISO/C standard for structures)
399
400TSDL meta-data representation of a named structure:
401
402struct name {
403  field_type field_name;
404  field_type field_name;
405  ...
406}; 
407
408Example:
409
410struct example {
411  integer {                       /* Nameless type */
412    size = 16;
413    signed = true;
414    align = 16;
415  } first_field_name;
416  uint64_t second_field_name;  /* Named type declared in the meta-data */
417};
418
419The fields are placed in a sequence next to each other. They each
420possess a field name, which is a unique identifier within the structure.
421The identifier is not allowed to use any reserved keyword
422(see Section C.1.2). Replacing reserved keywords with
423underscore-prefixed field names is recommended. Fields starting with an
424underscore should have their leading underscore removed by the CTF trace
425readers.
426
427A nameless structure can be declared as a field type or as part of a typedef:
428
429struct {
430  ...
431}
432
433Alignment for a structure compound type can be forced to a minimum value
434by adding an "align" specifier after the declaration of a structure
435body. This attribute is read as: align(value). The value is specified in
436bits. The structure will be aligned on the maximum value between this
437attribute and the alignment required by the basic types contained within
438the structure. e.g.
439
440struct {
441  ...
442} align(32)
443
4444.2.2 Variants (Discriminated/Tagged Unions)
445
446A CTF variant is a selection between different types. A CTF variant must
447always be defined within the scope of a structure or within fields
448contained within a structure (defined recursively). A "tag" enumeration
449field must appear in either the same static scope, prior to the variant
450field (in field declaration order), in an upper static scope , or in an
451upper dynamic scope (see Section 7.3.2). The type selection is indicated
452by the mapping from the enumeration value to the string used as variant
453type selector. The field to use as tag is specified by the "tag_field",
454specified between "< >" after the "variant" keyword for unnamed
455variants, and after "variant name" for named variants.
456
457The alignment of the variant is the alignment of the type as selected by the tag
458value for the specific instance of the variant. The alignment of the type
459containing the variant is independent of the variant alignment.  The size of the
460variant is the size as selected by the tag value for the specific instance of
461the variant.
462
463Each variant type selector possess a field name, which is a unique
464identifier within the variant. The identifier is not allowed to use any
465reserved keyword (see Section C.1.2). Replacing reserved keywords with
466underscore-prefixed field names is recommended. Fields starting with an
467underscore should have their leading underscore removed by the CTF trace
468readers.
469
470
471A named variant declaration followed by its definition within a structure
472declaration:
473
474variant name {
475  field_type sel1;
476  field_type sel2;
477  field_type sel3;
478  ...
479};
480
481struct {
482  enum : integer_type { sel1, sel2, sel3, ... } tag_field;
483  ...
484  variant name <tag_field> v;
485}
486
487An unnamed variant definition within a structure is expressed by the following
488TSDL meta-data:
489
490struct {
491  enum : integer_type { sel1, sel2, sel3, ... } tag_field;
492  ...
493  variant <tag_field> {
494    field_type sel1;
495    field_type sel2;
496    field_type sel3;
497    ...
498  } v;
499}
500
501Example of a named variant within a sequence that refers to a single tag field:
502
503variant example {
504  uint32_t a;
505  uint64_t b;
506  short c;
507};
508
509struct {
510  enum : uint2_t { a, b, c } choice;
511  unsigned int seqlen;
512  variant example <choice> v[seqlen];
513}
514
515Example of an unnamed variant:
516
517struct {
518  enum : uint2_t { a, b, c, d } choice;
519  /* Unrelated fields can be added between the variant and its tag */
520  int32_t somevalue;
521  variant <choice> {
522    uint32_t a;
523    uint64_t b;
524    short c;
525    struct {
526      unsigned int field1;
527      uint64_t field2;
528    } d;
529  } s;
530}
531
532Example of an unnamed variant within an array:
533
534struct {
535  enum : uint2_t { a, b, c } choice;
536  variant <choice> {
537    uint32_t a;
538    uint64_t b;
539    short c;
540  } v[10];
541}
542
543Example of a variant type definition within a structure, where the defined type
544is then declared within an array of structures. This variant refers to a tag
545located in an upper static scope. This example clearly shows that a variant
546type definition referring to the tag "x" uses the closest preceding field from
547the static scope of the type definition.
548
549struct {
550  enum : uint2_t { a, b, c, d } x;
551
552  typedef variant <x> {	/*
553			 * "x" refers to the preceding "x" enumeration in the
554			 * static scope of the type definition.
555			 */
556    uint32_t a;
557    uint64_t b;
558    short c;
559  } example_variant;
560
561  struct {
562    enum : int { x, y, z } x;	/* This enumeration is not used by "v". */
563    example_variant v; 		/*
564				 * "v" uses the "enum : uint2_t { a, b, c, d }"
565				 * tag.
566				 */
567  } a[10];
568}
569
5704.2.3 Arrays
571
572Arrays are fixed-length. Their length is declared in the type
573declaration within the meta-data. They contain an array of "inner type"
574elements, which can refer to any type not containing the type of the
575array being declared (no circular dependency). The length is the number
576of elements in an array.
577
578TSDL meta-data representation of a named array:
579
580typedef elem_type name[length];
581
582A nameless array can be declared as a field type within a structure, e.g.:
583
584  uint8_t field_name[10];
585
586Arrays are always aligned on their element alignment requirement.
