Clarify alignment requirements
[ctf.git] / common-trace-format-specification.txt
CommitLineData
f5cb0e57 1Common Trace Format (CTF) Specification (v1.8.2)
5ba9f198
MD
2
3Mathieu Desnoyers, EfficiOS Inc.
4
339a7dde 5The goal of the present document is to specify a trace format that suits the
cc089c3a 6needs of the embedded, telecom, high-performance and kernel communities. It is
5ba9f198 7based on the Common Trace Format Requirements (v1.4) document. It is designed to
cc089c3a 8allow traces to be natively generated by the Linux kernel, Linux user-space
6c7226e9
MD
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.
cc089c3a
MD
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
5ba9f198
MD
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
339a7dde
MD
25The CE Workgroup of the Linux Foundation, Ericsson, and EfficiOS have
26sponsored this work.
27
5ba9f198 28
beabf088
MD
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
37ab95c3 66 7.3.2 Static and Dynamic Scopes
beabf088 67 7.4 TSDL Examples
2fa70eba 688. Clocks
beabf088
MD
69
70
5ba9f198
MD
711. Preliminary definitions
72
3bf79539
MD
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.
5ba9f198
MD
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
3bf79539
MD
80 an event stream.
81 e.g. event: irq_entry.
5ba9f198
MD
82 - An event (or event record) relates to a specific instance of an event
83 class.
3bf79539
MD
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.
5ba9f198
MD
87
88
892. High-level representation of a trace
90
3bf79539
MD
91A trace is divided into multiple event streams. Each event stream contains a
92subset of the trace event types.
5ba9f198 93
3bf79539
MD
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
32254806
MD
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.
5ba9f198 100
6672e9e1
MD
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:
5ba9f198
MD
104
105- Trace version.
106- Types available.
6672e9e1 107- Per-trace event header description.
3bf79539 108- Per-stream event header description.
6672e9e1 109- Per-stream event context description.
5ba9f198 110- Per-event
3bf79539 111 - Event type to stream mapping.
5ba9f198
MD
112 - Event type to name mapping.
113 - Event type to ID mapping.
6672e9e1 114 - Event context description.
5ba9f198
MD
115 - Event fields description.
116
117
3bf79539 1183. Event stream
5ba9f198 119
6672e9e1 120An event stream can be divided into contiguous event packets of variable
12c9069f
MD
121size. An event packet can contain a certain amount of padding at the
122end. The stream header is repeated at the beginning of each event
123packet. The rationale for the event stream design choices is explained
124in Appendix B. Stream Header Rationale.
5ba9f198 125
3bf79539
MD
126The event stream header will therefore be referred to as the "event packet
127header" throughout the rest of this document.
5ba9f198
MD
128
129
1304. Types
131
1fad7a85
MD
132Types are organized as type classes. Each type class belong to either of two
133kind of types: basic types or compound types.
134
5ba9f198
MD
1354.1 Basic types
136
1fad7a85
MD
137A basic type is a scalar type, as described in this section. It includes
138integers, GNU/C bitfields, enumerations, and floating point values.
5ba9f198
MD
139
1404.1.1 Type inheritance
141
80fd2569
MD
142Type specifications can be inherited to allow deriving types from a
143type class. For example, see the uint32_t named type derived from the "integer"
144type class below ("Integers" section). Types have a precise binary
145representation in the trace. A type class has methods to read and write these
146types, but must be derived into a type to be usable in an event field.
5ba9f198
MD
147
1484.1.2 Alignment
149
150We define "byte-packed" types as aligned on the byte size, namely 8-bit.
151We define "bit-packed" types as following on the next bit, as defined by the
370eae99 152"Integers" section.
5ba9f198 153
6672e9e1 154Each basic type must specify its alignment, in bits. Examples of
ec4404a7
MD
155possible alignments are: bit-packed (align = 1), byte-packed (align =
1568), or word-aligned (e.g. align = 32 or align = 64). The choice depends
157on the architecture preference and compactness vs performance trade-offs
158of the implementation. Architectures providing fast unaligned write
159byte-packed basic types to save space, aligning each type on byte
160boundaries (8-bit). Architectures with slow unaligned writes align types
161on specific alignment values. If no specific alignment is declared for a
162type, it is assumed to be bit-packed for integers with size not multiple
163of 8 bits and for gcc bitfields. All other basic types are byte-packed
164by default. It is however recommended to always specify the alignment
165explicitly. Alignment values must be power of two. Compound types are
166aligned as specified in their individual specification.
5ba9f198 167
96bf5a3b
MD
168The base offset used for field alignment is the start of the packet
169containing the field. For instance, a field aligned on 32-bit needs to
170be at an offset multiple of 32-bit from the start of the packet that
171contains it.
172
6672e9e1 173TSDL meta-data attribute representation of a specific alignment:
5ba9f198
MD
174
175 align = value; /* value in bits */
176
1774.1.3 Byte order
178
237324a2
MD
179By default, byte order of a basic type is the byte order described in
180the trace description. It can be overridden by specifying a
181"byte_order" attribute for a basic type. Typical use-case is to specify
182the network byte order (big endian: "be") to save data captured from the
183network into the trace without conversion.
5ba9f198 184
6672e9e1 185TSDL meta-data representation:
5ba9f198
MD
186
187 byte_order = native OR network OR be OR le; /* network and be are aliases */
188
237324a2
MD
189The "native" keyword selects the byte order described in the trace
190description. The "network" byte order is an alias for big endian.
191
7e58b8bf 192Even though the trace description section is not per se a type, for sake
237324a2
MD
193of clarity, it should be noted that "native" and "network" byte orders
194are only allowed within type declaration. The byte_order specified in
195the trace description section only accepts "be" or "le" values.
7e58b8bf 196
5ba9f198
MD
1974.1.4 Size
198
199Type size, in bits, for integers and floats is that returned by "sizeof()" in C
200multiplied by CHAR_BIT.
201We require the size of "char" and "unsigned char" types (CHAR_BIT) to be fixed
202to 8 bits for cross-endianness compatibility.
203
6672e9e1 204TSDL meta-data representation:
5ba9f198
MD
205
206 size = value; (value is in bits)
207
2084.1.5 Integers
209
6672e9e1
MD
210Signed integers are represented in two-complement. Integer alignment,
211size, signedness and byte ordering are defined in the TSDL meta-data.
212Integers aligned on byte size (8-bit) and with length multiple of byte
213size (8-bit) correspond to the C99 standard integers. In addition,
214integers with alignment and/or size that are _not_ a multiple of the
215byte size are permitted; these correspond to the C99 standard bitfields,
216with the added specification that the CTF integer bitfields have a fixed
217binary representation. A MIT-licensed reference implementation of the
218CTF portable bitfields is available at:
5ba9f198
MD
219
220 http://git.efficios.com/?p=babeltrace.git;a=blob;f=include/babeltrace/bitfield.h
221
222Binary representation of integers:
223
224- On little and big endian:
225 - Within a byte, high bits correspond to an integer high bits, and low bits
226 correspond to low bits.
227- On little endian:
228 - Integer across multiple bytes are placed from the less significant to the
229 most significant.
230 - Consecutive integers are placed from lower bits to higher bits (even within
231 a byte).
232- On big endian:
233 - Integer across multiple bytes are placed from the most significant to the
234 less significant.
235 - Consecutive integers are placed from higher bits to lower bits (even within
236 a byte).
237
238This binary representation is derived from the bitfield implementation in GCC
239for little and big endian. However, contrary to what GCC does, integers can
6672e9e1 240cross units boundaries (no padding is required). Padding can be explicitly
5ba9f198
MD
241added (see 4.1.6 GNU/C bitfields) to follow the GCC layout if needed.
