71d9140704eb049dba1032c11c392f209f802b21
[ctf.git] / common-trace-format-proposal.txt
1
2 RFC: Common Trace Format (CTF) Proposal (pre-v1.7)
3
4 Mathieu Desnoyers, EfficiOS Inc.
5
6 The goal of the present document is to propose a trace format that suits the
7 needs of the embedded, telecom, high-performance and kernel communities. It is
8 based on the Common Trace Format Requirements (v1.4) document. It is designed to
9 allow traces to be natively generated by the Linux kernel, Linux user-space
10 applications written in C/C++, and hardware components. One major element of
11 CTF is the Trace Stream Description Language (TSDL) which flexibility
12 enables description of various binary trace stream layouts.
13
14 The latest version of this document can be found at:
15
16 git tree: git://git.efficios.com/ctf.git
17 gitweb: http://git.efficios.com/?p=ctf.git
18
19 A reference implementation of a library to read and write this trace format is
20 being implemented within the BabelTrace project, a converter between trace
21 formats. The development tree is available at:
22
23 git tree: git://git.efficios.com/babeltrace.git
24 gitweb: http://git.efficios.com/?p=babeltrace.git
25
26
27 1. Preliminary definitions
28
29 - Event Trace: An ordered sequence of events.
30 - Event Stream: An ordered sequence of events, containing a subset of the
31 trace event types.
32 - Event Packet: A sequence of physically contiguous events within an event
33 stream.
34 - Event: This is the basic entry in a trace. (aka: a trace record).
35 - An event identifier (ID) relates to the class (a type) of event within
36 an event stream.
37 e.g. event: irq_entry.
38 - An event (or event record) relates to a specific instance of an event
39 class.
40 e.g. event: irq_entry, at time X, on CPU Y
41 - Source Architecture: Architecture writing the trace.
42 - Reader Architecture: Architecture reading the trace.
43
44
45 2. High-level representation of a trace
46
47 A trace is divided into multiple event streams. Each event stream contains a
48 subset of the trace event types.
49
50 The final output of the trace, after its generation and optional transport over
51 the network, is expected to be either on permanent or temporary storage in a
52 virtual file system. Because each event stream is appended to while a trace is
53 being recorded, each is associated with a separate file for output. Therefore,
54 a stored trace can be represented as a directory containing one file per stream.
55
56 A metadata event stream contains information on trace event types
57 expressed in the Trace Stream Description Language (TSDL). It describes:
58
59 - Trace version.
60 - Types available.
61 - Per-stream event header description.
62 - Per-stream event header selection.
63 - Per-stream event context fields.
64 - Per-event
65 - Event type to stream mapping.
66 - Event type to name mapping.
67 - Event type to ID mapping.
68 - Event fields description.
69
70
71 3. Event stream
72
73 An event stream is divided in contiguous event packets of variable size. These
74 subdivisions have a variable size. An event packet can contain a certain
75 amount of padding at the end. The stream header is repeated at the
76 beginning of each event packet. The rationale for the event stream
77 design choices is explained in Appendix B. Stream Header Rationale.
78
79 The event stream header will therefore be referred to as the "event packet
80 header" throughout the rest of this document.
81
82
83 4. Types
84
85 Types are organized as type classes. Each type class belong to either of two
86 kind of types: basic types or compound types.
87
88 4.1 Basic types
89
90 A basic type is a scalar type, as described in this section. It includes
91 integers, GNU/C bitfields, enumerations, and floating point values.
92
93 4.1.1 Type inheritance
94
95 Type specifications can be inherited to allow deriving types from a
96 type class. For example, see the uint32_t named type derived from the "integer"
97 type class below ("Integers" section). Types have a precise binary
98 representation in the trace. A type class has methods to read and write these
99 types, but must be derived into a type to be usable in an event field.
100
101 4.1.2 Alignment
102
103 We define "byte-packed" types as aligned on the byte size, namely 8-bit.
104 We define "bit-packed" types as following on the next bit, as defined by the
105 "Integers" section.
106
107 All basic types, except bitfields, are either aligned on an architecture-defined
108 specific alignment or byte-packed, depending on the architecture preference.
109 Architectures providing fast unaligned write byte-packed basic types to save
110 space, aligning each type on byte boundaries (8-bit). Architectures with slow
111 unaligned writes align types on specific alignment values. If no specific
112 alignment is declared for a type, it is assumed to be bit-packed for
113 integers with size not multiple of 8 bits and for gcc bitfields. All
114 other types are byte-packed.
115
116 Metadata attribute representation of a specific alignment:
117
118 align = value; /* value in bits */
119
120 4.1.3 Byte order
121
122 By default, the native endianness of the source architecture the trace is used.
123 Byte order can be overridden for a basic type by specifying a "byte_order"
124 attribute. Typical use-case is to specify the network byte order (big endian:
125 "be") to save data captured from the network into the trace without conversion.
126 If not specified, the byte order is native.
127
128 Metadata representation:
129
130 byte_order = native OR network OR be OR le; /* network and be are aliases */
131
132 4.1.4 Size
133
134 Type size, in bits, for integers and floats is that returned by "sizeof()" in C
135 multiplied by CHAR_BIT.
136 We require the size of "char" and "unsigned char" types (CHAR_BIT) to be fixed
137 to 8 bits for cross-endianness compatibility.
