Add alignment explanation and struct align() attribute
[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 Meta-data description associated with the trace contains information on
57 trace event types expressed in the Trace Stream Description Language
58 (TSDL). This language describes:
59
60 - Trace version.
61 - Types available.
62 - Per-trace event header description.
63 - Per-stream event header description.
64 - Per-stream event context description.
65 - Per-event
66 - Event type to stream mapping.
67 - Event type to name mapping.
68 - Event type to ID mapping.
69 - Event context description.
70 - Event fields description.
71
72
73 3. Event stream
74
75 An event stream can be divided into contiguous event packets of variable
76 size. These subdivisions have a variable size. An event packet can
77 contain a certain amount of padding at the end. The stream header is
78 repeated at the beginning of each event packet. The rationale for the
79 event stream design choices is explained in Appendix B. Stream Header
80 Rationale.
81
82 The event stream header will therefore be referred to as the "event packet
83 header" throughout the rest of this document.
84
85
86 4. Types
87
88 Types are organized as type classes. Each type class belong to either of two
89 kind of types: basic types or compound types.
90
91 4.1 Basic types
92
93 A basic type is a scalar type, as described in this section. It includes
94 integers, GNU/C bitfields, enumerations, and floating point values.
95
96 4.1.1 Type inheritance
97
98 Type specifications can be inherited to allow deriving types from a
99 type class. For example, see the uint32_t named type derived from the "integer"
100 type class below ("Integers" section). Types have a precise binary
101 representation in the trace. A type class has methods to read and write these
102 types, but must be derived into a type to be usable in an event field.
103
104 4.1.2 Alignment
105
106 We define "byte-packed" types as aligned on the byte size, namely 8-bit.
107 We define "bit-packed" types as following on the next bit, as defined by the
108 "Integers" section.
109
110 Each basic type must specify its alignment, in bits. Examples of
111 possible alignments are: bit-packed (align = 1), byte-packed (align =
112 8), or word-aligned (e.g. align = 32 or align = 64). The choice depends
113 on the architecture preference and compactness vs performance trade-offs
114 of the implementation. Architectures providing fast unaligned write
115 byte-packed basic types to save space, aligning each type on byte
116 boundaries (8-bit). Architectures with slow unaligned writes align types
117 on specific alignment values. If no specific alignment is declared for a
118 type, it is assumed to be bit-packed for integers with size not multiple
119 of 8 bits and for gcc bitfields. All other basic types are byte-packed
120 by default. It is however recommended to always specify the alignment
121 explicitly. Alignment values must be power of two. Compound types are
122 aligned as specified in their individual specification.
123
124 TSDL meta-data attribute representation of a specific alignment:
125
126 align = value; /* value in bits */
127
128 4.1.3 Byte order
129
130 By default, the native endianness of the source architecture the trace is used.
131 Byte order can be overridden for a basic type by specifying a "byte_order"
132 attribute. Typical use-case is to specify the network byte order (big endian:
133 "be") to save data captured from the network into the trace without conversion.
134 If not specified, the byte order is native.
135
136 TSDL meta-data representation:
137
138 byte_order = native OR network OR be OR le; /* network and be are aliases */
139
140 4.1.4 Size
141
142 Type size, in bits, for integers and floats is that returned by "sizeof()" in C
143 multiplied by CHAR_BIT.
144 We require the size of "char" and "unsigned char" types (CHAR_BIT) to be fixed
145 to 8 bits for cross-endianness compatibility.
146
147 TSDL meta-data representation:
148
149 size = value; (value is in bits)
150
151 4.1.5 Integers
152
153 Signed integers are represented in two-complement. Integer alignment,
154 size, signedness and byte ordering are defined in the TSDL meta-data.
155 Integers aligned on byte size (8-bit) and with length multiple of byte
156 size (8-bit) correspond to the C99 standard integers. In addition,
157 integers with alignment and/or size that are _not_ a multiple of the
158 byte size are permitted; these correspond to the C99 standard bitfields,
159 with the added specification that the CTF integer bitfields have a fixed
160 binary representation. A MIT-licensed reference implementation of the
161 CTF portable bitfields is available at:
162
163 http://git.efficios.com/?p=babeltrace.git;a=blob;f=include/babeltrace/bitfield.h
164
165 Binary representation of integers:
166
167 - On little and big endian:
168 - Within a byte, high bits correspond to an integer high bits, and low bits
169 correspond to low bits.
170 - On little endian:
171 - Integer across multiple bytes are placed from the less significant to the
172 most significant.
173 - Consecutive integers are placed from lower bits to higher bits (even within
174 a byte).
175 - On big endian:
176 - Integer across multiple bytes are placed from the most significant to the
177 less significant.
178 - Consecutive integers are placed from higher bits to lower bits (even within
179 a byte).
180
181 This binary representation is derived from the bitfield implementation in GCC
182 for little and big endian. However, contrary to what GCC does, integers can
183 cross units boundaries (no padding is required). Padding can be explicitly
184 added (see 4.1.6 GNU/C bitfields) to follow the GCC layout if needed.
185
186 TSDL meta-data representation:
187
188 integer {
189 signed = true OR false; /* default false */
190 byte_order = native OR network OR be OR le; /* default native */
191 size = value; /* value in bits, no default */
192 align = value; /* value in bits */
193 }
194
195 Example of type inheritance (creation of a uint32_t named type):
196
197 typealias integer {
198 size = 32;
199 signed = false;
200 align = 32;
201 } := uint32_t;
202
203 Definition of a named 5-bit signed bitfield:
204
205 typealias integer {
206 size = 5;
207 signed = true;
208 align = 1;
209 } := int5_t;
210
211 4.1.6 GNU/C bitfields
212
213 The GNU/C bitfields follow closely the integer representation, with a
214 particularity on alignment: if a bitfield cannot fit in the current unit, the
215 unit is padded and the bitfield starts at the following unit. The unit size is
216 defined by the size of the type "unit_type".
