Update proposal to v1.5
[ctf.git] / common-trace-format-linux-proposal.txt
CommitLineData
5ba9f198 1
3bf79539 2RFC: Common Trace Format Proposal for Linux (v1.5)
5ba9f198
MD
3
4Mathieu Desnoyers, EfficiOS Inc.
5
6The goal of the present document is to propose a trace format that suits the
7needs of the embedded, telecom, high-performance and kernel communities. It is
8based on the Common Trace Format Requirements (v1.4) document. It is designed to
3bf79539
MD
9allow tracing that is natively generated by the Linux kernel and Linux
10user-space applications written in C/C++.
5ba9f198
MD
11
12A reference implementation of a library to read and write this trace format is
13being implemented within the BabelTrace project, a converter between trace
14formats. The development tree is available at:
15
16 git tree: git://git.efficios.com/babeltrace.git
17 gitweb: http://git.efficios.com/?p=babeltrace.git
18
19
201. Preliminary definitions
21
3bf79539
MD
22 - Event Trace: An ordered sequence of events.
23 - Event Stream: An ordered sequence of events, containing a subset of the
24 trace event types.
25 - Event Packet: A sequence of physically contiguous events within an event
26 stream.
5ba9f198
MD
27 - Event: This is the basic entry in a trace. (aka: a trace record).
28 - An event identifier (ID) relates to the class (a type) of event within
3bf79539
MD
29 an event stream.
30 e.g. event: irq_entry.
5ba9f198
MD
31 - An event (or event record) relates to a specific instance of an event
32 class.
3bf79539
MD
33 e.g. event: irq_entry, at time X, on CPU Y
34 - Source Architecture: Architecture writing the trace.
35 - Reader Architecture: Architecture reading the trace.
5ba9f198
MD
36
37
382. High-level representation of a trace
39
3bf79539
MD
40A trace is divided into multiple event streams. Each event stream contains a
41subset of the trace event types.
5ba9f198 42
3bf79539
MD
43The final output of the trace, after its generation and optional transport over
44the network, is expected to be either on permanent or temporary storage in a
45virtual file system. Because each event stream is appended to while a trace is
46being recorded, each is associated with a separate file for output. Therefore,
47a stored trace can be represented as a directory containing one file per stream.
5ba9f198 48
3bf79539 49A metadata event stream contains information on trace event types. It describes:
5ba9f198
MD
50
51- Trace version.
52- Types available.
3bf79539
MD
53- Per-stream event header description.
54- Per-stream event header selection.
55- Per-stream event context fields.
5ba9f198 56- Per-event
3bf79539 57 - Event type to stream mapping.
5ba9f198
MD
58 - Event type to name mapping.
59 - Event type to ID mapping.
60 - Event fields description.
61
62
3bf79539 633. Event stream
5ba9f198 64
3bf79539
MD
65An event stream is divided in contiguous event packets of variable size. These
66subdivisions have a variable size. An event packet can contain a certain amount
67of padding at the end. The rationale for the event stream design choices is
68explained in Appendix B. Stream Header Rationale.
5ba9f198 69
3bf79539
MD
70An event stream is divided in contiguous event packets of variable size. These
71subdivisions have a variable size. An event packet can contain a certain amount
72of padding at the end. The stream header is repeated at the beginning of each
73event packet.
5ba9f198 74
3bf79539
MD
75The event stream header will therefore be referred to as the "event packet
76header" throughout the rest of this document.
5ba9f198
MD
77
78
794. Types
80
814.1 Basic types
82
83A basic type is a scalar type, as described in this section.
84
854.1.1 Type inheritance
86
87Type specifications can be inherited to allow deriving concrete types from an
88abstract type. For example, see the uint32_t type derived from the "integer"
89abstract type below ("Integers" section). Concrete types have a precise binary
90representation in the trace. Abstract types have methods to read and write these
91types, but must be derived into a concrete type to be usable in an event field.
92
93Concrete types inherit from abstract types. Abstract types can inherit from
94other abstract types.
95
964.1.2 Alignment
97
98We define "byte-packed" types as aligned on the byte size, namely 8-bit.