138
139 Metadata representation:
140
141 size = value; (value is in bits)
142
143 4.1.5 Integers
144
145 Signed integers are represented in two-complement. Integer alignment, size,
146 signedness and byte ordering are defined in the metadata. Integers aligned on
147 byte size (8-bit) and with length multiple of byte size (8-bit) correspond to
148 the C99 standard integers. In addition, integers with alignment and/or size that
149 are _not_ a multiple of the byte size are permitted; these correspond to the C99
150 standard bitfields, with the added specification that the CTF integer bitfields
151 have a fixed binary representation. A MIT-licensed reference implementation of
152 the CTF portable bitfields is available at:
153
154 http://git.efficios.com/?p=babeltrace.git;a=blob;f=include/babeltrace/bitfield.h
155
156 Binary representation of integers:
157
158 - On little and big endian:
159 - Within a byte, high bits correspond to an integer high bits, and low bits
160 correspond to low bits.
161 - On little endian:
162 - Integer across multiple bytes are placed from the less significant to the
163 most significant.
164 - Consecutive integers are placed from lower bits to higher bits (even within
165 a byte).
166 - On big endian:
167 - Integer across multiple bytes are placed from the most significant to the
168 less significant.
169 - Consecutive integers are placed from higher bits to lower bits (even within
170 a byte).
171
172 This binary representation is derived from the bitfield implementation in GCC
173 for little and big endian. However, contrary to what GCC does, integers can
174 cross units boundaries (no padding is required). Padding can be explicitely
175 added (see 4.1.6 GNU/C bitfields) to follow the GCC layout if needed.
176
177 Metadata representation:
178
179 integer {
180 signed = true OR false; /* default false */
181 byte_order = native OR network OR be OR le; /* default native */
182 size = value; /* value in bits, no default */
183 align = value; /* value in bits */
184 }
185
186 Example of type inheritance (creation of a uint32_t named type):
187
188 typealias integer {
189 size = 32;
190 signed = false;
191 align = 32;
192 } := uint32_t;
193
194 Definition of a named 5-bit signed bitfield:
195
196 typealias integer {
197 size = 5;
198 signed = true;
199 align = 1;
200 } := int5_t;
201
202 4.1.6 GNU/C bitfields
203
204 The GNU/C bitfields follow closely the integer representation, with a
205 particularity on alignment: if a bitfield cannot fit in the current unit, the
206 unit is padded and the bitfield starts at the following unit. The unit size is
207 defined by the size of the type "unit_type".
208
209 Metadata representation:
210
211 unit_type name:size:
212
213 As an example, the following structure declared in C compiled by GCC:
214
215 struct example {
216 short a:12;
217 short b:5;
218 };
219
220 The example structure is aligned on the largest element (short). The second
221 bitfield would be aligned on the next unit boundary, because it would not fit in
222 the current unit.
223
224 4.1.7 Floating point
225
226 The floating point values byte ordering is defined in the metadata.
227
228 Floating point values follow the IEEE 754-2008 standard interchange formats.
229 Description of the floating point values include the exponent and mantissa size
230 in bits. Some requirements are imposed on the floating point values:
231
232 - FLT_RADIX must be 2.
233 - mant_dig is the number of digits represented in the mantissa. It is specified
234 by the ISO C99 standard, section 5.2.4, as FLT_MANT_DIG, DBL_MANT_DIG and
235 LDBL_MANT_DIG as defined by <float.h>.
236 - exp_dig is the number of digits represented in the exponent. Given that
237 mant_dig is one bit more than its actual size in bits (leading 1 is not
238 needed) and also given that the sign bit always takes one bit, exp_dig can be
239 specified as:
240
241 - sizeof(float) * CHAR_BIT - FLT_MANT_DIG
242 - sizeof(double) * CHAR_BIT - DBL_MANT_DIG
243 - sizeof(long double) * CHAR_BIT - LDBL_MANT_DIG
244
245 Metadata representation:
246
247 floating_point {
248 exp_dig = value;
249 mant_dig = value;
250 byte_order = native OR network OR be OR le;
251 }
252
253 Example of type inheritance:
254
255 typealias floating_point {
256 exp_dig = 8; /* sizeof(float) * CHAR_BIT - FLT_MANT_DIG */
257 mant_dig = 24; /* FLT_MANT_DIG */
258 byte_order = native;
259 } := float;
260
261 TODO: define NaN, +inf, -inf behavior.
262
263 4.1.8 Enumerations
264
265 Enumerations are a mapping between an integer type and a table of strings. The
266 numerical representation of the enumeration follows the integer type specified
267 by the metadata. The enumeration mapping table is detailed in the enumeration
268 description within the metadata. The mapping table maps inclusive value ranges
269 (or single values) to strings. Instead of being limited to simple
270 "value -> string" mappings, these enumerations map
271 "[ start_value ... end_value ] -> string", which map inclusive ranges of
272 values to strings. An enumeration from the C language can be represented in
273 this format by having the same start_value and end_value for each element, which
274 is in fact a range of size 1. This single-value range is supported without
275 repeating the start and end values with the value = string declaration.
276
277 enum name : integer_type {
278 somestring = start_value1 ... end_value1,
279 "other string" = start_value2 ... end_value2,
280 yet_another_string, /* will be assigned to end_value2 + 1 */
281 "some other string" = value,
282 ...
283 };
284
285 If the values are omitted, the enumeration starts at 0 and increment of 1 for
286 each entry:
287
288 enum name : unsigned int {
289 ZERO,
290 ONE,
291 TWO,
292 TEN = 10,
293 ELEVEN,
294 };
295
296 Overlapping ranges within a single enumeration are implementation defined.
297
298 A nameless enumeration can be declared as a field type or as part of a typedef:
299
300 enum : integer_type {
301 ...
302 }
303
304 Enumerations omitting the container type ": integer_type" use the "int"
305 type (for compatibility with C99). The "int" type must be previously
306 declared. E.g.:
307
308 typealias integer { size = 32; align = 32; signed = true } := int;
309
310 enum {
311 ...
312 }
313
314
315 4.2 Compound types
316
317 Compound are aggregation of type declarations. Compound types include
318 structures, variant, arrays, sequences, and strings.