217
218 TSDL meta-data representation:
219
220 unit_type name:size:
221
222 As an example, the following structure declared in C compiled by GCC:
223
224 struct example {
225 short a:12;
226 short b:5;
227 };
228
229 The example structure is aligned on the largest element (short). The second
230 bitfield would be aligned on the next unit boundary, because it would not fit in
231 the current unit.
232
233 4.1.7 Floating point
234
235 The floating point values byte ordering is defined in the TSDL meta-data.
236
237 Floating point values follow the IEEE 754-2008 standard interchange formats.
238 Description of the floating point values include the exponent and mantissa size
239 in bits. Some requirements are imposed on the floating point values:
240
241 - FLT_RADIX must be 2.
242 - mant_dig is the number of digits represented in the mantissa. It is specified
243 by the ISO C99 standard, section 5.2.4, as FLT_MANT_DIG, DBL_MANT_DIG and
244 LDBL_MANT_DIG as defined by <float.h>.
245 - exp_dig is the number of digits represented in the exponent. Given that
246 mant_dig is one bit more than its actual size in bits (leading 1 is not
247 needed) and also given that the sign bit always takes one bit, exp_dig can be
248 specified as:
249
250 - sizeof(float) * CHAR_BIT - FLT_MANT_DIG
251 - sizeof(double) * CHAR_BIT - DBL_MANT_DIG
252 - sizeof(long double) * CHAR_BIT - LDBL_MANT_DIG
253
254 TSDL meta-data representation:
255
256 floating_point {
257 exp_dig = value;
258 mant_dig = value;
259 byte_order = native OR network OR be OR le;
260 align = value;
261 }
262
263 Example of type inheritance:
264
265 typealias floating_point {
266 exp_dig = 8; /* sizeof(float) * CHAR_BIT - FLT_MANT_DIG */
267 mant_dig = 24; /* FLT_MANT_DIG */
268 byte_order = native;
269 align = 32;
270 } := float;
271
272 TODO: define NaN, +inf, -inf behavior.
273
274 Bit-packed, byte-packed or larger alignments can be used for floating
275 point values, similarly to integers.
276
277 4.1.8 Enumerations
278
279 Enumerations are a mapping between an integer type and a table of strings. The
280 numerical representation of the enumeration follows the integer type specified
281 by the meta-data. The enumeration mapping table is detailed in the enumeration
282 description within the meta-data. The mapping table maps inclusive value
283 ranges (or single values) to strings. Instead of being limited to simple
284 "value -> string" mappings, these enumerations map
285 "[ start_value ... end_value ] -> string", which map inclusive ranges of
286 values to strings. An enumeration from the C language can be represented in
287 this format by having the same start_value and end_value for each element, which
288 is in fact a range of size 1. This single-value range is supported without
289 repeating the start and end values with the value = string declaration.
290
291 enum name : integer_type {
292 somestring = start_value1 ... end_value1,
293 "other string" = start_value2 ... end_value2,
294 yet_another_string, /* will be assigned to end_value2 + 1 */
295 "some other string" = value,
296 ...
297 };
298
299 If the values are omitted, the enumeration starts at 0 and increment of 1 for
300 each entry:
301
302 enum name : unsigned int {
303 ZERO,
304 ONE,
305 TWO,
306 TEN = 10,
307 ELEVEN,
308 };
309
310 Overlapping ranges within a single enumeration are implementation defined.
311
312 A nameless enumeration can be declared as a field type or as part of a typedef:
313
314 enum : integer_type {
315 ...
316 }
317
318 Enumerations omitting the container type ": integer_type" use the "int"
319 type (for compatibility with C99). The "int" type must be previously
320 declared. E.g.:
321
322 typealias integer { size = 32; align = 32; signed = true } := int;
323
324 enum {
325 ...
326 }
327
328
329 4.2 Compound types
330
331 Compound are aggregation of type declarations. Compound types include
332 structures, variant, arrays, sequences, and strings.
333
334 4.2.1 Structures
335
336 Structures are aligned on the largest alignment required by basic types
337 contained within the structure. (This follows the ISO/C standard for structures)
338
339 TSDL meta-data representation of a named structure:
340
341 struct name {
342 field_type field_name;
343 field_type field_name;
344 ...
345 };
346
347 Example:
348
349 struct example {
350 integer { /* Nameless type */
351 size = 16;
352 signed = true;
353 align = 16;
354 } first_field_name;
355 uint64_t second_field_name; /* Named type declared in the meta-data */
356 };
357
358 The fields are placed in a sequence next to each other. They each possess a
359 field name, which is a unique identifier within the structure.
360
361 A nameless structure can be declared as a field type or as part of a typedef:
362
363 struct {
364 ...
365 }
366
367 Alignment for a structure compound type can be forced to a minimum value
368 by adding an "align" specifier after the declaration of a structure
369 body. This attribute is read as: align(value). The value is specified in
370 bits. The structure will be aligned on the maximum value between this
371 attribute and the alignment required by the basic types contained within
372 the structure. e.g.
373
374 struct {
375 ...
376 } align(32)
377
378 4.2.2 Variants (Discriminated/Tagged Unions)
379
380 A CTF variant is a selection between different types. A CTF variant must
381 always be defined within the scope of a structure or within fields
382 contained within a structure (defined recursively). A "tag" enumeration
383 field must appear in either the same lexical scope, prior to the variant
384 field (in field declaration order), in an uppermost lexical scope (see
385 Section 7.3.1), or in an uppermost dynamic scope (see Section 7.3.2).
386 The type selection is indicated by the mapping from the enumeration
387 value to the string used as variant type selector. The field to use as
388 tag is specified by the "tag_field", specified between "< >" after the
389 "variant" keyword for unnamed variants, and after "variant name" for
390 named variants.