99We define "bit-packed" types as following on the next bit, as defined by the
100"bitfields" section.
5ba9f198 101
3bf79539
MD
102All basic types, except bitfields, are either aligned on an architecture-defined
103specific alignment or byte-packed, depending on the architecture preference.
104Architectures providing fast unaligned write byte-packed basic types to save
5ba9f198 105space, aligning each type on byte boundaries (8-bit). Architectures with slow
3bf79539
MD
106unaligned writes align types on specific alignment values. If no specific
107alignment is declared for a type nor its parents, it is assumed to be bit-packed
108for bitfields and byte-packed for other types.
5ba9f198 109
3bf79539 110Metadata attribute representation of a specific alignment:
5ba9f198
MD
111
112 align = value; /* value in bits */
113
1144.1.3 Byte order
115
3bf79539
MD
116By default, the native endianness of the source architecture the trace is used.
117Byte order can be overridden for a basic type by specifying a "byte_order"
118attribute. Typical use-case is to specify the network byte order (big endian:
119"be") to save data captured from the network into the trace without conversion.
120If not specified, the byte order is native.
5ba9f198
MD
121
122Metadata representation:
123
124 byte_order = native OR network OR be OR le; /* network and be are aliases */
125
1264.1.4 Size
127
128Type size, in bits, for integers and floats is that returned by "sizeof()" in C
129multiplied by CHAR_BIT.
130We require the size of "char" and "unsigned char" types (CHAR_BIT) to be fixed
131to 8 bits for cross-endianness compatibility.
132
133Metadata representation:
134
135 size = value; (value is in bits)
136
1374.1.5 Integers
138
139Signed integers are represented in two-complement. Integer alignment, size,
140signedness and byte ordering are defined in the metadata. Integers aligned on
141byte size (8-bit) and with length multiple of byte size (8-bit) correspond to
142the C99 standard integers. In addition, integers with alignment and/or size that
143are _not_ a multiple of the byte size are permitted; these correspond to the C99
144standard bitfields, with the added specification that the CTF integer bitfields
145have a fixed binary representation. A MIT-licensed reference implementation of
146the CTF portable bitfields is available at:
147
148 http://git.efficios.com/?p=babeltrace.git;a=blob;f=include/babeltrace/bitfield.h
149
150Binary representation of integers:
151
152- On little and big endian:
153 - Within a byte, high bits correspond to an integer high bits, and low bits
154 correspond to low bits.
155- On little endian:
156 - Integer across multiple bytes are placed from the less significant to the
157 most significant.
158 - Consecutive integers are placed from lower bits to higher bits (even within
159 a byte).
160- On big endian:
161 - Integer across multiple bytes are placed from the most significant to the
162 less significant.
163 - Consecutive integers are placed from higher bits to lower bits (even within
164 a byte).
165
166This binary representation is derived from the bitfield implementation in GCC
167for little and big endian. However, contrary to what GCC does, integers can
168cross units boundaries (no padding is required). Padding can be explicitely
169added (see 4.1.6 GNU/C bitfields) to follow the GCC layout if needed.
170
171Metadata representation:
172
173 abstract_type integer {
174 signed = true OR false; /* default false */
175 byte_order = native OR network OR be OR le; /* default native */
176 size = value; /* value in bits, no default */
177 align = value; /* value in bits */
3bf79539 178 };
5ba9f198
MD
179
180Example of type inheritance (creation of a concrete type uint32_t):
181
182type uint32_t {
183 parent = integer;
184 size = 8;
185 signed = false;
186 align = 32;
3bf79539 187};
5ba9f198
MD
188
189Definition of a 5-bit signed bitfield:
190
191type int5_t {
192 parent = integer;
193 size = 5;
194 signed = true;
195 align = 1;
3bf79539 196};
5ba9f198
MD
197
1984.1.6 GNU/C bitfields
199
200The GNU/C bitfields follow closely the integer representation, with a
201particularity on alignment: if a bitfield cannot fit in the current unit, the
202unit is padded and the bitfield starts at the following unit. We therefore need
203to express the extra "unit size" information.