319
320 4.2.1 Structures
321
322 Structures are aligned on the largest alignment required by basic types
323 contained within the structure. (This follows the ISO/C standard for structures)
324
325 Metadata representation of a named structure:
326
327 struct name {
328 field_type field_name;
329 field_type field_name;
330 ...
331 };
332
333 Example:
334
335 struct example {
336 integer { /* Nameless type */
337 size = 16;
338 signed = true;
339 align = 16;
340 } first_field_name;
341 uint64_t second_field_name; /* Named type declared in the metadata */
342 };
343
344 The fields are placed in a sequence next to each other. They each possess a
345 field name, which is a unique identifier within the structure.
346
347 A nameless structure can be declared as a field type or as part of a typedef:
348
349 struct {
350 ...
351 }
352
353 4.2.2 Variants (Discriminated/Tagged Unions)
354
355 A CTF variant is a selection between different types. A CTF variant must
356 always be defined within the scope of a structure or within fields
357 contained within a structure (defined recursively). A "tag" enumeration
358 field must appear in either the same lexical scope, prior to the variant
359 field (in field declaration order), in an uppermost lexical scope (see
360 Section 7.3.1), or in an uppermost dynamic scope (see Section 7.3.2).
361 The type selection is indicated by the mapping from the enumeration
362 value to the string used as variant type selector. The field to use as
363 tag is specified by the "tag_field", specified between "< >" after the
364 "variant" keyword for unnamed variants, and after "variant name" for
365 named variants.
366
367 The alignment of the variant is the alignment of the type as selected by the tag
368 value for the specific instance of the variant. The alignment of the type
369 containing the variant is independent of the variant alignment. The size of the
370 variant is the size as selected by the tag value for the specific instance of
371 the variant.
372
373 A named variant declaration followed by its definition within a structure
374 declaration:
375
376 variant name {
377 field_type sel1;
378 field_type sel2;
379 field_type sel3;
380 ...
381 };
382
383 struct {
384 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
385 ...
386 variant name <tag_field> v;
387 }
388
389 An unnamed variant definition within a structure is expressed by the following
390 metadata:
391
392 struct {
393 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
394 ...
395 variant <tag_field> {
396 field_type sel1;
397 field_type sel2;
398 field_type sel3;
399 ...
400 } v;
401 }
402
403 Example of a named variant within a sequence that refers to a single tag field:
404
405 variant example {
406 uint32_t a;
407 uint64_t b;
408 short c;
409 };
410
411 struct {
412 enum : uint2_t { a, b, c } choice;
413 variant example <choice> v[unsigned int];
414 }
415
416 Example of an unnamed variant:
417
418 struct {
419 enum : uint2_t { a, b, c, d } choice;
420 /* Unrelated fields can be added between the variant and its tag */
421 int32_t somevalue;
422 variant <choice> {
423 uint32_t a;
424 uint64_t b;
425 short c;
426 struct {
427 unsigned int field1;
428 uint64_t field2;
429 } d;
430 } s;
431 }
432
433 Example of an unnamed variant within an array:
434
435 struct {
436 enum : uint2_t { a, b, c } choice;
437 variant <choice> {
438 uint32_t a;
439 uint64_t b;
440 short c;
441 } v[10];
442 }
443
444 Example of a variant type definition within a structure, where the defined type
445 is then declared within an array of structures. This variant refers to a tag
446 located in an upper lexical scope. This example clearly shows that a variant
447 type definition referring to the tag "x" uses the closest preceding field from
448 the lexical scope of the type definition.
449
450 struct {
451 enum : uint2_t { a, b, c, d } x;
452
453 typedef variant <x> { /*
454 * "x" refers to the preceding "x" enumeration in the
455 * lexical scope of the type definition.
456 */
457 uint32_t a;
458 uint64_t b;
459 short c;
460 } example_variant;
461
462 struct {
463 enum : int { x, y, z } x; /* This enumeration is not used by "v". */
464 example_variant v; /*
465 * "v" uses the "enum : uint2_t { a, b, c, d }"
466 * tag.
467 */
468 } a[10];
469 }
470
471 4.2.3 Arrays
472
473 Arrays are fixed-length. Their length is declared in the type declaration within
474 the metadata. They contain an array of "inner type" elements, which can refer to
475 any type not containing the type of the array being declared (no circular
476 dependency). The length is the number of elements in an array.
477
478 Metadata representation of a named array:
479
480 typedef elem_type name[length];
481
482 A nameless array can be declared as a field type within a structure, e.g.:
483
484 uint8_t field_name[10];
485
486
487 4.2.4 Sequences
488
489 Sequences are dynamically-sized arrays. They start with an integer that specify
490 the length of the sequence, followed by an array of "inner type" elements.
491 The length is the number of elements in the sequence.
492
493 Metadata representation for a named sequence:
494
495 typedef elem_type name[length_type];
496
497 A nameless sequence can be declared as a field type, e.g.:
498
499 long field_name[int];
500
501 The length type follows the integer types specifications, and the sequence
502 elements follow the "array" specifications.
503
504 4.2.5 Strings
505
506 Strings are an array of bytes of variable size and are terminated by a '\0'
507 "NULL" character. Their encoding is described in the metadata. In absence of
508 encoding attribute information, the default encoding is UTF-8.
509
510 Metadata representation of a named string type:
511
512 typealias string {
513 encoding = UTF8 OR ASCII;
514 } := name;
515
516 A nameless string type can be declared as a field type:
517
518 string field_name; /* Use default UTF8 encoding */
519
520 5. Event Packet Header
521
522 The event packet header consists of two part: one is mandatory and have a fixed
523 layout. The second part, the "event packet context", has its layout described in
524 the metadata.
525
526 - Aligned on page size. Fixed size. Fields either aligned or packed (depending
527 on the architecture preference).