391
392 The alignment of the variant is the alignment of the type as selected by the tag
393 value for the specific instance of the variant. The alignment of the type
394 containing the variant is independent of the variant alignment. The size of the
395 variant is the size as selected by the tag value for the specific instance of
396 the variant.
397
398 A named variant declaration followed by its definition within a structure
399 declaration:
400
401 variant name {
402 field_type sel1;
403 field_type sel2;
404 field_type sel3;
405 ...
406 };
407
408 struct {
409 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
410 ...
411 variant name <tag_field> v;
412 }
413
414 An unnamed variant definition within a structure is expressed by the following
415 TSDL meta-data:
416
417 struct {
418 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
419 ...
420 variant <tag_field> {
421 field_type sel1;
422 field_type sel2;
423 field_type sel3;
424 ...
425 } v;
426 }
427
428 Example of a named variant within a sequence that refers to a single tag field:
429
430 variant example {
431 uint32_t a;
432 uint64_t b;
433 short c;
434 };
435
436 struct {
437 enum : uint2_t { a, b, c } choice;
438 variant example <choice> v[unsigned int];
439 }
440
441 Example of an unnamed variant:
442
443 struct {
444 enum : uint2_t { a, b, c, d } choice;
445 /* Unrelated fields can be added between the variant and its tag */
446 int32_t somevalue;
447 variant <choice> {
448 uint32_t a;
449 uint64_t b;
450 short c;
451 struct {
452 unsigned int field1;
453 uint64_t field2;
454 } d;
455 } s;
456 }
457
458 Example of an unnamed variant within an array:
459
460 struct {
461 enum : uint2_t { a, b, c } choice;
462 variant <choice> {
463 uint32_t a;
464 uint64_t b;
465 short c;
466 } v[10];
467 }
468
469 Example of a variant type definition within a structure, where the defined type
470 is then declared within an array of structures. This variant refers to a tag
471 located in an upper lexical scope. This example clearly shows that a variant
472 type definition referring to the tag "x" uses the closest preceding field from
473 the lexical scope of the type definition.
474
475 struct {
476 enum : uint2_t { a, b, c, d } x;
477
478 typedef variant <x> { /*
479 * "x" refers to the preceding "x" enumeration in the
480 * lexical scope of the type definition.
481 */
482 uint32_t a;
483 uint64_t b;
484 short c;
485 } example_variant;
486
487 struct {
488 enum : int { x, y, z } x; /* This enumeration is not used by "v". */
489 example_variant v; /*
490 * "v" uses the "enum : uint2_t { a, b, c, d }"
491 * tag.
492 */
493 } a[10];
494 }
495
496 4.2.3 Arrays
497
498 Arrays are fixed-length. Their length is declared in the type
499 declaration within the meta-data. They contain an array of "inner type"
500 elements, which can refer to any type not containing the type of the
501 array being declared (no circular dependency). The length is the number
502 of elements in an array.
503
504 TSDL meta-data representation of a named array:
505
506 typedef elem_type name[length];
507
508 A nameless array can be declared as a field type within a structure, e.g.:
509
510 uint8_t field_name[10];
511
512 Arrays are always aligned on their element alignment requirement.
513
514 4.2.4 Sequences
515
516 Sequences are dynamically-sized arrays. They start with an integer that specify
517 the length of the sequence, followed by an array of "inner type" elements.
518 The length is the number of elements in the sequence.
519
520 TSDL meta-data representation for a named sequence:
521
522 typedef elem_type name[length_type];
523
524 A nameless sequence can be declared as a field type, e.g.:
525
526 long field_name[int];
527
528 The length type follows the integer types specifications, and the sequence
529 elements follow the "array" specifications.
530
531 4.2.5 Strings
532
533 Strings are an array of bytes of variable size and are terminated by a '\0'
534 "NULL" character. Their encoding is described in the TSDL meta-data. In
535 absence of encoding attribute information, the default encoding is
536 UTF-8.
537
538 TSDL meta-data representation of a named string type:
539
540 typealias string {
541 encoding = UTF8 OR ASCII;
542 } := name;
543
544 A nameless string type can be declared as a field type:
545
546 string field_name; /* Use default UTF8 encoding */
547
548 Strings are always aligned on byte size.
549
550 5. Event Packet Header
551
552 The event packet header consists of two parts: the "event packet header"
553 is the same for all streams of a trace. The second part, the "event
554 packet context", is described on a per-stream basis. Both are described
555 in the TSDL meta-data. The packets are aligned on architecture-page-sized
556 addresses.
557
558 Event packet header (all fields are optional, specified by TSDL meta-data):
559
560 - Magic number (CTF magic number: 0xC1FC1FC1) specifies that this is a
561 CTF packet. This magic number is optional, but when present, it should
562 come at the very beginning of the packet.
563 - Trace UUID, used to ensure the event packet match the meta-data used.
564 (note: we cannot use a meta-data checksum in every cases instead of a
565 UUID because meta-data can be appended to while tracing is active)
566 This field is optional.
567 - Stream ID, used as reference to stream description in meta-data.
568 This field is optional if there is only one stream description in the
569 meta-data, but becomes required if there are more than one stream in
570 the TSDL meta-data description.
571
572 Event packet context (all fields are optional, specified by TSDL meta-data):
573
574 - Event packet content size (in bytes).
575 - Event packet size (in bytes, includes padding).
576 - Event packet content checksum (optional). Checksum excludes the event packet
577 header.
578 - Per-stream event packet sequence count (to deal with UDP packet loss). The
579 number of significant sequence counter bits should also be present, so
580 wrap-arounds are dealt with correctly.