204
205Metadata representation:
206
207abstract_type gcc_bitfield {
208 parent = integer;
209 unit_size = value;
3bf79539 210};
5ba9f198
MD
211
212As an example, the following structure declared in C compiled by GCC:
213
214struct example {
215 short a:12;
216 short b:5;
217};
218
219Would correspond to the following structure, aligned on the largest element
220(short). The second bitfield would be aligned on the next unit boundary, because
221it would not fit in the current unit.
222
223type struct_example {
224 parent = struct;
225 fields = {
3bf79539
MD
226 type {
227 parent = gcc_bitfield;
228 unit_size = 16; /* sizeof(short) */
229 size = 12;
230 signed = true;
231 align = 1;
232 } a;
233 type {
234 parent = gcc_bitfield;
235 unit_size = 16; /* sizeof(short) */
236 size = 5;
237 signed = true;
238 align = 1;
239 } b;
5ba9f198 240 };
3bf79539 241};
5ba9f198
MD
242
2434.1.7 Floating point
244
245The floating point values byte ordering is defined in the metadata.
246
247Floating point values follow the IEEE 754-2008 standard interchange formats.
248Description of the floating point values include the exponent and mantissa size
249in bits. Some requirements are imposed on the floating point values:
250
251- FLT_RADIX must be 2.
252- mant_dig is the number of digits represented in the mantissa. It is specified
253 by the ISO C99 standard, section 5.2.4, as FLT_MANT_DIG, DBL_MANT_DIG and
254 LDBL_MANT_DIG as defined by <float.h>.
255- exp_dig is the number of digits represented in the exponent. Given that
256 mant_dig is one bit more than its actual size in bits (leading 1 is not
257 needed) and also given that the sign bit always takes one bit, exp_dig can be
258 specified as:
259
260 - sizeof(float) * CHAR_BIT - FLT_MANT_DIG
261 - sizeof(double) * CHAR_BIT - DBL_MANT_DIG
262 - sizeof(long double) * CHAR_BIT - LDBL_MANT_DIG
263
264Metadata representation:
265
266abstract_type floating_point {
267 exp_dig = value;
268 mant_dig = value;
269 byte_order = native OR network OR be OR le;
3bf79539 270};
5ba9f198
MD
271
272Example of type inheritance:
273
274type float {
275 exp_dig = 8; /* sizeof(float) * CHAR_BIT - FLT_MANT_DIG */
276 mant_dig = 24; /* FLT_MANT_DIG */
277 byte_order = native;
3bf79539 278};
5ba9f198
MD
279
280TODO: define NaN, +inf, -inf behavior.
281
2824.1.8 Enumerations
283
284Enumerations are a mapping between an integer type and a table of strings. The
285numerical representation of the enumeration follows the integer type specified
286by the metadata. The enumeration mapping table is detailed in the enumeration
3bf79539
MD
287description within the metadata. The mapping table maps inclusive value ranges
288(or single values) to strings. Instead of being limited to simple
289"value -> string" mappings, these enumerations map
290"[ start_value .. end_value ] -> string", which map inclusive ranges of
291values to strings. An enumeration from the C language can be represented in
292this format by having the same start_value and end_value for each element, which
293is in fact a range of size 1. This single-value range is supported without
294repeating the start and end values with the { value, string } declaration.
295
296abstract_type enum {
297 parent = integer;
298 map = {
299 { { start_value, end_value }, string },
300 { { start_value, end_value }, string },
301 { { start_value, end_value }, string },
302 { value, string },
303 { value, string },
5ba9f198
MD
304 ...
305 };
3bf79539 306};
5ba9f198
MD
307
308
3094.2 Compound types
310
3114.2.1 Structures
312
313Structures are aligned on the largest alignment required by basic types
314contained within the structure. (This follows the ISO/C standard for structures)
315
316Metadata representation:
317
318abstract_type struct {
319 fields = {
3bf79539
MD
320 field_type field_name;
321 field_type field_name;
5ba9f198
MD
322 ...