528 No padding at the end of the event packet header. Native architecture byte
529 ordering.
530
531 Fixed layout (event packet header):
532
533 - Magic number (CTF magic numbers: 0xC1FC1FC1 and its reverse endianness
534 representation: 0xC11FFCC1) It needs to have a non-symmetric bytewise
535 representation. Used to distinguish between big and little endian traces (this
536 information is determined by knowing the endianness of the architecture
537 reading the trace and comparing the magic number against its value and the
538 reverse, 0xC11FFCC1). This magic number specifies that we use the CTF metadata
539 description language described in this document. Different magic numbers
540 should be used for other metadata description languages.
541 - Trace UUID, used to ensure the event packet match the metadata used.
542 (note: we cannot use a metadata checksum because metadata can be appended to
543 while tracing is active)
544 - Stream ID, used as reference to stream description in metadata.
545
546 Metadata-defined layout (event packet context):
547
548 - Event packet content size (in bytes).
549 - Event packet size (in bytes, includes padding).
550 - Event packet content checksum (optional). Checksum excludes the event packet
551 header.
552 - Per-stream event packet sequence count (to deal with UDP packet loss). The
553 number of significant sequence counter bits should also be present, so
554 wrap-arounds are deal with correctly.
555 - Timestamp at the beginning and timestamp at the end of the event packet.
556 Both timestamps are written in the packet header, but sampled respectively
557 while (or before) writing the first event and while (or after) writing the
558 last event in the packet. The inclusive range between these timestamps should
559 include all event timestamps assigned to events contained within the packet.
560 - Events discarded count
561 - Snapshot of a per-stream free-running counter, counting the number of
562 events discarded that were supposed to be written in the stream prior to
563 the first event in the event packet.
564 * Note: producer-consumer buffer full condition should fill the current
565 event packet with padding so we know exactly where events have been
566 discarded.
567 - Lossless compression scheme used for the event packet content. Applied
568 directly to raw data. New types of compression can be added in following
569 versions of the format.
570 0: no compression scheme
571 1: bzip2
572 2: gzip
573 3: xz
574 - Cypher used for the event packet content. Applied after compression.
575 0: no encryption
576 1: AES
577 - Checksum scheme used for the event packet content. Applied after encryption.
578 0: no checksum
579 1: md5
580 2: sha1
581 3: crc32
582
583 5.1 Event Packet Header Fixed Layout Description
584
585 The event packet header layout is indicated by the trace packet.header
586 field. Here is an example structure type for the packet header with the
587 fields typically expected:
588
589 struct event_packet_header {
590 uint32_t magic;
591 uint8_t trace_uuid[16];
592 uint32_t stream_id;
593 };
594
595 trace {
596 ...
597 packet.header := struct event_packet_header;
598 };
599
600 If a packet header lacks the "magic" number field, the trace endianness
601 needs to be specified in the trace "endian" metadata attribute.
602
603 If the trace_uuid is not present, no validation that the metadata
604 actually corresponds to the stream is performed.
605
606 If the stream_id packet header field is missing, the trace can only
607 contain a single stream. Its "id" field can be left out, and its events
608 don't need to declare a "stream_id" field.
609
610
611 5.2 Event Packet Context Description
612
613 Event packet context example. These are declared within the stream declaration
614 in the metadata. All these fields are optional except for "content_size" and
615 "packet_size", which must be present in the context.
616
617 An example event packet context type:
618
619 struct event_packet_context {
620 uint64_t timestamp_begin;
621 uint64_t timestamp_end;
622 uint32_t checksum;
623 uint32_t stream_packet_count;
624 uint32_t events_discarded;
625 uint32_t cpu_id;
626 uint32_t/uint16_t content_size;
627 uint32_t/uint16_t packet_size;
628 uint8_t stream_packet_count_bits; /* Significant counter bits */
629 uint8_t compression_scheme;
630 uint8_t encryption_scheme;
631 uint8_t checksum_scheme;
632 };
633
634
635 6. Event Structure
636
637 The overall structure of an event is:
638
639 1 - Stream Packet Context (as specified by the stream metadata)
640 2 - Event Header (as specified by the stream metadata)
641 3 - Stream Event Context (as specified by the stream metadata)
642 4 - Event Context (as specified by the event metadata)
643 5 - Event Payload (as specified by the event metadata)
644
645 This structure defines an implicit dynamic scoping, where variants
646 located in inner structures (those with a higher number in the listing
647 above) can refer to the fields of outer structures (with lower number in
648 the listing above). See Section 7.3 TSDL Scopes for more detail.
649
650 6.1 Event Header
651
652 Event headers can be described within the metadata. We hereby propose, as an
653 example, two types of events headers. Type 1 accommodates streams with less than
654 31 event IDs. Type 2 accommodates streams with 31 or more event IDs.
655
656 One major factor can vary between streams: the number of event IDs assigned to
657 a stream. Luckily, this information tends to stay relatively constant (modulo
658 event registration while trace is being recorded), so we can specify different
659 representations for streams containing few event IDs and streams containing
660 many event IDs, so we end up representing the event ID and timestamp as densely
661 as possible in each case.
662
663 The header is extended in the rare occasions where the information cannot be
664 represented in the ranges available in the standard event header. They are also
665 used in the rare occasions where the data required for a field could not be
666 collected: the flag corresponding to the missing field within the missing_fields
667 array is then set to 1.
668
669 Types uintX_t represent an X-bit unsigned integer.
670
671
672 6.1.1 Type 1 - Few event IDs
673
674 - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
675 preference).
676 - Native architecture byte ordering.
677 - For "compact" selection
678 - Fixed size: 32 bits.
679 - For "extended" selection
680 - Size depends on the architecture and variant alignment.
681
682 struct event_header_1 {
683 /*
684 * id: range: 0 - 30.