242
6672e9e1 243TSDL meta-data representation:
5ba9f198 244
80fd2569 245 integer {
5ba9f198
MD
246 signed = true OR false; /* default false */
247 byte_order = native OR network OR be OR le; /* default native */
248 size = value; /* value in bits, no default */
249 align = value; /* value in bits */
56b77c06
MD
250 /* based used for pretty-printing output, default: decimal. */
251 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
252 OR octal OR oct OR o OR 8 OR binary OR b OR 2;
13441bf5
MD
253 /* character encoding, default: none */
254 encoding = none or UTF8 or ASCII;
2152348f 255 }
5ba9f198 256
80fd2569 257Example of type inheritance (creation of a uint32_t named type):
5ba9f198 258
359894ac 259typealias integer {
9e4e34e9 260 size = 32;
5ba9f198
MD
261 signed = false;
262 align = 32;
38b8da21 263} := uint32_t;
5ba9f198 264
80fd2569 265Definition of a named 5-bit signed bitfield:
5ba9f198 266
359894ac 267typealias integer {
5ba9f198
MD
268 size = 5;
269 signed = true;
270 align = 1;
38b8da21 271} := int5_t;
5ba9f198 272
13441bf5
MD
273The character encoding field can be used to specify that the integer
274must be printed as a text character when read. e.g.:
275
276typealias integer {
277 size = 8;
278 align = 8;
279 signed = false;
280 encoding = UTF8;
281} := utf_char;
282
283
5ba9f198
MD
2844.1.6 GNU/C bitfields
285
286The GNU/C bitfields follow closely the integer representation, with a
287particularity on alignment: if a bitfield cannot fit in the current unit, the
80fd2569
MD
288unit is padded and the bitfield starts at the following unit. The unit size is
289defined by the size of the type "unit_type".
5ba9f198 290
6672e9e1 291TSDL meta-data representation:
80fd2569 292
d674f4b8 293 unit_type name:size;
80fd2569 294
5ba9f198
MD
295As an example, the following structure declared in C compiled by GCC:
296
297struct example {
298 short a:12;
299 short b:5;
300};
301
2152348f
MD
302The example structure is aligned on the largest element (short). The second
303bitfield would be aligned on the next unit boundary, because it would not fit in
304the current unit.
5ba9f198
MD
305
3064.1.7 Floating point
307
6672e9e1 308The floating point values byte ordering is defined in the TSDL meta-data.
5ba9f198
MD
309
310Floating point values follow the IEEE 754-2008 standard interchange formats.
311Description of the floating point values include the exponent and mantissa size
312in bits. Some requirements are imposed on the floating point values:
313
314- FLT_RADIX must be 2.
315- mant_dig is the number of digits represented in the mantissa. It is specified
316 by the ISO C99 standard, section 5.2.4, as FLT_MANT_DIG, DBL_MANT_DIG and
317 LDBL_MANT_DIG as defined by <float.h>.
318- exp_dig is the number of digits represented in the exponent. Given that
319 mant_dig is one bit more than its actual size in bits (leading 1 is not
320 needed) and also given that the sign bit always takes one bit, exp_dig can be
321 specified as:
322
323 - sizeof(float) * CHAR_BIT - FLT_MANT_DIG
324 - sizeof(double) * CHAR_BIT - DBL_MANT_DIG
325 - sizeof(long double) * CHAR_BIT - LDBL_MANT_DIG
326
6672e9e1 327TSDL meta-data representation:
5ba9f198 328
80fd2569 329floating_point {
ec4404a7
MD
330 exp_dig = value;
331 mant_dig = value;
332 byte_order = native OR network OR be OR le;
333 align = value;
2152348f 334}
5ba9f198
MD
335
336Example of type inheritance:
337
359894ac 338typealias floating_point {
5ba9f198
MD
339 exp_dig = 8; /* sizeof(float) * CHAR_BIT - FLT_MANT_DIG */
340 mant_dig = 24; /* FLT_MANT_DIG */
341 byte_order = native;
ec4404a7 342 align = 32;
38b8da21 343} := float;
5ba9f198
MD
344
345TODO: define NaN, +inf, -inf behavior.
346
ec4404a7
MD
347Bit-packed, byte-packed or larger alignments can be used for floating
348point values, similarly to integers.
349
5ba9f198
MD
3504.1.8 Enumerations
351
352Enumerations are a mapping between an integer type and a table of strings. The
353numerical representation of the enumeration follows the integer type specified
6672e9e1
MD
354by the meta-data. The enumeration mapping table is detailed in the enumeration
355description within the meta-data. The mapping table maps inclusive value
356ranges (or single values) to strings. Instead of being limited to simple
3bf79539 357"value -> string" mappings, these enumerations map
80fd2569 358"[ start_value ... end_value ] -> string", which map inclusive ranges of
3bf79539
MD
359values to strings. An enumeration from the C language can be represented in
360this format by having the same start_value and end_value for each element, which
361is in fact a range of size 1. This single-value range is supported without
4767a9e7 362repeating the start and end values with the value = string declaration.
80fd2569 363
a9b83695 364enum name : integer_type {
359894ac 365 somestring = start_value1 ... end_value1,
80fd2569
MD
366 "other string" = start_value2 ... end_value2,
367 yet_another_string, /* will be assigned to end_value2 + 1 */
368 "some other string" = value,
369 ...
370};
371
372If the values are omitted, the enumeration starts at 0 and increment of 1 for
0cedf6e5
MD
373each entry. An entry with omitted value that follows a range entry takes
374as value the end_value of the previous range + 1:
80fd2569 375
a9b83695 376enum name : unsigned int {
80fd2569
MD
377 ZERO,
378 ONE,
379 TWO,
380 TEN = 10,
381 ELEVEN,
3bf79539 382};
5ba9f198 383
80fd2569 384Overlapping ranges within a single enumeration are implementation defined.
5ba9f198 385
2152348f
MD
386A nameless enumeration can be declared as a field type or as part of a typedef:
387
a9b83695 388enum : integer_type {
2152348f
MD
389 ...
390}
391
c2742c56
MD
392Enumerations omitting the container type ": integer_type" use the "int"
393type (for compatibility with C99). The "int" type must be previously
394declared. E.g.:
395
82cbdb16 396typealias integer { size = 32; align = 32; signed = true; } := int;
c2742c56
MD
397
398enum {
399 ...
400}
401
1fad7a85 402
5ba9f198
MD
4034.2 Compound types
404
1fad7a85
MD
405Compound are aggregation of type declarations. Compound types include
406structures, variant, arrays, sequences, and strings.
407
5ba9f198
MD
4084.2.1 Structures
409
410Structures are aligned on the largest alignment required by basic types
411contained within the structure. (This follows the ISO/C standard for structures)
412
6672e9e1 413TSDL meta-data representation of a named structure:
5ba9f198 414
80fd2569
MD
415struct name {
416 field_type field_name;
417 field_type field_name;
418 ...
419};
5ba9f198
MD
420
421Example:
422
80fd2569
MD
423struct example {
424 integer { /* Nameless type */
425 size = 16;
426 signed = true;
427 align = 16;
428 } first_field_name;
6672e9e1 429 uint64_t second_field_name; /* Named type declared in the meta-data */
3bf79539 430};
5ba9f198 431
4cac83ee
MD
432The fields are placed in a sequence next to each other. They each
433possess a field name, which is a unique identifier within the structure.
434The identifier is not allowed to use any reserved keyword
435(see Section C.1.2). Replacing reserved keywords with
70375f92 436underscore-prefixed field names is recommended. Fields starting with an
92250c71
MD
437underscore should have their leading underscore removed by the CTF trace
438readers.
5ba9f198 439
2152348f 440A nameless structure can be declared as a field type or as part of a typedef:
80fd2569
MD
441
442struct {
443 ...
2152348f 444}
80fd2569 445
ec4404a7
MD
446Alignment for a structure compound type can be forced to a minimum value
447by adding an "align" specifier after the declaration of a structure
448body. This attribute is read as: align(value). The value is specified in
449bits. The structure will be aligned on the maximum value between this
450attribute and the alignment required by the basic types contained within
451the structure. e.g.
452
453struct {
454 ...
455} align(32)
456
77a98c82 4574.2.2 Variants (Discriminated/Tagged Unions)
fcba70d4 458
fdf2bb05
MD
459A CTF variant is a selection between different types. A CTF variant must
460always be defined within the scope of a structure or within fields
461contained within a structure (defined recursively). A "tag" enumeration
37ab95c3
MD
462field must appear in either the same static scope, prior to the variant
463field (in field declaration order), in an upper static scope , or in an
464upper dynamic scope (see Section 7.3.2). The type selection is indicated
465by the mapping from the enumeration value to the string used as variant
466type selector. The field to use as tag is specified by the "tag_field",
467specified between "< >" after the "variant" keyword for unnamed
468variants, and after "variant name" for named variants.
fcba70d4 469
0bed08fc
MD
470The alignment of the variant is the alignment of the type as selected by
471the tag value for the specific instance of the variant. The size of the
472variant is the size as selected by the tag value for the specific
473instance of the variant.