323 };
324}
325
326Example:
327
328type struct_example {
329 parent = struct;
330 fields = {
3bf79539
MD
331 type { /* Nameless type */
332 parent = integer;
333 size = 16;
334 signed = true;
335 align = 16;
336 } first_field_name;
337 uint64_t second_field_name; /* Named type declared in the metadata */
5ba9f198 338 };
3bf79539 339};
5ba9f198
MD
340
341The fields are placed in a sequence next to each other. They each possess a
342field name, which is a unique identifier within the structure.
343
3444.2.2 Arrays
345
346Arrays are fixed-length. Their length is declared in the type declaration within
347the metadata. They contain an array of "inner type" elements, which can refer to
348any type not containing the type of the array being declared (no circular
3bf79539 349dependency). The length is the number of elements in an array.
5ba9f198
MD
350
351Metadata representation:
352
353abstract_type array {
354 length = value;
355 elem_type = type;
3bf79539 356};
5ba9f198
MD
357
358E.g.:
359
360type example_array {
361 parent = array;
362 length = 10;
363 elem_type = uint32_t;
3bf79539 364};
5ba9f198
MD
365
3664.2.3 Sequences
367
368Sequences are dynamically-sized arrays. They start with an integer that specify
369the length of the sequence, followed by an array of "inner type" elements.
3bf79539 370The length is the number of elements in the sequence.
5ba9f198
MD
371
372abstract_type sequence {
373 length_type = type; /* Inheriting from integer */
374 elem_type = type;
3bf79539 375};
5ba9f198
MD
376
377The integer type follows the integer types specifications, and the sequence
378elements follow the "array" specifications.
379
3804.2.4 Strings
381
382Strings are an array of bytes of variable size and are terminated by a '\0'
383"NULL" character. Their encoding is described in the metadata. In absence of
384encoding attribute information, the default encoding is UTF-8.
385
386abstract_type string {
387 encoding = UTF8 OR ASCII;
3bf79539 388};
5ba9f198
MD
389
390
3bf79539
MD
3915. Event Packet Header
392
393The event packet header consists of two part: one is mandatory and have a fixed
394layout. The second part, the "event packet context", has its layout described in
395the metadata.
5ba9f198 396
3bf79539
MD
397- Aligned on page size. Fixed size. Fields either aligned or packed (depending
398 on the architecture preference).
399 No padding at the end of the event packet header. Native architecture byte
5ba9f198 400 ordering.
3bf79539
MD
401
402Fixed layout (event packet header):
403
5ba9f198
MD
404- Magic number (CTF magic numbers: 0xC1FC1FC1 and its reverse endianness
405 representation: 0xC11FFCC1) It needs to have a non-symmetric bytewise
406 representation. Used to distinguish between big and little endian traces (this
407 information is determined by knowing the endianness of the architecture
408 reading the trace and comparing the magic number against its value and the
409 reverse, 0xC11FFCC1). This magic number specifies that we use the CTF metadata
410 description language described in this document. Different magic numbers
411 should be used for other metadata description languages.
3bf79539 412- Trace UUID, used to ensure the event packet match the metadata used.
5ba9f198
MD
413 (note: we cannot use a metadata checksum because metadata can be appended to
414 while tracing is active)
3bf79539
MD
415- Stream ID, used as reference to stream description in metadata.
416
417Metadata-defined layout (event packet context):
418
419- Event packet content size (in bytes).
420- Event packet size (in bytes, includes padding).
421- Event packet content checksum (optional). Checksum excludes the event packet
422 header.
423- Per-stream event packet sequence count (to deal with UDP packet loss). The
424 number of significant sequence counter bits should also be present, so
425 wrap-arounds are deal with correctly.
426- Timestamp at the beginning and timestamp at the end of the event packet.
427 Both timestamps are written in the packet header, but sampled respectively
428 while (or before) writing the first event and while (or after) writing the
429 last event in the packet. The inclusive range between these timestamps should
430 include all event timestamps assigned to events contained within the packet.
5ba9f198 431- Events discarded count
3bf79539
MD
432 - Snapshot of a per-stream free-running counter, counting the number of
433 events discarded that were supposed to be written in the stream prior to
434 the first event in the event packet.