685 * id 31 is reserved to indicate an extended header.
686 */
687 enum : uint5_t { compact = 0 ... 30, extended = 31 } id;
688 variant <id> {
689 struct {
690 uint27_t timestamp;
691 } compact;
692 struct {
693 uint32_t id; /* 32-bit event IDs */
694 uint64_t timestamp; /* 64-bit timestamps */
695 } extended;
696 } v;
697 };
698
699
700 6.1.2 Type 2 - Many event IDs
701
702 - Aligned on 16-bit (or 8-bit if byte-packed, depending on the architecture
703 preference).
704 - Native architecture byte ordering.
705 - For "compact" selection
706 - Size depends on the architecture and variant alignment.
707 - For "extended" selection
708 - Size depends on the architecture and variant alignment.
709
710 struct event_header_2 {
711 /*
712 * id: range: 0 - 65534.
713 * id 65535 is reserved to indicate an extended header.
714 */
715 enum : uint16_t { compact = 0 ... 65534, extended = 65535 } id;
716 variant <id> {
717 struct {
718 uint32_t timestamp;
719 } compact;
720 struct {
721 uint32_t id; /* 32-bit event IDs */
722 uint64_t timestamp; /* 64-bit timestamps */
723 } extended;
724 } v;
725 };
726
727
728 6.2 Event Context
729
730 The event context contains information relative to the current event. The choice
731 and meaning of this information is specified by the metadata "stream" and
732 "event" information. The "stream" context is applied to all events within the
733 stream. The "stream" context structure follows the event header. The "event"
734 context is applied to specific events. Its structure follows the "stream"
735 context stucture.
736
737 An example of stream-level event context is to save the event payload size with
738 each event, or to save the current PID with each event. These are declared
739 within the stream declaration within the metadata:
740
741 stream {
742 ...
743 event {
744 ...
745 context := struct {
746 uint pid;
747 uint16_t payload_size;
748 };
749 }
750 };
751
752 An example of event-specific event context is to declare a bitmap of missing
753 fields, only appended after the stream event context if the extended event
754 header is selected. NR_FIELDS is the number of fields within the event (a
755 numeric value).
756
757 event {
758 context = struct {
759 variant <id> {
760 struct { } compact;
761 struct {
762 uint1_t missing_fields[NR_FIELDS]; /* missing event fields bitmap */
763 } extended;
764 } v;
765 };
766 ...
767 }
768
769 6.3 Event Payload
770
771 An event payload contains fields specific to a given event type. The fields
772 belonging to an event type are described in the event-specific metadata
773 within a structure type.
774
775 6.3.1 Padding
776
777 No padding at the end of the event payload. This differs from the ISO/C standard
778 for structures, but follows the CTF standard for structures. In a trace, even
779 though it makes sense to align the beginning of a structure, it really makes no
780 sense to add padding at the end of the structure, because structures are usually
781 not followed by a structure of the same type.
782
783 This trick can be done by adding a zero-length "end" field at the end of the C
784 structures, and by using the offset of this field rather than using sizeof()
785 when calculating the size of a structure (see Appendix "A. Helper macros").
786
787 6.3.2 Alignment
788
789 The event payload is aligned on the largest alignment required by types
790 contained within the payload. (This follows the ISO/C standard for structures)
791
792
793 7. Trace Stream Description Language (TSDL)
794
795 The Trace Stream Description Language (TSDL) allows expression of the
796 binary trace streams layout in a C99-like Domain Specific Language
797 (DSL).
798
799
800 7.1 Metadata
801
802 The trace stream layout description is located in the trace meta-data.
803 The meta-data is itself located in a stream identified by its name:
804 "metadata".
805
806 It is made of "event packets", which each start with an event packet
807 header. The event type within the metadata stream have no event header
808 nor event context. Each event only contains a "string" payload without
809 any null-character. The events are packed one next to another. Each
810 event packet start with an event packet header, which contains, amongst
811 other fields, the magic number, trace UUID and packet length. In the
812 event packet header, the trace UUID is represented as an array of bytes.
813 Within the string-based metadata description, the trace UUID is
814 represented as a string of hexadecimal digits and dashes "-".
815
816 The metadata can be parsed by reading characters within the metadata
817 stream, for each packet starting after the packet header, for the length
818 of the packet payload specified in the header. Text contained within
819 "/*" and "*/", as well as within "//" and end of line, are treated as
820 comments. Boolean values can be represented as true, TRUE, or 1 for
821 true, and false, FALSE, or 0 for false.
822
823
824 7.2 Declaration vs Definition
825
826 A declaration associates a layout to a type, without specifying where
827 this type is located in the event structure hierarchy (see Section 6).
828 This therefore includes typedef, typealias, as well as all type
829 specifiers. In certain circumstances (typedef, structure field and
830 variant field), a declaration is followed by a declarator, which specify
831 the newly defined type name (for typedef), or the field name (for
832 declarations located within structure and variants). Array and sequence,
833 declared with square brackets ("[" "]"), are part of the declarator,
834 similarly to C99. The enumeration base type is specified by
835 ": enum_base", which is part of the type specifier. The variant tag
836 name, specified between "<" ">", is also part of the type specifier.
837
838 A definition associates a type to a location in the event structure
839 hierarchy (see Section 6). This association is denoted by ":=", as shown
840 in Section 7.3.
841
842
843 7.3 TSDL Scopes
844
845 TSDL uses two different types of scoping: a lexical scope is used for
846 declarations and type definitions, and a dynamic scope is used for
847 variants references to tag fields.