587
5884.2.4 Sequences
589
590Sequences are dynamically-sized arrays. They refer to a a "length"
591unsigned integer field, which must appear in either the same static scope,
592prior to the sequence field (in field declaration order), in an upper
593static scope, or in an upper dynamic scope (see Section 7.3.2). This
594length field represents the number of elements in the sequence. The
595sequence per se is an array of "inner type" elements.
596
597TSDL meta-data representation for a sequence type definition:
598
599struct {
600  unsigned int length_field;
601  typedef elem_type typename[length_field];
602  typename seq_field_name;
603}
604
605A sequence can also be declared as a field type, e.g.:
606
607struct {
608  unsigned int length_field;
609  long seq_field_name[length_field];
610}
611
612Multiple sequences can refer to the same length field, and these length
613fields can be in a different upper dynamic scope:
614
615e.g., assuming the stream.event.header defines:
616
617stream {
618  ...
619  id = 1;
620  event.header := struct {
621    uint16_t seq_len;
622  };
623};
624
625event {
626  ...
627  stream_id = 1;
628  fields := struct {
629    long seq_a[stream.event.header.seq_len];
630    char seq_b[stream.event.header.seq_len];
631  };
632};
633
634The sequence elements follow the "array" specifications.
635
6364.2.5 Strings
637
638Strings are an array of bytes of variable size and are terminated by a '\0'
639"NULL" character.  Their encoding is described in the TSDL meta-data. In
640absence of encoding attribute information, the default encoding is
641UTF-8.
642
643TSDL meta-data representation of a named string type:
644
645typealias string {
646  encoding = UTF8 OR ASCII;
647} := name;
648
649A nameless string type can be declared as a field type:
650
651string field_name;	/* Use default UTF8 encoding */
652
653Strings are always aligned on byte size.
654
6555. Event Packet Header
656
657The event packet header consists of two parts: the "event packet header"
658is the same for all streams of a trace. The second part, the "event
659packet context", is described on a per-stream basis. Both are described
660in the TSDL meta-data. The packets are aligned on architecture-page-sized
661addresses.
662
663Event packet header (all fields are optional, specified by TSDL meta-data):
664
665- Magic number (CTF magic number: 0xC1FC1FC1) specifies that this is a
666  CTF packet. This magic number is optional, but when present, it should
667  come at the very beginning of the packet.
668- Trace UUID, used to ensure the event packet match the meta-data used.
669  (note: we cannot use a meta-data checksum in every cases instead of a
670   UUID because meta-data can be appended to while tracing is active)
671  This field is optional.
672- Stream ID, used as reference to stream description in meta-data.
673  This field is optional if there is only one stream description in the
674  meta-data, but becomes required if there are more than one stream in
675  the TSDL meta-data description.
676
677Event packet context (all fields are optional, specified by TSDL meta-data):
678
679- Event packet content size (in bits).
680- Event packet size (in bits, includes padding).
681- Event packet content checksum. Checksum excludes the event packet
682  header.
683- Per-stream event packet sequence count (to deal with UDP packet loss). The
684  number of significant sequence counter bits should also be present, so
685  wrap-arounds are dealt with correctly.
686- Time-stamp at the beginning and time-stamp at the end of the event packet.
687  Both timestamps are written in the packet header, but sampled respectively
688  while (or before) writing the first event and while (or after) writing the
689  last event in the packet. The inclusive range between these timestamps should
690  include all event timestamps assigned to events contained within the packet.
691- Events discarded count
692  - Snapshot of a per-stream free-running counter, counting the number of
693    events discarded that were supposed to be written in the stream prior to
694    the first event in the event packet.
695    * Note: producer-consumer buffer full condition should fill the current
696            event packet with padding so we know exactly where events have been
697            discarded.
698- Lossless compression scheme used for the event packet content. Applied
699  directly to raw data. New types of compression can be added in following
700  versions of the format.
701  0: no compression scheme
702  1: bzip2
703  2: gzip
704  3: xz
705- Cypher used for the event packet content. Applied after compression.
706  0: no encryption
707  1: AES
708- Checksum scheme used for the event packet content. Applied after encryption.
709  0: no checksum
710  1: md5
711  2: sha1
712  3: crc32
713
7145.1 Event Packet Header Description
715
716The event packet header layout is indicated by the trace packet.header
717field. Here is a recommended structure type for the packet header with
718the fields typically expected (although these fields are each optional):
719
720struct event_packet_header {
721  uint32_t magic;
722  uint8_t  uuid[16];
723  uint32_t stream_id;
724};
725
726trace {
727  ...
728  packet.header := struct event_packet_header;
729};
730
731If the magic number is not present, tools such as "file" will have no
732mean to discover the file type.
733
734If the uuid is not present, no validation that the meta-data actually
735corresponds to the stream is performed.
736
737If the stream_id packet header field is missing, the trace can only
738contain a single stream. Its "id" field can be left out, and its events
739don't need to declare a "stream_id" field.
740
741
7425.2 Event Packet Context Description
743
744Event packet context example. These are declared within the stream declaration
745in the meta-data. All these fields are optional. If the packet size field is
746missing, the whole stream only contains a single packet. If the content
747size field is missing, the packet is filled (no padding). The content
748and packet sizes include all headers.