474
475The alignment of the type containing the variant is independent of the
476variant alignment. For instance, if a structure contains two fields, a
47732-bit integer, aligned on 32 bits, and a variant, which contains two
478choices: either a 32-bit field, aligned on 32 bits, or a 64-bit field,
479aligned on 64 bits, the alignment of the outmost structure will be
48032-bit (the alignment of its largest field, disregarding the alignment
481of the variant). The alignment of the variant will depend on the
482selector: if the variant's 32-bit field is selected, its alignment will
483be 32-bit, or 64-bit otherwise. It is important to note that variants
484are specifically tailored for compactness in a stream. Therefore, the
485relative offsets of compound type fields can vary depending on
486the offset at which the compound type starts if it contains a variant
487that itself contains a type with alignment larger than the largest field
488contained within the compound type. This is caused by the fact that the
489compound type may contain the enumeration that select the variant's
490choice, and therefore the alignment to be applied to the compound type
491cannot be determined before encountering the enumeration.
fcba70d4 492
4cac83ee
MD
493Each variant type selector possess a field name, which is a unique
494identifier within the variant. The identifier is not allowed to use any
495reserved keyword (see Section C.1.2). Replacing reserved keywords with
70375f92 496underscore-prefixed field names is recommended. Fields starting with an
92250c71
MD
497underscore should have their leading underscore removed by the CTF trace
498readers.
70375f92 499
4cac83ee 500
fcba70d4
MD
501A named variant declaration followed by its definition within a structure
502declaration:
503
504variant name {
505 field_type sel1;
506 field_type sel2;
507 field_type sel3;
508 ...
509};
510
511struct {
a9b83695 512 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
fcba70d4
MD
513 ...
514 variant name <tag_field> v;
515}
516
517An unnamed variant definition within a structure is expressed by the following
6672e9e1 518TSDL meta-data:
fcba70d4
MD
519
520struct {
a9b83695 521 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
fcba70d4
MD
522 ...
523 variant <tag_field> {
524 field_type sel1;
525 field_type sel2;
526 field_type sel3;
527 ...
528 } v;
529}
530
531Example of a named variant within a sequence that refers to a single tag field:
532
533variant example {
534 uint32_t a;
535 uint64_t b;
536 short c;
537};
538
539struct {
a9b83695 540 enum : uint2_t { a, b, c } choice;
1ab22b2a
MD
541 unsigned int seqlen;
542 variant example <choice> v[seqlen];
fcba70d4
MD
543}
544
545Example of an unnamed variant:
546
547struct {
a9b83695 548 enum : uint2_t { a, b, c, d } choice;
fcba70d4
MD
549 /* Unrelated fields can be added between the variant and its tag */
550 int32_t somevalue;
551 variant <choice> {
552 uint32_t a;
553 uint64_t b;
554 short c;
555 struct {
556 unsigned int field1;
557 uint64_t field2;
558 } d;
559 } s;
560}
561
562Example of an unnamed variant within an array:
563
564struct {
a9b83695 565 enum : uint2_t { a, b, c } choice;
fcba70d4
MD
566 variant <choice> {
567 uint32_t a;
568 uint64_t b;
569 short c;
15850440 570 } v[10];
fcba70d4
MD
571}
572
573Example of a variant type definition within a structure, where the defined type
574is then declared within an array of structures. This variant refers to a tag
37ab95c3 575located in an upper static scope. This example clearly shows that a variant
fcba70d4 576type definition referring to the tag "x" uses the closest preceding field from
37ab95c3 577the static scope of the type definition.
fcba70d4
MD
578
579struct {
a9b83695 580 enum : uint2_t { a, b, c, d } x;
fcba70d4
MD
581
582 typedef variant <x> { /*
583 * "x" refers to the preceding "x" enumeration in the
37ab95c3 584 * static scope of the type definition.
fcba70d4
MD
585 */
586 uint32_t a;
587 uint64_t b;
588 short c;
589 } example_variant;
590
591 struct {
a9b83695 592 enum : int { x, y, z } x; /* This enumeration is not used by "v". */
fcba70d4 593 example_variant v; /*
a9b83695 594 * "v" uses the "enum : uint2_t { a, b, c, d }"
fcba70d4
MD
595 * tag.
596 */
597 } a[10];
598}
599
6004.2.3 Arrays
5ba9f198 601
6672e9e1
MD
602Arrays are fixed-length. Their length is declared in the type
603declaration within the meta-data. They contain an array of "inner type"
604elements, which can refer to any type not containing the type of the
605array being declared (no circular dependency). The length is the number
606of elements in an array.
5ba9f198 607
6672e9e1 608TSDL meta-data representation of a named array:
80fd2569
MD
609
610typedef elem_type name[length];
5ba9f198 611
2152348f 612A nameless array can be declared as a field type within a structure, e.g.:
5ba9f198 613
2152348f 614 uint8_t field_name[10];
80fd2569 615
ec4404a7 616Arrays are always aligned on their element alignment requirement.
5ba9f198 617
fcba70d4 6184.2.4 Sequences
5ba9f198 619
f3afd1c7 620Sequences are dynamically-sized arrays. They refer to a "length"
37ab95c3 621unsigned integer field, which must appear in either the same static scope,
1ab22b2a 622prior to the sequence field (in field declaration order), in an upper
37ab95c3
MD
623static scope, or in an upper dynamic scope (see Section 7.3.2). This
624length field represents the number of elements in the sequence. The
625sequence per se is an array of "inner type" elements.
5ba9f198 626
1ab22b2a 627TSDL meta-data representation for a sequence type definition:
80fd2569 628
1ab22b2a
MD
629struct {
630 unsigned int length_field;
631 typedef elem_type typename[length_field];
632 typename seq_field_name;
633}
634
635A sequence can also be declared as a field type, e.g.:
80fd2569 636
1ab22b2a
MD
637struct {
638 unsigned int length_field;
639 long seq_field_name[length_field];
640}
80fd2569 641
1ab22b2a
MD
642Multiple sequences can refer to the same length field, and these length
643fields can be in a different upper dynamic scope:
644
645e.g., assuming the stream.event.header defines:
646
647stream {
648 ...
649 id = 1;
650 event.header := struct {
651 uint16_t seq_len;
652 };
653};
654
655event {
656 ...
657 stream_id = 1;
658 fields := struct {
659 long seq_a[stream.event.header.seq_len];
660 char seq_b[stream.event.header.seq_len];
661 };
662};
80fd2569 663
1ab22b2a 664The sequence elements follow the "array" specifications.
5ba9f198 665
fcba70d4 6664.2.5 Strings
5ba9f198
MD
667
668Strings are an array of bytes of variable size and are terminated by a '\0'
6672e9e1
MD
669"NULL" character. Their encoding is described in the TSDL meta-data. In
670absence of encoding attribute information, the default encoding is
671UTF-8.
5ba9f198 672
6672e9e1 673TSDL meta-data representation of a named string type:
80fd2569 674
359894ac 675typealias string {
5ba9f198 676 encoding = UTF8 OR ASCII;
38b8da21 677} := name;
5ba9f198 678
80fd2569
MD
679A nameless string type can be declared as a field type:
680
681string field_name; /* Use default UTF8 encoding */
5ba9f198 682
ec4404a7
MD
683Strings are always aligned on byte size.
684
3bf79539
MD
6855. Event Packet Header
686
6672e9e1
MD
687The event packet header consists of two parts: the "event packet header"
688is the same for all streams of a trace. The second part, the "event
689packet context", is described on a per-stream basis. Both are described
96bf5a3b 690in the TSDL meta-data.
3bf79539 691
6672e9e1 692Event packet header (all fields are optional, specified by TSDL meta-data):
3bf79539 693
6672e9e1
MD
694- Magic number (CTF magic number: 0xC1FC1FC1) specifies that this is a
695 CTF packet. This magic number is optional, but when present, it should
696 come at the very beginning of the packet.
697- Trace UUID, used to ensure the event packet match the meta-data used.
698 (note: we cannot use a meta-data checksum in every cases instead of a
699 UUID because meta-data can be appended to while tracing is active)
700 This field is optional.
701- Stream ID, used as reference to stream description in meta-data.