848
849 7.3.1 Lexical Scope
850
851 Each of "trace", "stream", "event", "struct" and "variant" have their own
852 nestable declaration scope, within which types can be declared using "typedef"
853 and "typealias". A root declaration scope also contains all declarations
854 located outside of any of the aforementioned declarations. An inner
855 declaration scope can refer to type declared within its container
856 lexical scope prior to the inner declaration scope. Redefinition of a
857 typedef or typealias is not valid, although hiding an upper scope
858 typedef or typealias is allowed within a sub-scope.
859
860 7.3.2 Dynamic Scope
861
862 A dynamic scope consists in the lexical scope augmented with the
863 implicit event structure definition hierarchy presented at Section 6.
864 The dynamic scope is only used for variant tag definitions. It is used
865 at definition time to look up the location of the tag field associated
866 with a variant.
867
868 Therefore, variants in lower levels in the dynamic scope (e.g. event
869 context) can refer to a tag field located in upper levels (e.g. in the
870 event header) by specifying, in this case, the associated tag with
871 <header.field_name>. This allows, for instance, the event context to
872 define a variant referring to the "id" field of the event header as
873 selector.
874
875 The target dynamic scope must be specified explicitly when referring to
876 a field outside of the local static scope. The dynamic scope prefixes
877 are thus:
878
879 - Stream Packet Context: <stream.packet.context. >,
880 - Event Header: <stream.event.header. >,
881 - Stream Event Context: <stream.event.context. >,
882 - Event Context: <event.context. >,
883 - Event Payload: <event.fields. >.
884
885 Multiple declarations of the same field name within a single scope is
886 not valid. It is however valid to re-use the same field name in
887 different scopes. There is no possible conflict, because the dynamic
888 scope must be specified when a variant refers to a tag field located in
889 a different dynamic scope.
890
891 The information available in the dynamic scopes can be thought of as the
892 current tracing context. At trace production, information about the
893 current context is saved into the specified scope field levels. At trace
894 consumption, for each event, the current trace context is therefore
895 readable by accessing the upper dynamic scopes.
896
897
898 7.4 TSDL Examples
899
900 The grammar representing the TSDL metadata is presented in Appendix C.
901 TSDL Grammar. This section presents a rather ligher reading that
902 consists in examples of TSDL metadata, with template values.
903
904 Note that the trace endianness is usually extracted from the packet
905 header magic number. The "endian" field is only needed when the magic
906 number packet header field is not present.
907
908 The stream "id" can be left out if there is only one stream in the
909 trace. The event "id" field can be left out if there is only one event
910 in a stream.
911
912 trace {
913 major = value; /* Trace format version */
914 minor = value;
915 uuid = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"; /* Trace UUID */
916 endian = big OR little; /* Endianness (optional) */
917 packet.header := struct {
918 uint32_t magic;
919 uint8_t trace_uuid[16];
920 uint32_t stream_id;
921 };
922 };
923
924 stream {
925 id = stream_id;
926 /* Type 1 - Few event IDs; Type 2 - Many event IDs. See section 6.1. */
927 event.header := event_header_1 OR event_header_2;
928 event.context := struct {
929 ...
930 };
931 packet.context := struct {
932 ...
933 };
934 };
935
936 event {
937 name = event_name;
938 id = value; /* Numeric identifier within the stream */
939 stream = stream_id;
940 context := struct {
941 ...
942 };
943 fields := struct {
944 ...
945 };
946 };
947
948 /* More detail on types in section 4. Types */
949
950 /*
951 * Named types:
952 *
953 * Type declarations behave similarly to the C standard.
954 */
955
956 typedef aliased_type_specifiers new_type_declarators;
957
958 /* e.g.: typedef struct example new_type_name[10]; */
959
960 /*
961 * typealias
962 *
963 * The "typealias" declaration can be used to give a name (including
964 * pointer declarator specifier) to a type. It should also be used to
965 * map basic C types (float, int, unsigned long, ...) to a CTF type.
966 * Typealias is a superset of "typedef": it also allows assignment of a
967 * simple variable identifier to a type.
968 */
969
970 typealias type_class {
971 ...
972 } := type_specifiers type_declarator;
973
974 /*
975 * e.g.:
976 * typealias integer {
977 * size = 32;
978 * align = 32;
979 * signed = false;
980 * } := struct page *;
981 *
982 * typealias integer {
983 * size = 32;
984 * align = 32;
985 * signed = true;
986 * } := int;
987 */
988
989 struct name {
990 ...
991 };
992
993 variant name {
994 ...
995 };
996
997 enum name : integer_type {
998 ...
999 };
1000
1001
1002 /*
1003 * Unnamed types, contained within compound type fields, typedef or typealias.
1004 */
1005
1006 struct {
1007 ...
1008 }
1009
1010 variant {
1011 ...
1012 }
1013
1014 enum : integer_type {
1015 ...
1016 }
1017
1018 typedef type new_type[length];
1019
1020 struct {
1021 type field_name[length];
1022 }
1023
1024 typedef type new_type[length_type];
1025
1026 struct {
1027 type field_name[length_type];
1028 }
1029
1030 integer {
1031 ...
1032 }
1033
1034 floating_point {
1035 ...
1036 }
1037
1038 struct {
1039 integer_type field_name:size; /* GNU/C bitfield */
1040 }
1041
1042 struct {
1043 string field_name;
1044 }
1045
1046
1047 A. Helper macros
1048
1049 The two following macros keep track of the size of a GNU/C structure without
1050 padding at the end by placing HEADER_END as the last field. A one byte end field
1051 is used for C90 compatibility (C99 flexible arrays could be used here). Note
1052 that this does not affect the effective structure size, which should always be
1053 calculated with the header_sizeof() helper.