581 - Time-stamp at the beginning and time-stamp at the end of the event packet.
582 Both timestamps are written in the packet header, but sampled respectively
583 while (or before) writing the first event and while (or after) writing the
584 last event in the packet. The inclusive range between these timestamps should
585 include all event timestamps assigned to events contained within the packet.
586 - Events discarded count
587 - Snapshot of a per-stream free-running counter, counting the number of
588 events discarded that were supposed to be written in the stream prior to
589 the first event in the event packet.
590 * Note: producer-consumer buffer full condition should fill the current
591 event packet with padding so we know exactly where events have been
592 discarded.
593 - Lossless compression scheme used for the event packet content. Applied
594 directly to raw data. New types of compression can be added in following
595 versions of the format.
596 0: no compression scheme
597 1: bzip2
598 2: gzip
599 3: xz
600 - Cypher used for the event packet content. Applied after compression.
601 0: no encryption
602 1: AES
603 - Checksum scheme used for the event packet content. Applied after encryption.
604 0: no checksum
605 1: md5
606 2: sha1
607 3: crc32
608
609 5.1 Event Packet Header Description
610
611 The event packet header layout is indicated by the trace packet.header
612 field. Here is a recommended structure type for the packet header with
613 the fields typically expected (although these fields are each optional):
614
615 struct event_packet_header {
616 uint32_t magic;
617 uint8_t trace_uuid[16];
618 uint32_t stream_id;
619 };
620
621 trace {
622 ...
623 packet.header := struct event_packet_header;
624 };
625
626 If the magic number is not present, tools such as "file" will have no
627 mean to discover the file type.
628
629 If the trace_uuid is not present, no validation that the meta-data
630 actually corresponds to the stream is performed.
631
632 If the stream_id packet header field is missing, the trace can only
633 contain a single stream. Its "id" field can be left out, and its events
634 don't need to declare a "stream_id" field.
635
636
637 5.2 Event Packet Context Description
638
639 Event packet context example. These are declared within the stream declaration
640 in the meta-data. All these fields are optional. If the packet size field is
641 missing, the whole stream only contains a single packet. If the content
642 size field is missing, the packet is filled (no padding). The content
643 and packet sizes include all headers.
644
645 An example event packet context type:
646
647 struct event_packet_context {
648 uint64_t timestamp_begin;
649 uint64_t timestamp_end;
650 uint32_t checksum;
651 uint32_t stream_packet_count;
652 uint32_t events_discarded;
653 uint32_t cpu_id;
654 uint32_t/uint16_t content_size;
655 uint32_t/uint16_t packet_size;
656 uint8_t stream_packet_count_bits; /* Significant counter bits */
657 uint8_t compression_scheme;
658 uint8_t encryption_scheme;
659 uint8_t checksum_scheme;
660 };
661
662
663 6. Event Structure
664
665 The overall structure of an event is:
666
667 1 - Stream Packet Context (as specified by the stream meta-data)
668 2 - Event Header (as specified by the stream meta-data)
669 3 - Stream Event Context (as specified by the stream meta-data)
670 4 - Event Context (as specified by the event meta-data)
671 5 - Event Payload (as specified by the event meta-data)
672
673 This structure defines an implicit dynamic scoping, where variants
674 located in inner structures (those with a higher number in the listing
675 above) can refer to the fields of outer structures (with lower number in
676 the listing above). See Section 7.3 TSDL Scopes for more detail.
677
678 6.1 Event Header
679
680 Event headers can be described within the meta-data. We hereby propose, as an
681 example, two types of events headers. Type 1 accommodates streams with less than
682 31 event IDs. Type 2 accommodates streams with 31 or more event IDs.
683
684 One major factor can vary between streams: the number of event IDs assigned to
685 a stream. Luckily, this information tends to stay relatively constant (modulo
686 event registration while trace is being recorded), so we can specify different
687 representations for streams containing few event IDs and streams containing
688 many event IDs, so we end up representing the event ID and time-stamp as
689 densely as possible in each case.
690
691 The header is extended in the rare occasions where the information cannot be
692 represented in the ranges available in the standard event header. They are also
693 used in the rare occasions where the data required for a field could not be
694 collected: the flag corresponding to the missing field within the missing_fields
695 array is then set to 1.
696
697 Types uintX_t represent an X-bit unsigned integer, as declared with
698 either:
699
700 typealias integer { size = X; align = X; signed = false } := uintX_t;
701
702 or
703
704 typealias integer { size = X; align = 1; signed = false } := uintX_t;
705
706 6.1.1 Type 1 - Few event IDs
707
708 - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
709 preference).
710 - Native architecture byte ordering.
711 - For "compact" selection
712 - Fixed size: 32 bits.
713 - For "extended" selection
714 - Size depends on the architecture and variant alignment.
715
716 struct event_header_1 {
717 /*
718 * id: range: 0 - 30.
719 * id 31 is reserved to indicate an extended header.
720 */
721 enum : uint5_t { compact = 0 ... 30, extended = 31 } id;
722 variant <id> {
723 struct {
724 uint27_t timestamp;
725 } compact;
726 struct {
727 uint32_t id; /* 32-bit event IDs */
728 uint64_t timestamp; /* 64-bit timestamps */
729 } extended;
730 } v;
731 };
732
733
734 6.1.2 Type 2 - Many event IDs
735
736 - Aligned on 16-bit (or 8-bit if byte-packed, depending on the architecture
737 preference).
738 - Native architecture byte ordering.
739 - For "compact" selection
740 - Size depends on the architecture and variant alignment.
741 - For "extended" selection
742 - Size depends on the architecture and variant alignment.
743
744 struct event_header_2 {
745 /*
746 * id: range: 0 - 65534.
747 * id 65535 is reserved to indicate an extended header.