749
750An example event packet context type:
751
752struct event_packet_context {
753  uint64_t timestamp_begin;
754  uint64_t timestamp_end;
755  uint32_t checksum;
756  uint32_t stream_packet_count;
757  uint32_t events_discarded;
758  uint32_t cpu_id;
759  uint32_t/uint16_t content_size;
760  uint32_t/uint16_t packet_size;
761  uint8_t  compression_scheme;
762  uint8_t  encryption_scheme;
763  uint8_t  checksum_scheme;
764};
765
766
7676. Event Structure
768
769The overall structure of an event is:
770
7711 - Stream Packet Context (as specified by the stream meta-data)
772 2 - Event Header (as specified by the stream meta-data)
773  3 - Stream Event Context (as specified by the stream meta-data)
774   4 - Event Context (as specified by the event meta-data)
775    5 - Event Payload (as specified by the event meta-data)
776
777This structure defines an implicit dynamic scoping, where variants
778located in inner structures (those with a higher number in the listing
779above) can refer to the fields of outer structures (with lower number in
780the listing above). See Section 7.3 TSDL Scopes for more detail.
781
7826.1 Event Header
783
784Event headers can be described within the meta-data. We hereby propose, as an
785example, two types of events headers. Type 1 accommodates streams with less than
78631 event IDs. Type 2 accommodates streams with 31 or more event IDs.
787
788One major factor can vary between streams: the number of event IDs assigned to
789a stream. Luckily, this information tends to stay relatively constant (modulo
790event registration while trace is being recorded), so we can specify different
791representations for streams containing few event IDs and streams containing
792many event IDs, so we end up representing the event ID and time-stamp as
793densely as possible in each case.
794
795The header is extended in the rare occasions where the information cannot be
796represented in the ranges available in the standard event header. They are also
797used in the rare occasions where the data required for a field could not be
798collected: the flag corresponding to the missing field within the missing_fields
799array is then set to 1.
800
801Types uintX_t represent an X-bit unsigned integer, as declared with
802either:
803
804  typealias integer { size = X; align = X; signed = false } := uintX_t;
805
806    or
807
808  typealias integer { size = X; align = 1; signed = false } := uintX_t;
809
8106.1.1 Type 1 - Few event IDs
811
812  - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
813    preference).
814  - Native architecture byte ordering.
815  - For "compact" selection
816    - Fixed size: 32 bits.
817  - For "extended" selection
818    - Size depends on the architecture and variant alignment.
819
820struct event_header_1 {
821  /*
822   * id: range: 0 - 30.
823   * id 31 is reserved to indicate an extended header.
824   */
825  enum : uint5_t { compact = 0 ... 30, extended = 31 } id;
826  variant <id> {
827    struct {
828      uint27_t timestamp;
829    } compact;
830    struct {
831      uint32_t id;			 /* 32-bit event IDs */
832      uint64_t timestamp;		 /* 64-bit timestamps */
833    } extended;
834  } v;
835} align(32);	/* or align(8) */
836
837
8386.1.2 Type 2 - Many event IDs
839
840  - Aligned on 16-bit (or 8-bit if byte-packed, depending on the architecture
841    preference).
842  - Native architecture byte ordering.
843  - For "compact" selection
844    - Size depends on the architecture and variant alignment.
845  - For "extended" selection
846    - Size depends on the architecture and variant alignment.
847
848struct event_header_2 {
849  /*
850   * id: range: 0 - 65534.
851   * id 65535 is reserved to indicate an extended header.
852   */
853  enum : uint16_t { compact = 0 ... 65534, extended = 65535 } id;
854  variant <id> {
855    struct {
856      uint32_t timestamp;
857    } compact;
858    struct {
859      uint32_t id;			 /* 32-bit event IDs */
860      uint64_t timestamp;		 /* 64-bit timestamps */ 
861    } extended;
862  } v;
863} align(16);	/* or align(8) */
864
865
8666.2 Event Context
867
868The event context contains information relative to the current event.
869The choice and meaning of this information is specified by the TSDL
870stream and event meta-data descriptions. The stream context is applied
871to all events within the stream. The stream context structure follows
872the event header. The event context is applied to specific events. Its
873structure follows the stream context structure.
874
875An example of stream-level event context is to save the event payload size with
876each event, or to save the current PID with each event.  These are declared
877within the stream declaration within the meta-data:
878
879  stream {
880    ...
881    event.context := struct {
882        uint pid;
883        uint16_t payload_size;
884    };
885  };
886
887An example of event-specific event context is to declare a bitmap of missing
888fields, only appended after the stream event context if the extended event
889header is selected. NR_FIELDS is the number of fields within the event (a
890numeric value).
891
892  event {
893    context = struct {
894      variant <id> {
895        struct { } compact;
896        struct {
897          uint1_t missing_fields[NR_FIELDS]; /* missing event fields bitmap */
898        } extended;
899      } v;
900    };
901    ...
902  }
903
9046.3 Event Payload
905
906An event payload contains fields specific to a given event type. The fields
907belonging to an event type are described in the event-specific meta-data
908within a structure type.