1054
1055 #define HEADER_END char end_field
1056 #define header_sizeof(type) offsetof(typeof(type), end_field)
1057
1058
1059 B. Stream Header Rationale
1060
1061 An event stream is divided in contiguous event packets of variable size. These
1062 subdivisions allow the trace analyzer to perform a fast binary search by time
1063 within the stream (typically requiring to index only the event packet headers)
1064 without reading the whole stream. These subdivisions have a variable size to
1065 eliminate the need to transfer the event packet padding when partially filled
1066 event packets must be sent when streaming a trace for live viewing/analysis.
1067 An event packet can contain a certain amount of padding at the end. Dividing
1068 streams into event packets is also useful for network streaming over UDP and
1069 flight recorder mode tracing (a whole event packet can be swapped out of the
1070 buffer atomically for reading).
1071
1072 The stream header is repeated at the beginning of each event packet to allow
1073 flexibility in terms of:
1074
1075 - streaming support,
1076 - allowing arbitrary buffers to be discarded without making the trace
1077 unreadable,
1078 - allow UDP packet loss handling by either dealing with missing event packet
1079 or asking for re-transmission.
1080 - transparently support flight recorder mode,
1081 - transparently support crash dump.
1082
1083 The event stream header will therefore be referred to as the "event packet
1084 header" throughout the rest of this document.
1085
1086
1087 C. TSDL Grammar
1088
1089 /*
1090 * Common Trace Format (CTF) Trace Stream Description Language (TSDL) Grammar.
1091 *
1092 * Inspired from the C99 grammar:
1093 * http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf (Annex A)
1094 * and c++1x grammar (draft)
1095 * http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3291.pdf (Annex A)
1096 *
1097 * Specialized for CTF needs by including only constant and declarations from
1098 * C99 (excluding function declarations), and by adding support for variants,
1099 * sequences and CTF-specific specifiers. Enumeration container types
1100 * semantic is inspired from c++1x enum-base.
1101 */
1102
1103 1) Lexical grammar
1104
1105 1.1) Lexical elements
1106
1107 token:
1108 keyword
1109 identifier
1110 constant
1111 string-literal
1112 punctuator
1113
1114 1.2) Keywords
1115
1116 keyword: is one of
1117
1118 const
1119 char
1120 double
1121 enum
1122 event
1123 floating_point
1124 float
1125 integer
1126 int
1127 long
1128 short
1129 signed
1130 stream
1131 string
1132 struct
1133 trace
1134 typealias
1135 typedef
1136 unsigned
1137 variant
1138 void
1139 _Bool
1140 _Complex
1141 _Imaginary
1142
1143
1144 1.3) Identifiers
1145
1146 identifier:
1147 identifier-nondigit
1148 identifier identifier-nondigit
1149 identifier digit
1150
1151 identifier-nondigit:
1152 nondigit
1153 universal-character-name
1154 any other implementation-defined characters
1155
1156 nondigit:
1157 _
1158 [a-zA-Z] /* regular expression */
1159
1160 digit:
1161 [0-9] /* regular expression */
1162
1163 1.4) Universal character names
1164
1165 universal-character-name:
1166 \u hex-quad
1167 \U hex-quad hex-quad
1168
1169 hex-quad:
1170 hexadecimal-digit hexadecimal-digit hexadecimal-digit hexadecimal-digit
1171
1172 1.5) Constants
1173
1174 constant:
1175 integer-constant
1176 enumeration-constant
1177 character-constant
1178
1179 integer-constant:
1180 decimal-constant integer-suffix-opt
1181 octal-constant integer-suffix-opt
1182 hexadecimal-constant integer-suffix-opt
1183
1184 decimal-constant:
1185 nonzero-digit
1186 decimal-constant digit
1187
1188 octal-constant:
1189 0
1190 octal-constant octal-digit
1191
1192 hexadecimal-constant:
1193 hexadecimal-prefix hexadecimal-digit
1194 hexadecimal-constant hexadecimal-digit
1195
1196 hexadecimal-prefix:
1197 0x
1198 0X
1199
1200 nonzero-digit:
1201 [1-9]
1202
1203 integer-suffix:
1204 unsigned-suffix long-suffix-opt
1205 unsigned-suffix long-long-suffix
1206 long-suffix unsigned-suffix-opt
1207 long-long-suffix unsigned-suffix-opt
1208
1209 unsigned-suffix:
1210 u
1211 U
1212
1213 long-suffix:
1214 l
1215 L
1216
1217 long-long-suffix:
1218 ll
1219 LL
1220
1221 digit-sequence:
1222 digit
1223 digit-sequence digit
1224
1225 hexadecimal-digit-sequence:
1226 hexadecimal-digit
1227 hexadecimal-digit-sequence hexadecimal-digit
1228
1229 enumeration-constant:
1230 identifier
1231 string-literal
1232
1233 character-constant:
1234 ' c-char-sequence '
1235 L' c-char-sequence '
1236
1237 c-char-sequence:
1238 c-char
1239 c-char-sequence c-char
1240
1241 c-char:
1242 any member of source charset except single-quote ('), backslash
1243 (\), or new-line character.
1244 escape-sequence
1245
1246 escape-sequence:
1247 simple-escape-sequence
1248 octal-escape-sequence
1249 hexadecimal-escape-sequence
1250 universal-character-name
1251
1252 simple-escape-sequence: one of
1253 \' \" \? \\ \a \b \f \n \r \t \v
1254
1255 octal-escape-sequence:
1256 \ octal-digit
1257 \ octal-digit octal-digit
1258 \ octal-digit octal-digit octal-digit
1259
1260 hexadecimal-escape-sequence:
1261 \x hexadecimal-digit
1262 hexadecimal-escape-sequence hexadecimal-digit
1263
1264 1.6) String literals
1265
1266 string-literal:
1267 " s-char-sequence-opt "
1268 L" s-char-sequence-opt "
1269
1270 s-char-sequence:
1271 s-char
1272 s-char-sequence s-char
1273
1274 s-char:
1275 any member of source charset except double-quote ("), backslash
1276 (\), or new-line character.