748 */
749 enum : uint16_t { compact = 0 ... 65534, extended = 65535 } id;
750 variant <id> {
751 struct {
752 uint32_t timestamp;
753 } compact;
754 struct {
755 uint32_t id; /* 32-bit event IDs */
756 uint64_t timestamp; /* 64-bit timestamps */
757 } extended;
758 } v;
759 };
760
761
762 6.2 Event Context
763
764 The event context contains information relative to the current event.
765 The choice and meaning of this information is specified by the TSDL
766 stream and event meta-data descriptions. The stream context is applied
767 to all events within the stream. The stream context structure follows
768 the event header. The event context is applied to specific events. Its
769 structure follows the stream context structure.
770
771 An example of stream-level event context is to save the event payload size with
772 each event, or to save the current PID with each event. These are declared
773 within the stream declaration within the meta-data:
774
775 stream {
776 ...
777 event.context := struct {
778 uint pid;
779 uint16_t payload_size;
780 };
781 };
782
783 An example of event-specific event context is to declare a bitmap of missing
784 fields, only appended after the stream event context if the extended event
785 header is selected. NR_FIELDS is the number of fields within the event (a
786 numeric value).
787
788 event {
789 context = struct {
790 variant <id> {
791 struct { } compact;
792 struct {
793 uint1_t missing_fields[NR_FIELDS]; /* missing event fields bitmap */
794 } extended;
795 } v;
796 };
797 ...
798 }
799
800 6.3 Event Payload
801
802 An event payload contains fields specific to a given event type. The fields
803 belonging to an event type are described in the event-specific meta-data
804 within a structure type.
805
806 6.3.1 Padding
807
808 No padding at the end of the event payload. This differs from the ISO/C standard
809 for structures, but follows the CTF standard for structures. In a trace, even
810 though it makes sense to align the beginning of a structure, it really makes no
811 sense to add padding at the end of the structure, because structures are usually
812 not followed by a structure of the same type.
813
814 This trick can be done by adding a zero-length "end" field at the end of the C
815 structures, and by using the offset of this field rather than using sizeof()
816 when calculating the size of a structure (see Appendix "A. Helper macros").
817
818 6.3.2 Alignment
819
820 The event payload is aligned on the largest alignment required by types
821 contained within the payload. (This follows the ISO/C standard for structures)
822
823
824 7. Trace Stream Description Language (TSDL)
825
826 The Trace Stream Description Language (TSDL) allows expression of the
827 binary trace streams layout in a C99-like Domain Specific Language
828 (DSL).
829
830
831 7.1 Meta-data
832
833 The trace stream layout description is located in the trace meta-data.
834 The meta-data is itself located in a stream identified by its name:
835 "metadata".
836
837 The meta-data description can be expressed in two different formats:
838 text-only and packet-based. The text-only description facilitates
839 generation of meta-data and provides a convenient way to enter the
840 meta-data information by hand. The packet-based meta-data provides the
841 CTF stream packet facilities (checksumming, compression, encryption,
842 network-readiness) for meta-data stream generated and transported by a
843 tracer.
844
845 The text-only meta-data file is a plain text TSDL description.
846
847 The packet-based meta-data is made of "meta-data packets", which each
848 start with a meta-data packet header. The packet-based meta-data
849 description is detected by reading the magic number "0x75D11D57" at the
850 beginning of the file. This magic number is also used to detect the
851 endianness of the architecture by trying to read the CTF magic number
852 and its counterpart in reversed endianness. The events within the
853 meta-data stream have no event header nor event context. Each event only
854 contains a "string" payload. Each meta-data packet start with a special
855 packet header, specific to the meta-data stream, which contains,
856 exactly:
857
858 struct metadata_packet_header {
859 uint32_t magic; /* 0x75D11D57 */
860 uint8_t trace_uuid[16]; /* Unique Universal Identifier */
861 uint32_t checksum; /* 0 if unused */
862 uint32_t content_size; /* in bits */
863 uint32_t packet_size; /* in bits */
864 uint8_t compression_scheme; /* 0 if unused */
865 uint8_t encryption_scheme; /* 0 if unused */
866 uint8_t checksum_scheme; /* 0 if unused */
867 };
868
869 The packet-based meta-data can be converted to a text-only meta-data by
870 concatenating all the strings in contains.
871
872 In the textual representation of the meta-data, the text contained
873 within "/*" and "*/", as well as within "//" and end of line, are
874 treated as comments. Boolean values can be represented as true, TRUE,
875 or 1 for true, and false, FALSE, or 0 for false. Within the string-based
876 meta-data description, the trace UUID is represented as a string of
877 hexadecimal digits and dashes "-". In the event packet header, the trace
878 UUID is represented as an array of bytes.
879
880
881 7.2 Declaration vs Definition
882
883 A declaration associates a layout to a type, without specifying where
884 this type is located in the event structure hierarchy (see Section 6).
885 This therefore includes typedef, typealias, as well as all type
886 specifiers. In certain circumstances (typedef, structure field and
887 variant field), a declaration is followed by a declarator, which specify
888 the newly defined type name (for typedef), or the field name (for
889 declarations located within structure and variants). Array and sequence,
890 declared with square brackets ("[" "]"), are part of the declarator,
891 similarly to C99. The enumeration base type is specified by
892 ": enum_base", which is part of the type specifier. The variant tag
893 name, specified between "<" ">", is also part of the type specifier.
894
895 A definition associates a type to a location in the event structure
896 hierarchy (see Section 6). This association is denoted by ":=", as shown
897 in Section 7.3.
898
899
900 7.3 TSDL Scopes
901
902 TSDL uses two different types of scoping: a lexical scope is used for
903 declarations and type definitions, and a dynamic scope is used for
904 variants references to tag fields.