909
9106.3.1 Padding
911
912No padding at the end of the event payload. This differs from the ISO/C standard
913for structures, but follows the CTF standard for structures. In a trace, even
914though it makes sense to align the beginning of a structure, it really makes no
915sense to add padding at the end of the structure, because structures are usually
916not followed by a structure of the same type.
917
918This trick can be done by adding a zero-length "end" field at the end of the C
919structures, and by using the offset of this field rather than using sizeof()
920when calculating the size of a structure (see Appendix "A. Helper macros").
921
9226.3.2 Alignment
923
924The event payload is aligned on the largest alignment required by types
925contained within the payload. (This follows the ISO/C standard for structures)
926
927
9287. Trace Stream Description Language (TSDL)
929
930The Trace Stream Description Language (TSDL) allows expression of the
931binary trace streams layout in a C99-like Domain Specific Language
932(DSL).
933
934
9357.1 Meta-data
936
937The trace stream layout description is located in the trace meta-data.
938The meta-data is itself located in a stream identified by its name:
939"metadata".
940
941The meta-data description can be expressed in two different formats:
942text-only and packet-based. The text-only description facilitates
943generation of meta-data and provides a convenient way to enter the
944meta-data information by hand. The packet-based meta-data provides the
945CTF stream packet facilities (checksumming, compression, encryption,
946network-readiness) for meta-data stream generated and transported by a
947tracer.
948
949The text-only meta-data file is a plain-text TSDL description. This file
950must begin with the following characters to identify the file as a CTF
951TSDL text-based metadata file (without the double-quotes) :
952
953"/* CTF"
954
955It must be followed by a space, and the version of the specification
956followed by the CTF trace, e.g.:
957
958" 1.8"
959
960These characters allow automated discovery of file type and CTF
961specification version. They are interpreted as a the beginning of a
962comment by the TSDL metadata parser.  The comment can be continued to
963contain extra commented characters before it is closed.
964
965The packet-based meta-data is made of "meta-data packets", which each
966start with a meta-data packet header. The packet-based meta-data
967description is detected by reading the magic number "0x75D11D57" at the
968beginning of the file. This magic number is also used to detect the
969endianness of the architecture by trying to read the CTF magic number
970and its counterpart in reversed endianness. The events within the
971meta-data stream have no event header nor event context. Each event only
972contains a "sequence" payload, which is a sequence of bits using the
973"trace.packet.header.content_size" field as a placeholder for its length
974(the packet header size should be substracted). The formatting of this
975sequence of bits is a plain-text representation of the TSDL description.
976Each meta-data packet start with a special packet header, specific to
977the meta-data stream, which contains, exactly:
978
979struct metadata_packet_header {
980  uint32_t magic;			/* 0x75D11D57 */
981  uint8_t  uuid[16];			/* Unique Universal Identifier */
982  uint32_t checksum;			/* 0 if unused */
983  uint32_t content_size;		/* in bits */
984  uint32_t packet_size;			/* in bits */
985  uint8_t  compression_scheme;		/* 0 if unused */
986  uint8_t  encryption_scheme;		/* 0 if unused */
987  uint8_t  checksum_scheme;		/* 0 if unused */
988  uint8_t  major;			/* CTF spec version major number */
989  uint8_t  minor;			/* CTF spec version minor number */
990};
991
992The packet-based meta-data can be converted to a text-only meta-data by
993concatenating all the strings in contains.
994
995In the textual representation of the meta-data, the text contained
996within "/*" and "*/", as well as within "//" and end of line, are
997treated as comments.  Boolean values can be represented as true, TRUE,
998or 1 for true, and false, FALSE, or 0 for false. Within the string-based
999meta-data description, the trace UUID is represented as a string of
1000hexadecimal digits and dashes "-". In the event packet header, the trace
1001UUID is represented as an array of bytes.
1002
1003
10047.2 Declaration vs Definition
1005
1006A declaration associates a layout to a type, without specifying where
1007this type is located in the event structure hierarchy (see Section 6).
1008This therefore includes typedef, typealias, as well as all type
1009specifiers. In certain circumstances (typedef, structure field and
1010variant field), a declaration is followed by a declarator, which specify
1011the newly defined type name (for typedef), or the field name (for
1012declarations located within structure and variants). Array and sequence,
1013declared with square brackets ("[" "]"), are part of the declarator,
1014similarly to C99. The enumeration base type is specified by
1015": enum_base", which is part of the type specifier. The variant tag
1016name, specified between "<" ">", is also part of the type specifier.
1017
1018A definition associates a type to a location in the event structure
1019hierarchy (see Section 6). This association is denoted by ":=", as shown
1020in Section 7.3.
1021
1022
10237.3 TSDL Scopes
1024
1025TSDL uses three different types of scoping: a lexical scope is used for
1026declarations and type definitions, and static and dynamic scopes are
1027used for variants references to tag fields (with relative and absolute
1028path lookups) and for sequence references to length fields.
1029
10307.3.1 Lexical Scope
1031
1032Each of "trace", "env", "stream", "event", "struct" and "variant" have
1033their own nestable declaration scope, within which types can be declared
1034using "typedef" and "typealias". A root declaration scope also contains
1035all declarations located outside of any of the aforementioned
1036declarations. An inner declaration scope can refer to type declared
1037within its container lexical scope prior to the inner declaration scope.
1038Redefinition of a typedef or typealias is not valid, although hiding an
1039upper scope typedef or typealias is allowed within a sub-scope.