702 This field is optional if there is only one stream description in the
703 meta-data, but becomes required if there are more than one stream in
704 the TSDL meta-data description.
3bf79539 705
6672e9e1 706Event packet context (all fields are optional, specified by TSDL meta-data):
3bf79539 707
4dfca05b
MD
708- Event packet content size (in bits).
709- Event packet size (in bits, includes padding).
cda89682 710- Event packet content checksum. Checksum excludes the event packet
3bf79539
MD
711 header.
712- Per-stream event packet sequence count (to deal with UDP packet loss). The
713 number of significant sequence counter bits should also be present, so
b11853af 714 wrap-arounds are dealt with correctly.
6672e9e1 715- Time-stamp at the beginning and time-stamp at the end of the event packet.
3bf79539
MD
716 Both timestamps are written in the packet header, but sampled respectively
717 while (or before) writing the first event and while (or after) writing the
718 last event in the packet. The inclusive range between these timestamps should
719 include all event timestamps assigned to events contained within the packet.
5ba9f198 720- Events discarded count
3bf79539 721 - Snapshot of a per-stream free-running counter, counting the number of
8f34dd4f
MD
722 events discarded that were supposed to be written in the stream after
723 the last event in the event packet.
724 * Note: producer-consumer buffer full condition can fill the current
3bf79539 725 event packet with padding so we know exactly where events have been
8f34dd4f
MD
726 discarded. However, if the buffer full condition chooses not
727 to fill the current event packet with padding, all we know
728 about the timestamp range in which the events have been
729 discarded is that it is somewhere between the beginning and
730 the end of the packet.
3bf79539
MD
731- Lossless compression scheme used for the event packet content. Applied
732 directly to raw data. New types of compression can be added in following
733 versions of the format.
5ba9f198
MD
734 0: no compression scheme
735 1: bzip2
736 2: gzip
3bf79539
MD
737 3: xz
738- Cypher used for the event packet content. Applied after compression.
5ba9f198
MD
739 0: no encryption
740 1: AES
3bf79539 741- Checksum scheme used for the event packet content. Applied after encryption.
5ba9f198
MD
742 0: no checksum
743 1: md5
744 2: sha1
745 3: crc32
746
6672e9e1 7475.1 Event Packet Header Description
3bf79539 748
fc5425db 749The event packet header layout is indicated by the trace packet.header
6672e9e1
MD
750field. Here is a recommended structure type for the packet header with
751the fields typically expected (although these fields are each optional):
fc5425db 752
80fd2569
MD
753struct event_packet_header {
754 uint32_t magic;
3fde5da1 755 uint8_t uuid[16];
3bf79539 756 uint32_t stream_id;
80fd2569 757};
5ba9f198 758
fc5425db
MD
759trace {
760 ...
761 packet.header := struct event_packet_header;
762};
763
6672e9e1
MD
764If the magic number is not present, tools such as "file" will have no
765mean to discover the file type.
766
3fde5da1
MD
767If the uuid is not present, no validation that the meta-data actually
768corresponds to the stream is performed.
37b39e47
MD
769
770If the stream_id packet header field is missing, the trace can only
771contain a single stream. Its "id" field can be left out, and its events
772don't need to declare a "stream_id" field.
773
774
3bf79539
MD
7755.2 Event Packet Context Description
776
777Event packet context example. These are declared within the stream declaration
6672e9e1 778in the meta-data. All these fields are optional. If the packet size field is
6a7c61df 779missing, the whole stream only contains a single packet. If the content
f6b81726
MD
780size field is missing, the packet is filled (no padding). The content
781and packet sizes include all headers.
3bf79539
MD
782
783An example event packet context type:
784
80fd2569 785struct event_packet_context {
3bf79539
MD
786 uint64_t timestamp_begin;
787 uint64_t timestamp_end;
788 uint32_t checksum;
789 uint32_t stream_packet_count;
790 uint32_t events_discarded;
791 uint32_t cpu_id;
47911ef2
MD
792 uint64_t/uint32_t/uint16_t content_size;
793 uint64_t/uint32_t/uint16_t packet_size;
3bf79539
MD
794 uint8_t compression_scheme;
795 uint8_t encryption_scheme;
3b0f8e4d 796 uint8_t checksum_scheme;
3bf79539 797};
5ba9f198 798
fcba70d4 799
5ba9f198
MD
8006. Event Structure
801
802The overall structure of an event is:
803
6672e9e1
MD
8041 - Stream Packet Context (as specified by the stream meta-data)
805 2 - Event Header (as specified by the stream meta-data)
806 3 - Stream Event Context (as specified by the stream meta-data)
807 4 - Event Context (as specified by the event meta-data)
808 5 - Event Payload (as specified by the event meta-data)
5ba9f198 809
fdf2bb05 810This structure defines an implicit dynamic scoping, where variants
7d9d7e92
MD
811located in inner structures (those with a higher number in the listing
812above) can refer to the fields of outer structures (with lower number in
6c7226e9 813the listing above). See Section 7.3 TSDL Scopes for more detail.
5ba9f198 814
fdf2bb05 8156.1 Event Header
fcba70d4 816
6672e9e1 817Event headers can be described within the meta-data. We hereby propose, as an
fcba70d4
MD
818example, two types of events headers. Type 1 accommodates streams with less than
81931 event IDs. Type 2 accommodates streams with 31 or more event IDs.
5ba9f198 820
3bf79539
MD
821One major factor can vary between streams: the number of event IDs assigned to
822a stream. Luckily, this information tends to stay relatively constant (modulo
5ba9f198 823event registration while trace is being recorded), so we can specify different
3bf79539 824representations for streams containing few event IDs and streams containing
6672e9e1
MD
825many event IDs, so we end up representing the event ID and time-stamp as
826densely as possible in each case.
5ba9f198 827
fcba70d4
MD
828The header is extended in the rare occasions where the information cannot be
829represented in the ranges available in the standard event header. They are also
3bf79539
MD
830used in the rare occasions where the data required for a field could not be
831collected: the flag corresponding to the missing field within the missing_fields
832array is then set to 1.
5ba9f198 833
6672e9e1
MD
834Types uintX_t represent an X-bit unsigned integer, as declared with
835either:
5ba9f198 836
82cbdb16 837 typealias integer { size = X; align = X; signed = false; } := uintX_t;
6672e9e1
MD
838
839 or
840
82cbdb16 841 typealias integer { size = X; align = 1; signed = false; } := uintX_t;
5ba9f198 842
fdf2bb05 8436.1.1 Type 1 - Few event IDs
5ba9f198
MD
844
845 - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
846 preference).
5ba9f198 847 - Native architecture byte ordering.
fcba70d4
MD
848 - For "compact" selection
849 - Fixed size: 32 bits.
850 - For "extended" selection
851 - Size depends on the architecture and variant alignment.
5ba9f198 852
80fd2569 853struct event_header_1 {
fcba70d4
MD
854 /*
855 * id: range: 0 - 30.
856 * id 31 is reserved to indicate an extended header.
857 */
a9b83695 858 enum : uint5_t { compact = 0 ... 30, extended = 31 } id;
fcba70d4
MD
859 variant <id> {
860 struct {
861 uint27_t timestamp;
862 } compact;
863 struct {
864 uint32_t id; /* 32-bit event IDs */
865 uint64_t timestamp; /* 64-bit timestamps */
866 } extended;
867 } v;
cb108fea 868} align(32); /* or align(8) */
5ba9f198 869
5ba9f198 870
fdf2bb05 8716.1.2 Type 2 - Many event IDs
5ba9f198 872
fcba70d4 873 - Aligned on 16-bit (or 8-bit if byte-packed, depending on the architecture
5ba9f198 874 preference).
5ba9f198 875 - Native architecture byte ordering.
fcba70d4
MD
876 - For "compact" selection
877 - Size depends on the architecture and variant alignment.
878 - For "extended" selection
879 - Size depends on the architecture and variant alignment.
5ba9f198 880
80fd2569 881struct event_header_2 {
fcba70d4
MD
882 /*
883 * id: range: 0 - 65534.
884 * id 65535 is reserved to indicate an extended header.