905
906 7.3.1 Lexical Scope
907
908 Each of "trace", "stream", "event", "struct" and "variant" have their own
909 nestable declaration scope, within which types can be declared using "typedef"
910 and "typealias". A root declaration scope also contains all declarations
911 located outside of any of the aforementioned declarations. An inner
912 declaration scope can refer to type declared within its container
913 lexical scope prior to the inner declaration scope. Redefinition of a
914 typedef or typealias is not valid, although hiding an upper scope
915 typedef or typealias is allowed within a sub-scope.
916
917 7.3.2 Dynamic Scope
918
919 A dynamic scope consists in the lexical scope augmented with the
920 implicit event structure definition hierarchy presented at Section 6.
921 The dynamic scope is only used for variant tag definitions. It is used
922 at definition time to look up the location of the tag field associated
923 with a variant.
924
925 Therefore, variants in lower levels in the dynamic scope (e.g. event
926 context) can refer to a tag field located in upper levels (e.g. in the
927 event header) by specifying, in this case, the associated tag with
928 <header.field_name>. This allows, for instance, the event context to
929 define a variant referring to the "id" field of the event header as
930 selector.
931
932 The target dynamic scope must be specified explicitly when referring to
933 a field outside of the local static scope. The dynamic scope prefixes
934 are thus:
935
936 - Trace Packet Header: <trace.packet.header. >,
937 - Stream Packet Context: <stream.packet.context. >,
938 - Event Header: <stream.event.header. >,
939 - Stream Event Context: <stream.event.context. >,
940 - Event Context: <event.context. >,
941 - Event Payload: <event.fields. >.
942
943 Multiple declarations of the same field name within a single scope is
944 not valid. It is however valid to re-use the same field name in
945 different scopes. There is no possible conflict, because the dynamic
946 scope must be specified when a variant refers to a tag field located in
947 a different dynamic scope.
948
949 The information available in the dynamic scopes can be thought of as the
950 current tracing context. At trace production, information about the
951 current context is saved into the specified scope field levels. At trace
952 consumption, for each event, the current trace context is therefore
953 readable by accessing the upper dynamic scopes.
954
955
956 7.4 TSDL Examples
957
958 The grammar representing the TSDL meta-data is presented in Appendix C.
959 TSDL Grammar. This section presents a rather lighter reading that
960 consists in examples of TSDL meta-data, with template values.
961
962 The stream "id" can be left out if there is only one stream in the
963 trace. The event "id" field can be left out if there is only one event
964 in a stream.
965
966 trace {
967 major = value; /* Trace format version */
968 minor = value;
969 uuid = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"; /* Trace UUID */
970 byte_order = be OR le; /* Endianness (required) */
971 packet.header := struct {
972 uint32_t magic;
973 uint8_t trace_uuid[16];
974 uint32_t stream_id;
975 };
976 };
977
978 stream {
979 id = stream_id;
980 /* Type 1 - Few event IDs; Type 2 - Many event IDs. See section 6.1. */
981 event.header := event_header_1 OR event_header_2;
982 event.context := struct {
983 ...
984 };
985 packet.context := struct {
986 ...
987 };
988 };
989
990 event {
991 name = event_name;
992 id = value; /* Numeric identifier within the stream */
993 stream_id = stream_id;
994 context := struct {
995 ...
996 };
997 fields := struct {
998 ...
999 };
1000 };
1001
1002 /* More detail on types in section 4. Types */
1003
1004 /*
1005 * Named types:
1006 *
1007 * Type declarations behave similarly to the C standard.
1008 */
1009
1010 typedef aliased_type_specifiers new_type_declarators;
1011
1012 /* e.g.: typedef struct example new_type_name[10]; */
1013
1014 /*
1015 * typealias
1016 *
1017 * The "typealias" declaration can be used to give a name (including
1018 * pointer declarator specifier) to a type. It should also be used to
1019 * map basic C types (float, int, unsigned long, ...) to a CTF type.
1020 * Typealias is a superset of "typedef": it also allows assignment of a
1021 * simple variable identifier to a type.
1022 */
1023
1024 typealias type_class {
1025 ...
1026 } := type_specifiers type_declarator;
1027
1028 /*
1029 * e.g.:
1030 * typealias integer {
1031 * size = 32;
1032 * align = 32;
1033 * signed = false;
1034 * } := struct page *;
1035 *
1036 * typealias integer {
1037 * size = 32;
1038 * align = 32;
1039 * signed = true;
1040 * } := int;
1041 */
1042
1043 struct name {
1044 ...
1045 };
1046
1047 variant name {
1048 ...
1049 };
1050
1051 enum name : integer_type {
1052 ...
1053 };
1054
1055
1056 /*
1057 * Unnamed types, contained within compound type fields, typedef or typealias.
1058 */
1059
1060 struct {
1061 ...
1062 }
1063
1064 struct {
1065 ...
1066 } align(value)
1067
1068 variant {
1069 ...
1070 }
1071
1072 enum : integer_type {
1073 ...
1074 }
1075
1076 typedef type new_type[length];
1077
1078 struct {
1079 type field_name[length];
1080 }
1081
1082 typedef type new_type[length_type];
1083
1084 struct {
1085 type field_name[length_type];
1086 }
1087
1088 integer {
1089 ...
1090 }
1091
1092 floating_point {
1093 ...
1094 }
1095
1096 struct {
1097 integer_type field_name:size; /* GNU/C bitfield */
1098 }
1099
1100 struct {
1101 string field_name;
1102 }
1103
1104
1105 A. Helper macros
1106
1107 The two following macros keep track of the size of a GNU/C structure without
1108 padding at the end by placing HEADER_END as the last field. A one byte end field
1109 is used for C90 compatibility (C99 flexible arrays could be used here). Note
1110 that this does not affect the effective structure size, which should always be
1111 calculated with the header_sizeof() helper.