1040
10417.3.2 Static and Dynamic Scopes
1042
1043A local static scope consists in the scope generated by the declaration
1044of fields within a compound type. A static scope is a local static scope
1045augmented with the nested sub-static-scopes it contains.
1046
1047A dynamic scope consists in the static scope augmented with the
1048implicit event structure definition hierarchy presented at Section 6.
1049
1050Multiple declarations of the same field name within a local static scope
1051is not valid. It is however valid to re-use the same field name in
1052different local scopes.
1053
1054Nested static and dynamic scopes form lookup paths. These are used for
1055variant tag and sequence length references. They are used at the variant
1056and sequence definition site to look up the location of the tag field
1057associated with a variant, and to lookup up the location of the length
1058field associated with a sequence.
1059
1060Variants and sequences can refer to a tag field either using a relative
1061path or an absolute path. The relative path is relative to the scope in
1062which the variant or sequence performing the lookup is located.
1063Relative paths are only allowed to lookup within the same static scope,
1064which includes its nested static scopes. Lookups targeting parent static
1065scopes need to be performed with an absolute path.
1066
1067Absolute path lookups use the full path including the dynamic scope
1068followed by a "." and then the static scope. Therefore, variants (or
1069sequences) in lower levels in the dynamic scope (e.g. event context) can
1070refer to a tag (or length) field located in upper levels (e.g. in the
1071event header) by specifying, in this case, the associated tag with
1072<stream.event.header.field_name>. This allows, for instance, the event
1073context to define a variant referring to the "id" field of the event
1074header as selector.
1075
1076The dynamic scope prefixes are thus:
1077
1078 - Trace Environment: <env. >,
1079 - Trace Packet Header: <trace.packet.header. >,
1080 - Stream Packet Context: <stream.packet.context. >,
1081 - Event Header: <stream.event.header. >,
1082 - Stream Event Context: <stream.event.context. >,
1083 - Event Context: <event.context. >,
1084 - Event Payload: <event.fields. >.
1085
1086
1087The target dynamic scope must be specified explicitly when referring to
1088a field outside of the static scope (absolute scope reference). No
1089conflict can occur between relative and dynamic paths, because the
1090keywords "trace", "stream", and "event" are reserved, and thus
1091not permitted as field names. It is recommended that field names
1092clashing with CTF and C99 reserved keywords use an underscore prefix to
1093eliminate the risk of generating a description containing an invalid
1094field name. Consequently, fields starting with an underscore should have
1095their leading underscore removed by the CTF trace readers.
1096
1097
1098The information available in the dynamic scopes can be thought of as the
1099current tracing context. At trace production, information about the
1100current context is saved into the specified scope field levels. At trace
1101consumption, for each event, the current trace context is therefore
1102readable by accessing the upper dynamic scopes.
1103
1104
11057.4 TSDL Examples
1106
1107The grammar representing the TSDL meta-data is presented in Appendix C.
1108TSDL Grammar. This section presents a rather lighter reading that
1109consists in examples of TSDL meta-data, with template values.
1110
1111The stream "id" can be left out if there is only one stream in the
1112trace. The event "id" field can be left out if there is only one event
1113in a stream.
1114
1115trace {
1116  major = value;			/* CTF spec version major number */
1117  minor = value;			/* CTF spec version minor number */
1118  uuid = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa";	/* Trace UUID */
1119  byte_order = be OR le;			/* Endianness (required) */
1120  packet.header := struct {
1121    uint32_t magic;
1122    uint8_t  uuid[16];
1123    uint32_t stream_id;
1124  };
1125};
1126
1127/*
1128 * The "env" (environment) scope contains assignment expressions. The
1129 * field names and content are implementation-defined.
1130 */
1131env {
1132  pid = value;			/* example */
1133  proc_name = "name";		/* example */
1134  ...
1135};
1136
1137stream {
1138  id = stream_id;
1139  /* Type 1 - Few event IDs; Type 2 - Many event IDs. See section 6.1. */
1140  event.header := event_header_1 OR event_header_2;
1141  event.context := struct {
1142    ...
1143  };
1144  packet.context := struct {
1145    ...
1146  };
1147};
1148
1149event {
1150  name = "event_name";
1151  id = value;			/* Numeric identifier within the stream */
1152  stream_id = stream_id;
1153  loglevel = value;
1154  context := struct {
1155    ...
1156  };
1157  fields := struct {
1158    ...
1159  };
1160};
1161
1162/* More detail on types in section 4. Types */
1163
1164/*
1165 * Named types:
1166 *
1167 * Type declarations behave similarly to the C standard.
1168 */
1169
1170typedef aliased_type_specifiers new_type_declarators;
1171
1172/* e.g.: typedef struct example new_type_name[10]; */
1173
1174/*
1175 * typealias
1176 *
1177 * The "typealias" declaration can be used to give a name (including
1178 * pointer declarator specifier) to a type. It should also be used to
1179 * map basic C types (float, int, unsigned long, ...) to a CTF type.
1180 * Typealias is a superset of "typedef": it also allows assignment of a
1181 * simple variable identifier to a type.
1182 */
1183
1184typealias type_class {
1185  ...