885 */
a9b83695 886 enum : uint16_t { compact = 0 ... 65534, extended = 65535 } id;
fcba70d4
MD
887 variant <id> {
888 struct {
889 uint32_t timestamp;
890 } compact;
891 struct {
892 uint32_t id; /* 32-bit event IDs */
893 uint64_t timestamp; /* 64-bit timestamps */
894 } extended;
895 } v;
cb108fea 896} align(16); /* or align(8) */
5ba9f198 897
5ba9f198
MD
898
8996.2 Event Context
900
6672e9e1
MD
901The event context contains information relative to the current event.
902The choice and meaning of this information is specified by the TSDL
903stream and event meta-data descriptions. The stream context is applied
904to all events within the stream. The stream context structure follows
905the event header. The event context is applied to specific events. Its
906structure follows the stream context structure.
5ba9f198 907
fcba70d4
MD
908An example of stream-level event context is to save the event payload size with
909each event, or to save the current PID with each event. These are declared
6672e9e1 910within the stream declaration within the meta-data:
5ba9f198 911
fcba70d4
MD
912 stream {
913 ...
6672e9e1 914 event.context := struct {
80fd2569
MD
915 uint pid;
916 uint16_t payload_size;
6672e9e1 917 };
fcba70d4
MD
918 };
919
920An example of event-specific event context is to declare a bitmap of missing
921fields, only appended after the stream event context if the extended event
922header is selected. NR_FIELDS is the number of fields within the event (a
923numeric value).
5ba9f198 924
fcba70d4
MD
925 event {
926 context = struct {
927 variant <id> {
928 struct { } compact;
929 struct {
930 uint1_t missing_fields[NR_FIELDS]; /* missing event fields bitmap */
931 } extended;
932 } v;
933 };
934 ...
935 }
5ba9f198
MD
936
9376.3 Event Payload
938
939An event payload contains fields specific to a given event type. The fields
6672e9e1 940belonging to an event type are described in the event-specific meta-data
5ba9f198
MD
941within a structure type.
942
9436.3.1 Padding
944
945No padding at the end of the event payload. This differs from the ISO/C standard
946for structures, but follows the CTF standard for structures. In a trace, even
947though it makes sense to align the beginning of a structure, it really makes no
948sense to add padding at the end of the structure, because structures are usually
949not followed by a structure of the same type.
950
951This trick can be done by adding a zero-length "end" field at the end of the C
952structures, and by using the offset of this field rather than using sizeof()
3bf79539 953when calculating the size of a structure (see Appendix "A. Helper macros").
5ba9f198
MD
954
9556.3.2 Alignment
956
957The event payload is aligned on the largest alignment required by types
958contained within the payload. (This follows the ISO/C standard for structures)
959
960
6c7226e9
MD
9617. Trace Stream Description Language (TSDL)
962
963The Trace Stream Description Language (TSDL) allows expression of the
964binary trace streams layout in a C99-like Domain Specific Language
965(DSL).
966
967
6672e9e1 9687.1 Meta-data
6c7226e9
MD
969
970The trace stream layout description is located in the trace meta-data.
971The meta-data is itself located in a stream identified by its name:
972"metadata".
5ba9f198 973
6672e9e1
MD
974The meta-data description can be expressed in two different formats:
975text-only and packet-based. The text-only description facilitates
976generation of meta-data and provides a convenient way to enter the
977meta-data information by hand. The packet-based meta-data provides the
978CTF stream packet facilities (checksumming, compression, encryption,
979network-readiness) for meta-data stream generated and transported by a
980tracer.
981
1b4d35eb
MD
982The text-only meta-data file is a plain-text TSDL description. This file
983must begin with the following characters to identify the file as a CTF
9486a18c 984TSDL text-based metadata file (without the double-quotes) :
1b4d35eb 985
ec2b4db8 986"/* CTF"
1b4d35eb 987
ec2b4db8
MD
988It must be followed by a space, and the version of the specification
989followed by the CTF trace, e.g.:
990
991" 1.8"
992
993These characters allow automated discovery of file type and CTF
994specification version. They are interpreted as a the beginning of a
995comment by the TSDL metadata parser. The comment can be continued to
996contain extra commented characters before it is closed.
6672e9e1
MD
997
998The packet-based meta-data is made of "meta-data packets", which each
999start with a meta-data packet header. The packet-based meta-data
1000description is detected by reading the magic number "0x75D11D57" at the
1001beginning of the file. This magic number is also used to detect the
1002endianness of the architecture by trying to read the CTF magic number
1003and its counterpart in reversed endianness. The events within the
1004meta-data stream have no event header nor event context. Each event only
46400bde
MD
1005contains a special "sequence" payload, which is a sequence of bits which
1006length is implicitly calculated by using the
1007"trace.packet.header.content_size" field, minus the packet header size.
1008The formatting of this sequence of bits is a plain-text representation
1009of the TSDL description. Each meta-data packet start with a special
1010packet header, specific to the meta-data stream, which contains,
1011exactly:
6672e9e1
MD
1012
1013struct metadata_packet_header {
2daeaa3a 1014 uint32_t magic; /* 0x75D11D57 */
3fde5da1 1015 uint8_t uuid[16]; /* Unique Universal Identifier */
6672e9e1
MD
1016 uint32_t checksum; /* 0 if unused */
1017 uint32_t content_size; /* in bits */
1018 uint32_t packet_size; /* in bits */
1019 uint8_t compression_scheme; /* 0 if unused */
1020 uint8_t encryption_scheme; /* 0 if unused */
1021 uint8_t checksum_scheme; /* 0 if unused */
ec2b4db8
MD
1022 uint8_t major; /* CTF spec version major number */
1023 uint8_t minor; /* CTF spec version minor number */
6672e9e1
MD
1024};
1025
1026The packet-based meta-data can be converted to a text-only meta-data by
3568031f 1027concatenating all the strings it contains.
4fafe1ad 1028
6672e9e1
MD
1029In the textual representation of the meta-data, the text contained
1030within "/*" and "*/", as well as within "//" and end of line, are
1031treated as comments. Boolean values can be represented as true, TRUE,
1032or 1 for true, and false, FALSE, or 0 for false. Within the string-based
1033meta-data description, the trace UUID is represented as a string of
1034hexadecimal digits and dashes "-". In the event packet header, the trace
1035UUID is represented as an array of bytes.
fcba70d4 1036
fdf2bb05 1037
6c7226e9 10387.2 Declaration vs Definition
fdf2bb05
MD
1039
1040A declaration associates a layout to a type, without specifying where
1041this type is located in the event structure hierarchy (see Section 6).
1042This therefore includes typedef, typealias, as well as all type
1043specifiers. In certain circumstances (typedef, structure field and
1044variant field), a declaration is followed by a declarator, which specify
1045the newly defined type name (for typedef), or the field name (for
1046declarations located within structure and variants). Array and sequence,
1047declared with square brackets ("[" "]"), are part of the declarator,
a9b83695 1048similarly to C99. The enumeration base type is specified by
6c7226e9 1049": enum_base", which is part of the type specifier. The variant tag
a9b83695 1050name, specified between "<" ">", is also part of the type specifier.
fdf2bb05
MD
1051
1052A definition associates a type to a location in the event structure
b9606a77
MD
1053hierarchy (see Section 6). This association is denoted by ":=", as shown
1054in Section 7.3.
fdf2bb05
MD
1055
1056
6c7226e9 10577.3 TSDL Scopes
fdf2bb05 1058
37ab95c3
MD
1059TSDL uses three different types of scoping: a lexical scope is used for
1060declarations and type definitions, and static and dynamic scopes are
1061used for variants references to tag fields (with relative and absolute
1062path lookups) and for sequence references to length fields.
fdf2bb05 1063
6c7226e9 10647.3.1 Lexical Scope
fdf2bb05 1065
570ecabe
MD
1066Each of "trace", "env", "stream", "event", "struct" and "variant" have
1067their own nestable declaration scope, within which types can be declared
1068using "typedef" and "typealias". A root declaration scope also contains
1069all declarations located outside of any of the aforementioned
1070declarations. An inner declaration scope can refer to type declared
1071within its container lexical scope prior to the inner declaration scope.
1072Redefinition of a typedef or typealias is not valid, although hiding an
1073upper scope typedef or typealias is allowed within a sub-scope.
fdf2bb05 1074
37ab95c3 10757.3.2 Static and Dynamic Scopes
fdf2bb05 1076
37ab95c3
MD
1077A local static scope consists in the scope generated by the declaration
1078of fields within a compound type. A static scope is a local static scope
1079augmented with the nested sub-static-scopes it contains.