1112
1113 #define HEADER_END char end_field
1114 #define header_sizeof(type) offsetof(typeof(type), end_field)
1115
1116
1117 B. Stream Header Rationale
1118
1119 An event stream is divided in contiguous event packets of variable size. These
1120 subdivisions allow the trace analyzer to perform a fast binary search by time
1121 within the stream (typically requiring to index only the event packet headers)
1122 without reading the whole stream. These subdivisions have a variable size to
1123 eliminate the need to transfer the event packet padding when partially filled
1124 event packets must be sent when streaming a trace for live viewing/analysis.
1125 An event packet can contain a certain amount of padding at the end. Dividing
1126 streams into event packets is also useful for network streaming over UDP and
1127 flight recorder mode tracing (a whole event packet can be swapped out of the
1128 buffer atomically for reading).
1129
1130 The stream header is repeated at the beginning of each event packet to allow
1131 flexibility in terms of:
1132
1133 - streaming support,
1134 - allowing arbitrary buffers to be discarded without making the trace
1135 unreadable,
1136 - allow UDP packet loss handling by either dealing with missing event packet
1137 or asking for re-transmission.
1138 - transparently support flight recorder mode,
1139 - transparently support crash dump.
1140
1141
1142 C. TSDL Grammar
1143
1144 /*
1145 * Common Trace Format (CTF) Trace Stream Description Language (TSDL) Grammar.
1146 *
1147 * Inspired from the C99 grammar:
1148 * http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf (Annex A)
1149 * and c++1x grammar (draft)
1150 * http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3291.pdf (Annex A)
1151 *
1152 * Specialized for CTF needs by including only constant and declarations from
1153 * C99 (excluding function declarations), and by adding support for variants,
1154 * sequences and CTF-specific specifiers. Enumeration container types
1155 * semantic is inspired from c++1x enum-base.
1156 */
1157
1158 1) Lexical grammar
1159
1160 1.1) Lexical elements
1161
1162 token:
1163 keyword
1164 identifier
1165 constant
1166 string-literal
1167 punctuator
1168
1169 1.2) Keywords
1170
1171 keyword: is one of
1172
1173 align
1174 const
1175 char
1176 double
1177 enum
1178 event
1179 floating_point
1180 float
1181 integer
1182 int
1183 long
1184 short
1185 signed
1186 stream
1187 string
1188 struct
1189 trace
1190 typealias
1191 typedef
1192 unsigned
1193 variant
1194 void
1195 _Bool
1196 _Complex
1197 _Imaginary
1198
1199
1200 1.3) Identifiers
1201
1202 identifier:
1203 identifier-nondigit
1204 identifier identifier-nondigit
1205 identifier digit
1206
1207 identifier-nondigit:
1208 nondigit
1209 universal-character-name
1210 any other implementation-defined characters
1211
1212 nondigit:
1213 _
1214 [a-zA-Z] /* regular expression */
1215
1216 digit:
1217 [0-9] /* regular expression */
1218
1219 1.4) Universal character names
1220
1221 universal-character-name:
1222 \u hex-quad
1223 \U hex-quad hex-quad
1224
1225 hex-quad:
1226 hexadecimal-digit hexadecimal-digit hexadecimal-digit hexadecimal-digit
1227
1228 1.5) Constants
1229
1230 constant:
1231 integer-constant
1232 enumeration-constant
1233 character-constant
1234
1235 integer-constant:
1236 decimal-constant integer-suffix-opt
1237 octal-constant integer-suffix-opt
1238 hexadecimal-constant integer-suffix-opt
1239
1240 decimal-constant:
1241 nonzero-digit
1242 decimal-constant digit
1243
1244 octal-constant:
1245 0
1246 octal-constant octal-digit
1247
1248 hexadecimal-constant:
1249 hexadecimal-prefix hexadecimal-digit
1250 hexadecimal-constant hexadecimal-digit
1251
1252 hexadecimal-prefix:
1253 0x
1254 0X
1255
1256 nonzero-digit:
1257 [1-9]
1258
1259 integer-suffix:
1260 unsigned-suffix long-suffix-opt
1261 unsigned-suffix long-long-suffix
1262 long-suffix unsigned-suffix-opt
1263 long-long-suffix unsigned-suffix-opt
1264
1265 unsigned-suffix:
1266 u
1267 U
1268
1269 long-suffix:
1270 l
1271 L
1272
1273 long-long-suffix:
1274 ll
1275 LL
1276
1277 enumeration-constant:
1278 identifier
1279 string-literal
1280
1281 character-constant:
1282 ' c-char-sequence '
1283 L' c-char-sequence '
1284
1285 c-char-sequence:
1286 c-char
1287 c-char-sequence c-char
1288
1289 c-char:
1290 any member of source charset except single-quote ('), backslash
1291 (\), or new-line character.
1292 escape-sequence
1293
1294 escape-sequence:
1295 simple-escape-sequence
1296 octal-escape-sequence
1297 hexadecimal-escape-sequence
1298 universal-character-name
1299
1300 simple-escape-sequence: one of
1301 \' \" \? \\ \a \b \f \n \r \t \v
1302
1303 octal-escape-sequence:
1304 \ octal-digit
1305 \ octal-digit octal-digit
1306 \ octal-digit octal-digit octal-digit
1307
1308 hexadecimal-escape-sequence:
1309 \x hexadecimal-digit
1310 hexadecimal-escape-sequence hexadecimal-digit
1311
1312 1.6) String literals
1313
1314 string-literal:
1315 " s-char-sequence-opt "
1316 L" s-char-sequence-opt "
1317
1318 s-char-sequence:
1319 s-char
1320 s-char-sequence s-char
1321
1322 s-char:
1323 any member of source charset except double-quote ("), backslash
1324 (\), or new-line character.