1186} := type_specifiers type_declarator;
1187
1188/*
1189 * e.g.: 
1190 * typealias integer {
1191 *   size = 32;
1192 *   align = 32;
1193 *   signed = false;
1194 * } := struct page *;
1195 *
1196 * typealias integer {
1197 *  size = 32;
1198 *  align = 32;
1199 *  signed = true;
1200 * } := int;
1201 */
1202
1203struct name {
1204  ...
1205};
1206
1207variant name {
1208  ...
1209};
1210
1211enum name : integer_type {
1212  ...
1213};
1214
1215
1216/*
1217 * Unnamed types, contained within compound type fields, typedef or typealias.
1218 */
1219
1220struct {
1221  ...
1222}
1223
1224struct {
1225  ...
1226} align(value)
1227
1228variant {
1229  ...
1230}
1231
1232enum : integer_type {
1233  ...
1234}
1235
1236typedef type new_type[length];
1237
1238struct {
1239  type field_name[length];
1240}
1241
1242typedef type new_type[length_type];
1243
1244struct {
1245  type field_name[length_type];
1246}
1247
1248integer {
1249  ...
1250}
1251
1252floating_point {
1253  ...
1254}
1255
1256struct {
1257  integer_type field_name:size;		/* GNU/C bitfield */
1258}
1259
1260struct {
1261  string field_name;
1262}
1263
1264
12658. Clocks
1266
1267Clock metadata allows to describe the clock topology of the system, as
1268well as to detail each clock parameter. In absence of clock description,
1269it is assumed that all fields named "timestamp" use the same clock
1270source, which increments once per nanosecond.
1271
1272Describing a clock and how it is used by streams is threefold: first,
1273the clock and clock topology should be described in a "clock"
1274description block, e.g.:
1275
1276clock {
1277	name = cycle_counter_sync;
1278	uuid = "62189bee-96dc-11e0-91a8-cfa3d89f3923";
1279	description = "Cycle counter synchronized across CPUs";
1280	freq = 1000000000;	       /* frequency, in Hz */
1281	/* precision in seconds is: 1000 * (1/freq) */
1282	precision = 1000;
1283	/*
1284	 * clock value offset from Epoch is:
1285	 * offset_s + (offset * (1/freq))
1286	 */
1287	offset_s = 1326476837;
1288	offset = 897235420;
1289	absolute = FALSE;
1290};
1291
1292The mandatory "name" field specifies the name of the clock identifier,
1293which can later be used as a reference. The optional field "uuid" is the
1294unique identifier of the clock. It can be used to correlate different
1295traces that use the same clock. An optional textual description string
1296can be added with the "description" field. The "freq" field is the
1297initial frequency of the clock, in Hz. If the "freq" field is not
1298present, the frequency is assumed to be 1000000000 (providing clock
1299increment of 1 ns). The optional "precision" field details the
1300uncertainty on the clock measurements, in (1/freq) units. The "offset_s"
1301and "offset" fields indicate the offset from POSIX.1 Epoch, 1970-01-01
130200:00:00 +0000 (UTC), to the zero of value of the clock. The "offset_s"
1303field is in seconds. The "offset" field is in (1/freq) units. If any of
1304the "offset_s" or "offset" field is not present, it is assigned the 0
1305value. The field "absolute" is TRUE if the clock is a global reference
1306across different clock uuid (e.g. NTP time). Otherwise, "absolute" is
1307FALSE, and the clock can be considered as synchronized only with other
1308clocks that have the same uuid.
1309
1310
1311Secondly, a reference to this clock should be added within an integer
1312type:
1313
1314typealias integer {
1315	size = 64; align = 1; signed = false;
1316	map = clock.cycle_counter_sync.value;
1317} := uint64_ccnt_t;
1318
1319Thirdly, stream declarations can reference the clock they use as a
1320time-stamp source:
1321
1322struct packet_context {
1323	uint64_ccnt_t ccnt_begin;
1324	uint64_ccnt_t ccnt_end;
1325	/* ... */
1326};
1327
1328stream {
1329	/* ... */
1330	event.header := struct {
1331		uint64_ccnt_t timestamp;
1332		/* ... */
1333	}
1334	packet.context := struct packet_context;
1335};
1336
1337For a N-bit integer type referring to a clock, if the integer overflows
1338compared to the N low order bits of the clock prior value, then it is
1339assumed that one, and only one, overflow occurred. It is therefore
1340important that events encoding time on a small number of bits happen
1341frequently enough to detect when more than one N-bit overflow occurs.
1342
1343In a packet context, clock field names ending with "_begin" and "_end"
1344have a special meaning: this refers to the time-stamps at, respectively,
1345the beginning and the end of each packet.
1346
1347
1348A. Helper macros
1349
1350The two following macros keep track of the size of a GNU/C structure without
1351padding at the end by placing HEADER_END as the last field. A one byte end field
1352is used for C90 compatibility (C99 flexible arrays could be used here). Note
1353that this does not affect the effective structure size, which should always be
1354calculated with the header_sizeof() helper.
1355
1356#define HEADER_END		char end_field
1357#define header_sizeof(type)	offsetof(typeof(type), end_field)
1358
1359
1360B. Stream Header Rationale
1361
1362An event stream is divided in contiguous event packets of variable size. These
1363subdivisions allow the trace analyzer to perform a fast binary search by time
1364within the stream (typically requiring to index only the event packet headers)
1365without reading the whole stream. These subdivisions have a variable size to
1366eliminate the need to transfer the event packet padding when partially filled
1367event packets must be sent when streaming a trace for live viewing/analysis.