1080
1081A dynamic scope consists in the static scope augmented with the
7d9d7e92 1082implicit event structure definition hierarchy presented at Section 6.
fdf2bb05 1083
37ab95c3
MD
1084Multiple declarations of the same field name within a local static scope
1085is not valid. It is however valid to re-use the same field name in
1086different local scopes.
1087
1088Nested static and dynamic scopes form lookup paths. These are used for
1089variant tag and sequence length references. They are used at the variant
1090and sequence definition site to look up the location of the tag field
1091associated with a variant, and to lookup up the location of the length
1092field associated with a sequence.
1093
1094Variants and sequences can refer to a tag field either using a relative
4cac83ee
MD
1095path or an absolute path. The relative path is relative to the scope in
1096which the variant or sequence performing the lookup is located.
37ab95c3
MD
1097Relative paths are only allowed to lookup within the same static scope,
1098which includes its nested static scopes. Lookups targeting parent static
1099scopes need to be performed with an absolute path.
1100
1101Absolute path lookups use the full path including the dynamic scope
1102followed by a "." and then the static scope. Therefore, variants (or
1103sequences) in lower levels in the dynamic scope (e.g. event context) can
1104refer to a tag (or length) field located in upper levels (e.g. in the
1105event header) by specifying, in this case, the associated tag with
1106<stream.event.header.field_name>. This allows, for instance, the event
1107context to define a variant referring to the "id" field of the event
1108header as selector.
1109
284724ae 1110The dynamic scope prefixes are thus:
fdf2bb05 1111
570ecabe 1112 - Trace Environment: <env. >,
e0d9e2c7 1113 - Trace Packet Header: <trace.packet.header. >,
7d9d7e92
MD
1114 - Stream Packet Context: <stream.packet.context. >,
1115 - Event Header: <stream.event.header. >,
1116 - Stream Event Context: <stream.event.context. >,
1117 - Event Context: <event.context. >,
1118 - Event Payload: <event.fields. >.
fdf2bb05 1119
37ab95c3
MD
1120
1121The target dynamic scope must be specified explicitly when referring to
4cac83ee
MD
1122a field outside of the static scope (absolute scope reference). No
1123conflict can occur between relative and dynamic paths, because the
1124keywords "trace", "stream", and "event" are reserved, and thus
1125not permitted as field names. It is recommended that field names
1126clashing with CTF and C99 reserved keywords use an underscore prefix to
1127eliminate the risk of generating a description containing an invalid
70375f92 1128field name. Consequently, fields starting with an underscore should have
92250c71 1129their leading underscore removed by the CTF trace readers.
70375f92 1130
fdf2bb05 1131
457d8b0a
MD
1132The information available in the dynamic scopes can be thought of as the
1133current tracing context. At trace production, information about the
1134current context is saved into the specified scope field levels. At trace
1135consumption, for each event, the current trace context is therefore
1136readable by accessing the upper dynamic scopes.
1137
fdf2bb05 1138
6c7226e9 11397.4 TSDL Examples
d285084f 1140
6672e9e1 1141The grammar representing the TSDL meta-data is presented in Appendix C.
7df6b93a 1142TSDL Grammar. This section presents a rather lighter reading that
6672e9e1 1143consists in examples of TSDL meta-data, with template values.
969f30c0 1144
350bb633
MD
1145The stream "id" can be left out if there is only one stream in the
1146trace. The event "id" field can be left out if there is only one event
1147in a stream.
1148
5ba9f198 1149trace {
1e9de2d1
MD
1150 major = value; /* CTF spec version major number */
1151 minor = value; /* CTF spec version minor number */
fdf2bb05 1152 uuid = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"; /* Trace UUID */
58997e9e 1153 byte_order = be OR le; /* Endianness (required) */
fc5425db
MD
1154 packet.header := struct {
1155 uint32_t magic;
3fde5da1 1156 uint8_t uuid[16];
fc5425db
MD
1157 uint32_t stream_id;
1158 };
3bf79539 1159};
5ba9f198 1160
570ecabe
MD
1161/*
1162 * The "env" (environment) scope contains assignment expressions. The
1163 * field names and content are implementation-defined.
1164 */
1165env {
1166 pid = value; /* example */
1167 proc_name = "name"; /* example */
1168 ...
1169};
1170
3bf79539
MD
1171stream {
1172 id = stream_id;
fdf2bb05 1173 /* Type 1 - Few event IDs; Type 2 - Many event IDs. See section 6.1. */
4fa992a5
MD
1174 event.header := event_header_1 OR event_header_2;
1175 event.context := struct {
77a98c82 1176 ...
3bf79539 1177 };
4fa992a5 1178 packet.context := struct {
77a98c82 1179 ...
3bf79539
MD
1180 };
1181};
5ba9f198
MD
1182
1183event {
980015f9 1184 name = "event_name";
3bf79539 1185 id = value; /* Numeric identifier within the stream */
67f02e24 1186 stream_id = stream_id;
dc56f167 1187 loglevel = value;
8e9060f2 1188 model.emf.uri = "string";
4fa992a5 1189 context := struct {
fcba70d4
MD
1190 ...
1191 };
4fa992a5 1192 fields := struct {
80fd2569
MD
1193 ...
1194 };
3bf79539 1195};
5ba9f198 1196
555f54e6
MD
1197callsite {
1198 name = "event_name";
1199 func = "func_name";
1200 file = "myfile.c";
1201 line = 39;
5a6b4ee1 1202 ip = 0x40096c;
555f54e6
MD
1203};
1204
5ba9f198
MD
1205/* More detail on types in section 4. Types */
1206
3d13ef1a
MD
1207/*
1208 * Named types:
1209 *
4fa992a5 1210 * Type declarations behave similarly to the C standard.
3d13ef1a
MD
1211 */
1212
80af8ac6 1213typedef aliased_type_specifiers new_type_declarators;
2152348f 1214
3d13ef1a 1215/* e.g.: typedef struct example new_type_name[10]; */
80fd2569 1216
4fa992a5
MD
1217/*
1218 * typealias
1219 *
1220 * The "typealias" declaration can be used to give a name (including
80af8ac6
MD
1221 * pointer declarator specifier) to a type. It should also be used to
1222 * map basic C types (float, int, unsigned long, ...) to a CTF type.
1223 * Typealias is a superset of "typedef": it also allows assignment of a
38b8da21 1224 * simple variable identifier to a type.
4fa992a5
MD
1225 */
1226
1227typealias type_class {
80fd2569 1228 ...
38b8da21 1229} := type_specifiers type_declarator;
2152348f 1230
3d13ef1a
MD
1231/*
1232 * e.g.:
4fa992a5 1233 * typealias integer {
3d13ef1a
MD
1234 * size = 32;
1235 * align = 32;
1236 * signed = false;
38b8da21 1237 * } := struct page *;
359894ac
MD
1238 *
1239 * typealias integer {
1240 * size = 32;
1241 * align = 32;
1242 * signed = true;
38b8da21 1243 * } := int;
3d13ef1a 1244 */
80fd2569
MD
1245
1246struct name {
3bf79539
MD
1247 ...
1248};
5ba9f198 1249
fcba70d4
MD
1250variant name {
1251 ...
1252};
1253
a9b83695 1254enum name : integer_type {
3bf79539
MD
1255 ...
1256};
1257
2152348f 1258
4fa992a5
MD
1259/*
1260 * Unnamed types, contained within compound type fields, typedef or typealias.
1261 */
2152348f 1262
80fd2569
MD
1263struct {
1264 ...
2152348f 1265}
5ba9f198 1266
ec4404a7
MD
1267struct {
1268 ...
1269} align(value)
1270
fcba70d4
MD
1271variant {
1272 ...
1273}
1274
a9b83695 1275enum : integer_type {
80fd2569 1276 ...
2152348f
MD
1277}
1278
1279typedef type new_type[length];
3bf79539 1280
2152348f
MD
1281struct {
1282 type field_name[length];
1283}
1284
1285typedef type new_type[length_type];
1286
1287struct {
1288 type field_name[length_type];
1289}
1290
1291integer {
80fd2569 1292 ...
2152348f 1293}
3bf79539 1294
2152348f 1295floating_point {
80fd2569 1296 ...