1277 escape-sequence
1278
1279 1.7) Punctuators
1280
1281 punctuator: one of
1282 [ ] ( ) { } . -> * + - < > : ; ... = ,
1283
1284
1285 2) Phrase structure grammar
1286
1287 primary-expression:
1288 identifier
1289 constant
1290 string-literal
1291 ( unary-expression )
1292
1293 postfix-expression:
1294 primary-expression
1295 postfix-expression [ unary-expression ]
1296 postfix-expression . identifier
1297 postfix-expressoin -> identifier
1298
1299 unary-expression:
1300 postfix-expression
1301 unary-operator postfix-expression
1302
1303 unary-operator: one of
1304 + -
1305
1306 assignment-operator:
1307 =
1308
1309 type-assignment-operator:
1310 :=
1311
1312 constant-expression:
1313 unary-expression
1314
1315 constant-expression-range:
1316 constant-expression ... constant-expression
1317
1318 2.2) Declarations:
1319
1320 declaration:
1321 declaration-specifiers declarator-list-opt ;
1322 ctf-specifier ;
1323
1324 declaration-specifiers:
1325 storage-class-specifier declaration-specifiers-opt
1326 type-specifier declaration-specifiers-opt
1327 type-qualifier declaration-specifiers-opt
1328
1329 declarator-list:
1330 declarator
1331 declarator-list , declarator
1332
1333 abstract-declarator-list:
1334 abstract-declarator
1335 abstract-declarator-list , abstract-declarator
1336
1337 storage-class-specifier:
1338 typedef
1339
1340 type-specifier:
1341 void
1342 char
1343 short
1344 int
1345 long
1346 float
1347 double
1348 signed
1349 unsigned
1350 _Bool
1351 _Complex
1352 _Imaginary
1353 struct-specifier
1354 variant-specifier
1355 enum-specifier
1356 typedef-name
1357 ctf-type-specifier
1358
1359 struct-specifier:
1360 struct identifier-opt { struct-or-variant-declaration-list-opt }
1361 struct identifier
1362
1363 struct-or-variant-declaration-list:
1364 struct-or-variant-declaration
1365 struct-or-variant-declaration-list struct-or-variant-declaration
1366
1367 struct-or-variant-declaration:
1368 specifier-qualifier-list struct-or-variant-declarator-list ;
1369 declaration-specifiers storage-class-specifier declaration-specifiers declarator-list ;
1370 typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list ;
1371 typealias declaration-specifiers abstract-declarator-list := declarator-list ;
1372
1373 specifier-qualifier-list:
1374 type-specifier specifier-qualifier-list-opt
1375 type-qualifier specifier-qualifier-list-opt
1376
1377 struct-or-variant-declarator-list:
1378 struct-or-variant-declarator
1379 struct-or-variant-declarator-list , struct-or-variant-declarator
1380
1381 struct-or-variant-declarator:
1382 declarator
1383 declarator-opt : constant-expression
1384
1385 variant-specifier:
1386 variant identifier-opt variant-tag-opt { struct-or-variant-declaration-list }
1387 variant identifier variant-tag
1388
1389 variant-tag:
1390 < identifier >
1391
1392 enum-specifier:
1393 enum identifier-opt { enumerator-list }
1394 enum identifier-opt { enumerator-list , }
1395 enum identifier
1396 enum identifier-opt : declaration-specifiers { enumerator-list }
1397 enum identifier-opt : declaration-specifiers { enumerator-list , }
1398
1399 enumerator-list:
1400 enumerator
1401 enumerator-list , enumerator
1402
1403 enumerator:
1404 enumeration-constant
1405 enumeration-constant = constant-expression
1406 enumeration-constant = constant-expression-range
1407
1408 type-qualifier:
1409 const
1410
1411 declarator:
1412 pointer-opt direct-declarator
1413
1414 direct-declarator:
1415 identifier
1416 ( declarator )
1417 direct-declarator [ type-specifier ]
1418 direct-declarator [ constant-expression ]
1419
1420 abstract-declarator:
1421 pointer-opt direct-abstract-declarator
1422
1423 direct-abstract-declarator:
1424 identifier-opt
1425 ( abstract-declarator )
1426 direct-abstract-declarator [ type-specifier ]
1427 direct-abstract-declarator [ constant-expression ]
1428 direct-abstract-declarator [ ]
1429
1430 pointer:
1431 * type-qualifier-list-opt
1432 * type-qualifier-list-opt pointer
1433
1434 type-qualifier-list:
1435 type-qualifier
1436 type-qualifier-list type-qualifier
1437
1438 typedef-name:
1439 identifier
1440
1441 2.3) CTF-specific declarations
1442
1443 ctf-specifier:
1444 event { ctf-assignment-expression-list-opt }
1445 stream { ctf-assignment-expression-list-opt }
1446 trace { ctf-assignment-expression-list-opt }
1447 typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list ;
1448 typealias declaration-specifiers abstract-declarator-list := declarator-list ;
1449
1450 ctf-type-specifier:
1451 floating_point { ctf-assignment-expression-list-opt }
1452 integer { ctf-assignment-expression-list-opt }
1453 string { ctf-assignment-expression-list-opt }
1454
1455 ctf-assignment-expression-list:
1456 ctf-assignment-expression
1457 ctf-assignment-expression-list ; ctf-assignment-expression
1458
1459 ctf-assignment-expression:
1460 unary-expression assignment-operator unary-expression
1461 unary-expression type-assignment-operator type-specifier
1462 declaration-specifiers storage-class-specifier declaration-specifiers declarator-list
1463 typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list
1464 typealias declaration-specifiers abstract-declarator-list := declarator-list
This page took 0.096756 seconds and 3 git commands to generate.