1368An event packet can contain a certain amount of padding at the end. Dividing
1369streams into event packets is also useful for network streaming over UDP and
1370flight recorder mode tracing (a whole event packet can be swapped out of the
1371buffer atomically for reading).
1372
1373The stream header is repeated at the beginning of each event packet to allow
1374flexibility in terms of:
1375
1376  - streaming support,
1377  - allowing arbitrary buffers to be discarded without making the trace
1378    unreadable,
1379  - allow UDP packet loss handling by either dealing with missing event packet
1380    or asking for re-transmission.
1381  - transparently support flight recorder mode,
1382  - transparently support crash dump.
1383
1384
1385C. TSDL Grammar
1386
1387/*
1388 * Common Trace Format (CTF) Trace Stream Description Language (TSDL) Grammar.
1389 *
1390 * Inspired from the C99 grammar:
1391 * http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf (Annex A)
1392 * and c++1x grammar (draft)
1393 * http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3291.pdf (Annex A)
1394 *
1395 * Specialized for CTF needs by including only constant and declarations from
1396 * C99 (excluding function declarations), and by adding support for variants,
1397 * sequences and CTF-specific specifiers. Enumeration container types
1398 * semantic is inspired from c++1x enum-base.
1399 */
1400
14011) Lexical grammar
1402
14031.1) Lexical elements
1404
1405token:
1406	keyword
1407	identifier
1408	constant
1409	string-literal
1410	punctuator
1411
14121.2) Keywords
1413
1414keyword: is one of
1415
1416align
1417const
1418char
1419clock
1420double
1421enum
1422env
1423event
1424floating_point
1425float
1426integer
1427int
1428long
1429short
1430signed
1431stream
1432string
1433struct
1434trace
1435typealias
1436typedef
1437unsigned
1438variant
1439void
1440_Bool
1441_Complex
1442_Imaginary
1443
1444
14451.3) Identifiers
1446
1447identifier:
1448	identifier-nondigit
1449	identifier identifier-nondigit
1450	identifier digit
1451
1452identifier-nondigit:
1453	nondigit
1454	universal-character-name
1455	any other implementation-defined characters
1456
1457nondigit:
1458	_
1459	[a-zA-Z]	/* regular expression */
1460
1461digit:
1462	[0-9]		/* regular expression */
1463
14641.4) Universal character names
1465
1466universal-character-name:
1467	\u hex-quad
1468	\U hex-quad hex-quad
1469
1470hex-quad:
1471	hexadecimal-digit hexadecimal-digit hexadecimal-digit hexadecimal-digit
1472
14731.5) Constants
1474
1475constant:
1476	integer-constant
1477	enumeration-constant
1478	character-constant
1479
1480integer-constant:
1481	decimal-constant integer-suffix-opt
1482	octal-constant integer-suffix-opt
1483	hexadecimal-constant integer-suffix-opt
1484
1485decimal-constant:
1486	nonzero-digit
1487	decimal-constant digit
1488
1489octal-constant:
1490	0
1491	octal-constant octal-digit
1492
1493hexadecimal-constant:
1494	hexadecimal-prefix hexadecimal-digit
1495	hexadecimal-constant hexadecimal-digit
1496
1497hexadecimal-prefix:
1498	0x
1499	0X
1500
1501nonzero-digit:
1502	[1-9]
1503
1504integer-suffix:
1505	unsigned-suffix long-suffix-opt
1506	unsigned-suffix long-long-suffix
1507	long-suffix unsigned-suffix-opt
1508	long-long-suffix unsigned-suffix-opt
1509
1510unsigned-suffix:
1511	u
1512	U
1513
1514long-suffix:
1515	l
1516	L
1517
1518long-long-suffix:
1519	ll
1520	LL
1521
1522enumeration-constant:
1523	identifier
1524	string-literal
1525
1526character-constant:
1527	' c-char-sequence '
1528	L' c-char-sequence '
1529
1530c-char-sequence:
1531	c-char
1532	c-char-sequence c-char
1533
1534c-char:
1535	any member of source charset except single-quote ('), backslash
1536	(\), or new-line character.
1537	escape-sequence
1538
1539escape-sequence:
1540	simple-escape-sequence
1541	octal-escape-sequence
1542	hexadecimal-escape-sequence
1543	universal-character-name
1544
1545simple-escape-sequence: one of
1546	\' \" \? \\ \a \b \f \n \r \t \v
1547
1548octal-escape-sequence:
1549	\ octal-digit
1550	\ octal-digit octal-digit
1551	\ octal-digit octal-digit octal-digit
1552
1553hexadecimal-escape-sequence:
1554	\x hexadecimal-digit
1555	hexadecimal-escape-sequence hexadecimal-digit
1556
15571.6) String literals
1558
1559string-literal:
1560	" s-char-sequence-opt "
1561	L" s-char-sequence-opt "
1562
1563s-char-sequence:
1564	s-char
1565	s-char-sequence s-char
1566
1567s-char:
1568	any member of source charset except double-quote ("), backslash
1569	(\), or new-line character.