1325 escape-sequence
1326
1327 1.7) Punctuators
1328
1329 punctuator: one of
1330 [ ] ( ) { } . -> * + - < > : ; ... = ,
1331
1332
1333 2) Phrase structure grammar
1334
1335 primary-expression:
1336 identifier
1337 constant
1338 string-literal
1339 ( unary-expression )
1340
1341 postfix-expression:
1342 primary-expression
1343 postfix-expression [ unary-expression ]
1344 postfix-expression . identifier
1345 postfix-expressoin -> identifier
1346
1347 unary-expression:
1348 postfix-expression
1349 unary-operator postfix-expression
1350
1351 unary-operator: one of
1352 + -
1353
1354 assignment-operator:
1355 =
1356
1357 type-assignment-operator:
1358 :=
1359
1360 constant-expression:
1361 unary-expression
1362
1363 constant-expression-range:
1364 constant-expression ... constant-expression
1365
1366 2.2) Declarations:
1367
1368 declaration:
1369 declaration-specifiers declarator-list-opt ;
1370 ctf-specifier ;
1371
1372 declaration-specifiers:
1373 storage-class-specifier declaration-specifiers-opt
1374 type-specifier declaration-specifiers-opt
1375 type-qualifier declaration-specifiers-opt
1376
1377 declarator-list:
1378 declarator
1379 declarator-list , declarator
1380
1381 abstract-declarator-list:
1382 abstract-declarator
1383 abstract-declarator-list , abstract-declarator
1384
1385 storage-class-specifier:
1386 typedef
1387
1388 type-specifier:
1389 void
1390 char
1391 short
1392 int
1393 long
1394 float
1395 double
1396 signed
1397 unsigned
1398 _Bool
1399 _Complex
1400 _Imaginary
1401 struct-specifier
1402 variant-specifier
1403 enum-specifier
1404 typedef-name
1405 ctf-type-specifier
1406
1407 align-attribute:
1408 align ( constant-expression )
1409
1410 struct-specifier:
1411 struct identifier-opt { struct-or-variant-declaration-list-opt } align-attribute-opt
1412 struct identifier align-attribute-opt
1413
1414 struct-or-variant-declaration-list:
1415 struct-or-variant-declaration
1416 struct-or-variant-declaration-list struct-or-variant-declaration
1417
1418 struct-or-variant-declaration:
1419 specifier-qualifier-list struct-or-variant-declarator-list ;
1420 declaration-specifiers storage-class-specifier declaration-specifiers declarator-list ;
1421 typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list ;
1422 typealias declaration-specifiers abstract-declarator-list := declarator-list ;
1423
1424 specifier-qualifier-list:
1425 type-specifier specifier-qualifier-list-opt
1426 type-qualifier specifier-qualifier-list-opt
1427
1428 struct-or-variant-declarator-list:
1429 struct-or-variant-declarator
1430 struct-or-variant-declarator-list , struct-or-variant-declarator
1431
1432 struct-or-variant-declarator:
1433 declarator
1434 declarator-opt : constant-expression
1435
1436 variant-specifier:
1437 variant identifier-opt variant-tag-opt { struct-or-variant-declaration-list }
1438 variant identifier variant-tag
1439
1440 variant-tag:
1441 < identifier >
1442
1443 enum-specifier:
1444 enum identifier-opt { enumerator-list }
1445 enum identifier-opt { enumerator-list , }
1446 enum identifier
1447 enum identifier-opt : declaration-specifiers { enumerator-list }
1448 enum identifier-opt : declaration-specifiers { enumerator-list , }
1449
1450 enumerator-list:
1451 enumerator
1452 enumerator-list , enumerator
1453
1454 enumerator:
1455 enumeration-constant
1456 enumeration-constant = constant-expression
1457 enumeration-constant = constant-expression-range
1458
1459 type-qualifier:
1460 const
1461
1462 declarator:
1463 pointer-opt direct-declarator
1464
1465 direct-declarator:
1466 identifier
1467 ( declarator )
1468 direct-declarator [ type-specifier ]
1469 direct-declarator [ constant-expression ]
1470
1471 abstract-declarator:
1472 pointer-opt direct-abstract-declarator
1473
1474 direct-abstract-declarator:
1475 identifier-opt
1476 ( abstract-declarator )
1477 direct-abstract-declarator [ type-specifier ]
1478 direct-abstract-declarator [ constant-expression ]
1479 direct-abstract-declarator [ ]
1480
1481 pointer:
1482 * type-qualifier-list-opt
1483 * type-qualifier-list-opt pointer
1484
1485 type-qualifier-list:
1486 type-qualifier
1487 type-qualifier-list type-qualifier
1488
1489 typedef-name:
1490 identifier
1491
1492 2.3) CTF-specific declarations
1493
1494 ctf-specifier:
1495 event { ctf-assignment-expression-list-opt }
1496 stream { ctf-assignment-expression-list-opt }
1497 trace { ctf-assignment-expression-list-opt }
1498 typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list ;
1499 typealias declaration-specifiers abstract-declarator-list := declarator-list ;
1500
1501 ctf-type-specifier:
1502 floating_point { ctf-assignment-expression-list-opt }
1503 integer { ctf-assignment-expression-list-opt }
1504 string { ctf-assignment-expression-list-opt }
1505
1506 ctf-assignment-expression-list:
1507 ctf-assignment-expression
1508 ctf-assignment-expression-list ; ctf-assignment-expression
1509
1510 ctf-assignment-expression:
1511 unary-expression assignment-operator unary-expression
1512 unary-expression type-assignment-operator type-specifier
1513 declaration-specifiers storage-class-specifier declaration-specifiers declarator-list
1514 typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list
1515 typealias declaration-specifiers abstract-declarator-list := declarator-list
This page took 0.098997 seconds and 5 git commands to generate.