2152348f
MD
1297}
1298
1299struct {
1300 integer_type field_name:size; /* GNU/C bitfield */
1301}
1302
1303struct {
1304 string field_name;
1305}
3bf79539 1306
fcba70d4 1307
2fa70eba
MD
13088. Clocks
1309
1310Clock metadata allows to describe the clock topology of the system, as
1311well as to detail each clock parameter. In absence of clock description,
1312it is assumed that all fields named "timestamp" use the same clock
aed18b5e 1313source, which increments once per nanosecond.
2fa70eba
MD
1314
1315Describing a clock and how it is used by streams is threefold: first,
1316the clock and clock topology should be described in a "clock"
1317description block, e.g.:
1318
d803bfcb 1319clock {
58262d97 1320 name = cycle_counter_sync;
2fa70eba 1321 uuid = "62189bee-96dc-11e0-91a8-cfa3d89f3923";
58262d97 1322 description = "Cycle counter synchronized across CPUs";
2fa70eba
MD
1323 freq = 1000000000; /* frequency, in Hz */
1324 /* precision in seconds is: 1000 * (1/freq) */
1325 precision = 1000;
a40cccda
MD
1326 /*
1327 * clock value offset from Epoch is:
1328 * offset_s + (offset * (1/freq))
1329 */
1330 offset_s = 1326476837;
1331 offset = 897235420;
ce0fadbd 1332 absolute = FALSE;
2fa70eba
MD
1333};
1334
d803bfcb
MD
1335The mandatory "name" field specifies the name of the clock identifier,
1336which can later be used as a reference. The optional field "uuid" is the
1337unique identifier of the clock. It can be used to correlate different
1338traces that use the same clock. An optional textual description string
1339can be added with the "description" field. The "freq" field is the
1340initial frequency of the clock, in Hz. If the "freq" field is not
1341present, the frequency is assumed to be 1000000000 (providing clock
1342increment of 1 ns). The optional "precision" field details the
a40cccda
MD
1343uncertainty on the clock measurements, in (1/freq) units. The "offset_s"
1344and "offset" fields indicate the offset from POSIX.1 Epoch, 1970-01-01
134500:00:00 +0000 (UTC), to the zero of value of the clock. The "offset_s"
1346field is in seconds. The "offset" field is in (1/freq) units. If any of
1347the "offset_s" or "offset" field is not present, it is assigned the 0
ce0fadbd
MD
1348value. The field "absolute" is TRUE if the clock is a global reference
1349across different clock uuid (e.g. NTP time). Otherwise, "absolute" is
1350FALSE, and the clock can be considered as synchronized only with other
1351clocks that have the same uuid.
1352
2fa70eba
MD
1353
1354Secondly, a reference to this clock should be added within an integer
1355type:
1356
1357typealias integer {
1358 size = 64; align = 1; signed = false;
58262d97 1359 map = clock.cycle_counter_sync.value;
2fa70eba
MD
1360} := uint64_ccnt_t;
1361
1362Thirdly, stream declarations can reference the clock they use as a
1363time-stamp source:
1364
1365struct packet_context {
1366 uint64_ccnt_t ccnt_begin;
1367 uint64_ccnt_t ccnt_end;
1368 /* ... */
1369};
1370
1371stream {
1372 /* ... */
1373 event.header := struct {
1374 uint64_ccnt_t timestamp;
1375 /* ... */
1376 }
1377 packet.context := struct packet_context;
1378};
1379
1380For a N-bit integer type referring to a clock, if the integer overflows
1381compared to the N low order bits of the clock prior value, then it is
1382assumed that one, and only one, overflow occurred. It is therefore
1383important that events encoding time on a small number of bits happen
1384frequently enough to detect when more than one N-bit overflow occurs.
1385
1386In a packet context, clock field names ending with "_begin" and "_end"
1387have a special meaning: this refers to the time-stamps at, respectively,
1388the beginning and the end of each packet.
1389
1390
3bf79539 1391A. Helper macros
5ba9f198
MD
1392
1393The two following macros keep track of the size of a GNU/C structure without
1394padding at the end by placing HEADER_END as the last field. A one byte end field
1395is used for C90 compatibility (C99 flexible arrays could be used here). Note
1396that this does not affect the effective structure size, which should always be
1397calculated with the header_sizeof() helper.
1398
1399#define HEADER_END char end_field
1400#define header_sizeof(type) offsetof(typeof(type), end_field)
3bf79539
MD
1401
1402
1403B. Stream Header Rationale
1404
1405An event stream is divided in contiguous event packets of variable size. These
1406subdivisions allow the trace analyzer to perform a fast binary search by time
1407within the stream (typically requiring to index only the event packet headers)
1408without reading the whole stream. These subdivisions have a variable size to
1409eliminate the need to transfer the event packet padding when partially filled
1410event packets must be sent when streaming a trace for live viewing/analysis.
1411An event packet can contain a certain amount of padding at the end. Dividing
1412streams into event packets is also useful for network streaming over UDP and
1413flight recorder mode tracing (a whole event packet can be swapped out of the
1414buffer atomically for reading).
1415
1416The stream header is repeated at the beginning of each event packet to allow
1417flexibility in terms of:
1418
1419 - streaming support,
1420 - allowing arbitrary buffers to be discarded without making the trace
1421 unreadable,
1422 - allow UDP packet loss handling by either dealing with missing event packet
1423 or asking for re-transmission.
1424 - transparently support flight recorder mode,
1425 - transparently support crash dump.
1426
6c7226e9
MD
1427
1428C. TSDL Grammar
fcba70d4 1429
4fa992a5 1430/*
6c7226e9 1431 * Common Trace Format (CTF) Trace Stream Description Language (TSDL) Grammar.
4fa992a5
MD
1432 *
1433 * Inspired from the C99 grammar:
1434 * http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf (Annex A)
6c7226e9
MD
1435 * and c++1x grammar (draft)
1436 * http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3291.pdf (Annex A)
4fa992a5
MD
1437 *
1438 * Specialized for CTF needs by including only constant and declarations from
1439 * C99 (excluding function declarations), and by adding support for variants,
6c7226e9
MD
1440 * sequences and CTF-specific specifiers. Enumeration container types
1441 * semantic is inspired from c++1x enum-base.
4fa992a5
MD
1442 */
1443
14441) Lexical grammar
1445
14461.1) Lexical elements
1447
1448token:
1449 keyword
1450 identifier
1451 constant
1452 string-literal
1453 punctuator
1454
14551.2) Keywords
1456
1457keyword: is one of
1458
ec4404a7 1459align
dbb8b280 1460callsite
4fa992a5
MD
1461const
1462char
2fa70eba 1463clock
4fa992a5
MD
1464double
1465enum
570ecabe 1466env
4fa992a5
MD
1467event
1468floating_point
1469float
1470integer
1471int
1472long
1473short
1474signed
1475stream
1476string
1477struct
1478trace
3e1e1a78 1479typealias
4fa992a5
MD
1480typedef
1481unsigned
1482variant
1483void
1484_Bool
1485_Complex
1486_Imaginary
1487
1488
14891.3) Identifiers
1490
1491identifier:
1492 identifier-nondigit
1493 identifier identifier-nondigit
1494 identifier digit
1495
1496identifier-nondigit:
1497 nondigit
1498 universal-character-name
1499 any other implementation-defined characters
1500
1501nondigit:
1502 _
1503 [a-zA-Z] /* regular expression */
1504
1505digit:
1506 [0-9] /* regular expression */
1507
15081.4) Universal character names
1509
1510universal-character-name:
1511 \u hex-quad
1512 \U hex-quad hex-quad
1513
1514hex-quad:
1515 hexadecimal-digit hexadecimal-digit hexadecimal-digit hexadecimal-digit
1516
15171.5) Constants
1518
1519constant:
1520 integer-constant
1521 enumeration-constant
1522 character-constant
1523
1524integer-constant:
1525 decimal-constant integer-suffix-opt
1526 octal-constant integer-suffix-opt
1527 hexadecimal-constant integer-suffix-opt
1528
1529decimal-constant:
1530 nonzero-digit
1531 decimal-constant digit
1532
1533octal-constant:
1534 0
1535 octal-constant octal-digit
1536
1537hexadecimal-constant:
1538 hexadecimal-prefix hexadecimal-digit
1539 hexadecimal-constant hexadecimal-digit
1540
1541hexadecimal-prefix:
1542 0x
1543 0X
1544
1545nonzero-digit:
1546 [1-9]
1547
1548integer-suffix:
1549 unsigned-suffix long-suffix-opt
1550 unsigned-suffix long-long-suffix
1551 long-suffix unsigned-suffix-opt
1552 long-long-suffix unsigned-suffix-opt
1553
1554unsigned-suffix:
1555 u
1556 U
1557
1558long-suffix:
1559 l
1560 L
1561
1562long-long-suffix:
1563 ll
1564 LL
1565
4fa992a5
MD
1566enumeration-constant:
1567 identifier
1568 string-literal
1569
1570character-constant:
1571 ' c-char-sequence '
1572 L' c-char-sequence '
1573
1574c-char-sequence:
1575 c-char
1576 c-char-sequence c-char
1577
1578c-char:
1579 any member of source charset except single-quote ('), backslash
1580 (\), or new-line character.