1570	escape-sequence
1571
15721.7) Punctuators
1573
1574punctuator: one of
1575	[ ] ( ) { } . -> * + - < > : ; ... = ,
1576
1577
15782) Phrase structure grammar
1579
1580primary-expression:
1581	identifier
1582	constant
1583	string-literal
1584	( unary-expression )
1585
1586postfix-expression:
1587	primary-expression
1588	postfix-expression [ unary-expression ]
1589	postfix-expression . identifier
1590	postfix-expressoin -> identifier
1591
1592unary-expression:
1593	postfix-expression
1594	unary-operator postfix-expression
1595
1596unary-operator: one of
1597	+ -
1598
1599assignment-operator:
1600	=
1601
1602type-assignment-operator:
1603	:=
1604
1605constant-expression-range:
1606	unary-expression ... unary-expression
1607
16082.2) Declarations:
1609
1610declaration:
1611	declaration-specifiers declarator-list-opt ;
1612	ctf-specifier ;
1613
1614declaration-specifiers:
1615	storage-class-specifier declaration-specifiers-opt
1616	type-specifier declaration-specifiers-opt
1617	type-qualifier declaration-specifiers-opt
1618
1619declarator-list:
1620	declarator
1621	declarator-list , declarator
1622
1623abstract-declarator-list:
1624	abstract-declarator
1625	abstract-declarator-list , abstract-declarator
1626
1627storage-class-specifier:
1628	typedef
1629
1630type-specifier:
1631	void
1632	char
1633	short
1634	int
1635	long
1636	float
1637	double
1638	signed
1639	unsigned
1640	_Bool
1641	_Complex
1642	_Imaginary
1643	struct-specifier
1644	variant-specifier
1645	enum-specifier
1646	typedef-name
1647	ctf-type-specifier
1648
1649align-attribute:
1650	align ( unary-expression )
1651
1652struct-specifier:
1653	struct identifier-opt { struct-or-variant-declaration-list-opt } align-attribute-opt
1654	struct identifier align-attribute-opt
1655
1656struct-or-variant-declaration-list:
1657	struct-or-variant-declaration
1658	struct-or-variant-declaration-list struct-or-variant-declaration
1659
1660struct-or-variant-declaration:
1661	specifier-qualifier-list struct-or-variant-declarator-list ;
1662	declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list ;
1663	typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list ;
1664	typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list ;
1665
1666specifier-qualifier-list:
1667	type-specifier specifier-qualifier-list-opt
1668	type-qualifier specifier-qualifier-list-opt
1669
1670struct-or-variant-declarator-list:
1671	struct-or-variant-declarator
1672	struct-or-variant-declarator-list , struct-or-variant-declarator
1673
1674struct-or-variant-declarator:
1675	declarator
1676	declarator-opt : unary-expression
1677
1678variant-specifier:
1679	variant identifier-opt variant-tag-opt { struct-or-variant-declaration-list }
1680	variant identifier variant-tag
1681
1682variant-tag:
1683	< unary-expression >
1684
1685enum-specifier:
1686	enum identifier-opt { enumerator-list }
1687	enum identifier-opt { enumerator-list , }
1688	enum identifier
1689	enum identifier-opt : declaration-specifiers { enumerator-list }
1690	enum identifier-opt : declaration-specifiers { enumerator-list , }
1691
1692enumerator-list:
1693	enumerator
1694	enumerator-list , enumerator
1695
1696enumerator:
1697	enumeration-constant
1698	enumeration-constant assignment-operator unary-expression
1699	enumeration-constant assignment-operator constant-expression-range
1700
1701type-qualifier:
1702	const
1703
1704declarator:
1705	pointer-opt direct-declarator
1706
1707direct-declarator:
1708	identifier
1709	( declarator )
1710	direct-declarator [ unary-expression ]
1711
1712abstract-declarator:
1713	pointer-opt direct-abstract-declarator
1714
1715direct-abstract-declarator:
1716	identifier-opt
1717	( abstract-declarator )
1718	direct-abstract-declarator [ unary-expression ]
1719	direct-abstract-declarator [ ]
1720
1721pointer:
1722	* type-qualifier-list-opt
1723	* type-qualifier-list-opt pointer
1724
1725type-qualifier-list:
1726	type-qualifier
1727	type-qualifier-list type-qualifier
1728
1729typedef-name:
1730	identifier
1731
17322.3) CTF-specific declarations
1733
1734ctf-specifier:
1735	clock { ctf-assignment-expression-list-opt }
1736	event { ctf-assignment-expression-list-opt }
1737	stream { ctf-assignment-expression-list-opt }
1738	env { ctf-assignment-expression-list-opt }
1739	trace { ctf-assignment-expression-list-opt }
1740	typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1741	typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list
1742
1743ctf-type-specifier:
1744	floating_point { ctf-assignment-expression-list-opt }
1745	integer { ctf-assignment-expression-list-opt }
1746	string { ctf-assignment-expression-list-opt }
1747	string
1748
1749ctf-assignment-expression-list:
1750	ctf-assignment-expression ;
1751	ctf-assignment-expression-list ctf-assignment-expression ;
1752
1753ctf-assignment-expression:
1754	unary-expression assignment-operator unary-expression
1755	unary-expression type-assignment-operator type-specifier
1756	declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list
1757	typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1758	typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list