1581 escape-sequence
1582
1583escape-sequence:
1584 simple-escape-sequence
1585 octal-escape-sequence
1586 hexadecimal-escape-sequence
1587 universal-character-name
1588
1589simple-escape-sequence: one of
1590 \' \" \? \\ \a \b \f \n \r \t \v
1591
1592octal-escape-sequence:
1593 \ octal-digit
1594 \ octal-digit octal-digit
1595 \ octal-digit octal-digit octal-digit
1596
1597hexadecimal-escape-sequence:
1598 \x hexadecimal-digit
1599 hexadecimal-escape-sequence hexadecimal-digit
1600
16011.6) String literals
1602
1603string-literal:
1604 " s-char-sequence-opt "
1605 L" s-char-sequence-opt "
1606
1607s-char-sequence:
1608 s-char
1609 s-char-sequence s-char
1610
1611s-char:
1612 any member of source charset except double-quote ("), backslash
1613 (\), or new-line character.
1614 escape-sequence
1615
16161.7) Punctuators
1617
1618punctuator: one of
1619 [ ] ( ) { } . -> * + - < > : ; ... = ,
1620
1621
16222) Phrase structure grammar
1623
1624primary-expression:
1625 identifier
1626 constant
1627 string-literal
1628 ( unary-expression )
1629
1630postfix-expression:
1631 primary-expression
1632 postfix-expression [ unary-expression ]
1633 postfix-expression . identifier
1634 postfix-expressoin -> identifier
1635
1636unary-expression:
1637 postfix-expression
1638 unary-operator postfix-expression
1639
1640unary-operator: one of
1641 + -
1642
4fa992a5
MD
1643assignment-operator:
1644 =
1645
b9606a77
MD
1646type-assignment-operator:
1647 :=
1648
4fa992a5 1649constant-expression-range:
73d61ac3 1650 unary-expression ... unary-expression
4fa992a5
MD
1651
16522.2) Declarations:
1653
1654declaration:
689e04b4 1655 declaration-specifiers declarator-list-opt ;
4fa992a5
MD
1656 ctf-specifier ;
1657
1658declaration-specifiers:
689e04b4 1659 storage-class-specifier declaration-specifiers-opt
4fa992a5
MD
1660 type-specifier declaration-specifiers-opt
1661 type-qualifier declaration-specifiers-opt
1662
1663declarator-list:
1664 declarator
1665 declarator-list , declarator
1666
d285084f
MD
1667abstract-declarator-list:
1668 abstract-declarator
1669 abstract-declarator-list , abstract-declarator
1670
4fa992a5
MD
1671storage-class-specifier:
1672 typedef
1673
1674type-specifier:
1675 void
1676 char
1677 short
1678 int
1679 long
1680 float
1681 double
1682 signed
1683 unsigned
1684 _Bool
1685 _Complex
cfdd51ec 1686 _Imaginary
9dfcfc0f
MD
1687 struct-specifier
1688 variant-specifier
4fa992a5
MD
1689 enum-specifier
1690 typedef-name
1691 ctf-type-specifier
1692
ec4404a7 1693align-attribute:
73d61ac3 1694 align ( unary-expression )
ec4404a7 1695
4fa992a5 1696struct-specifier:
ec4404a7
MD
1697 struct identifier-opt { struct-or-variant-declaration-list-opt } align-attribute-opt
1698 struct identifier align-attribute-opt
4fa992a5
MD
1699
1700struct-or-variant-declaration-list:
1701 struct-or-variant-declaration
1702 struct-or-variant-declaration-list struct-or-variant-declaration
1703
1704struct-or-variant-declaration:
1705 specifier-qualifier-list struct-or-variant-declarator-list ;
eacb16d1 1706 declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list ;
6143bab7
MD
1707 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list ;
1708 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list ;
4fa992a5
MD
1709
1710specifier-qualifier-list:
1711 type-specifier specifier-qualifier-list-opt
1712 type-qualifier specifier-qualifier-list-opt
1713
1714struct-or-variant-declarator-list:
1715 struct-or-variant-declarator
1716 struct-or-variant-declarator-list , struct-or-variant-declarator
1717
1718struct-or-variant-declarator:
1719 declarator
73d61ac3 1720 declarator-opt : unary-expression
4fa992a5
MD
1721
1722variant-specifier:
1723 variant identifier-opt variant-tag-opt { struct-or-variant-declaration-list }
1724 variant identifier variant-tag
1725
1726variant-tag:
37ab95c3 1727 < unary-expression >
4fa992a5
MD
1728
1729enum-specifier:
1730 enum identifier-opt { enumerator-list }
1731 enum identifier-opt { enumerator-list , }
1732 enum identifier
a9b83695
MD
1733 enum identifier-opt : declaration-specifiers { enumerator-list }
1734 enum identifier-opt : declaration-specifiers { enumerator-list , }
4fa992a5
MD
1735
1736enumerator-list:
1737 enumerator
1738 enumerator-list , enumerator
1739
1740enumerator:
1741 enumeration-constant
8d2d41f7
MD
1742 enumeration-constant assignment-operator unary-expression
1743 enumeration-constant assignment-operator constant-expression-range
4fa992a5
MD
1744
1745type-qualifier:
1746 const
1747
1748declarator:
1749 pointer-opt direct-declarator
1750
1751direct-declarator:
1752 identifier
1753 ( declarator )
1ab22b2a 1754 direct-declarator [ unary-expression ]
4fa992a5 1755
d285084f
MD
1756abstract-declarator:
1757 pointer-opt direct-abstract-declarator
1758
1759direct-abstract-declarator:
1760 identifier-opt
1761 ( abstract-declarator )
1ab22b2a 1762 direct-abstract-declarator [ unary-expression ]
d285084f
MD
1763 direct-abstract-declarator [ ]
1764
4fa992a5 1765pointer:
3b0f8e4d
MD
1766 * type-qualifier-list-opt
1767 * type-qualifier-list-opt pointer
4fa992a5
MD
1768
1769type-qualifier-list:
1770 type-qualifier
1771 type-qualifier-list type-qualifier
1772
4fa992a5
MD
1773typedef-name:
1774 identifier
1775
17762.3) CTF-specific declarations
1777
1778ctf-specifier:
662baf84 1779 clock { ctf-assignment-expression-list-opt }
4fa992a5
MD
1780 event { ctf-assignment-expression-list-opt }
1781 stream { ctf-assignment-expression-list-opt }
570ecabe 1782 env { ctf-assignment-expression-list-opt }
4fa992a5 1783 trace { ctf-assignment-expression-list-opt }
555f54e6 1784 callsite { ctf-assignment-expression-list-opt }
b12919a5
MD
1785 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1786 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list
4fa992a5
MD
1787
1788ctf-type-specifier:
1789 floating_point { ctf-assignment-expression-list-opt }
1790 integer { ctf-assignment-expression-list-opt }
1791 string { ctf-assignment-expression-list-opt }
7609d3c7 1792 string
4fa992a5
MD
1793
1794ctf-assignment-expression-list:
13441bf5
MD
1795 ctf-assignment-expression ;
1796 ctf-assignment-expression-list ctf-assignment-expression ;
4fa992a5
MD
1797
1798ctf-assignment-expression:
1799 unary-expression assignment-operator unary-expression
1800 unary-expression type-assignment-operator type-specifier
eacb16d1 1801 declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list
6143bab7
MD
1802 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1803 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list
This page took 0.231086 seconds and 4 git commands to generate.