5ba9f198 435 * Note: producer-consumer buffer full condition should fill the current
3bf79539 436 event packet with padding so we know exactly where events have been
5ba9f198 437 discarded.
3bf79539
MD
438- Lossless compression scheme used for the event packet content. Applied
439 directly to raw data. New types of compression can be added in following
440 versions of the format.
5ba9f198
MD
441 0: no compression scheme
442 1: bzip2
443 2: gzip
3bf79539
MD
444 3: xz
445- Cypher used for the event packet content. Applied after compression.
5ba9f198
MD
446 0: no encryption
447 1: AES
3bf79539 448- Checksum scheme used for the event packet content. Applied after encryption.
5ba9f198
MD
449 0: no checksum
450 1: md5
451 2: sha1
452 3: crc32
453
3bf79539
MD
4545.1 Event Packet Header Fixed Layout Description
455
456type event_packet_header {
5ba9f198
MD
457 parent = struct;
458 fields = {
3bf79539
MD
459 uint32_t magic;
460 type {
461 parent = array;
462 length = 16;
463 elem_type = uint8_t;
464 } trace_uuid;
465 uint32_t stream_id;
466}
5ba9f198 467
3bf79539
MD
4685.2 Event Packet Context Description
469
470Event packet context example. These are declared within the stream declaration
471in the metadata. All these fields are optional except for "content_size" and
472"packet_size", which must be present in the context.
473
474An example event packet context type:
475
476type event_packet_context {
477 uint64_t timestamp_begin;
478 uint64_t timestamp_end;
479 uint32_t checksum;
480 uint32_t stream_packet_count;
481 uint32_t events_discarded;
482 uint32_t cpu_id;
483 uint32_t/uint16_t content_size;
484 uint32_t/uint16_t packet_size;
485 uint8_t stream_packet_count_bits; /* Significant counter bits */
486 uint8_t compression_scheme;
487 uint8_t encryption_scheme;
488 uint8_t checksum;
489};
5ba9f198
MD
490
4916. Event Structure
492
493The overall structure of an event is:
494
3bf79539 495 - Event Header (as specifed by the stream metadata)
5ba9f198 496 - Extended Event Header (as specified by the event header)
3bf79539 497 - Event Context (as specified by the stream metadata)
5ba9f198
MD
498 - Event Payload (as specified by the event metadata)
499
500
5016.1 Event Header
502
3bf79539
MD
503One major factor can vary between streams: the number of event IDs assigned to
504a stream. Luckily, this information tends to stay relatively constant (modulo
5ba9f198 505event registration while trace is being recorded), so we can specify different
3bf79539 506representations for streams containing few event IDs and streams containing
5ba9f198
MD
507many event IDs, so we end up representing the event ID and timestamp as densely
508as possible in each case.
509
3bf79539
MD
510We therefore provide two types of events headers. Type 1 accommodates streams
511with less than 31 event IDs. Type 2 accommodates streams with 31 or more event
5ba9f198
MD
512IDs.
513
514The "extended headers" are used in the rare occasions where the information
3bf79539
MD
515cannot be represented in the ranges available in the event header. They are also
516used in the rare occasions where the data required for a field could not be
517collected: the flag corresponding to the missing field within the missing_fields
518array is then set to 1.
5ba9f198
MD
519
520Types uintX_t represent an X-bit unsigned integer.
521
522
5236.1.1 Type 1 - Few event IDs
524
525 - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
526 preference).
527 - Fixed size: 32 bits.
528 - Native architecture byte ordering.
529
530type event_header_1 {
531 parent = struct;
532 fields = {
3bf79539 533 uint5_t id; /*
5ba9f198
MD
534 * id: range: 0 - 30.
535 * id 31 is reserved to indicate a following
536 * extended header.
537 */
3bf79539 538 uint27_t timestamp;
5ba9f198
MD
539 };
540};
541
542The end of a type 1 header is aligned on a 32-bit boundary (or packed).
543
544
5456.1.2 Extended Type 1 Event Header
546
547 - Follows struct event_header_1, which is aligned on 32-bit, so no need to
548 realign.
3bf79539 549 - Variable size (depends on the number of fields per event).
5ba9f198
MD
550 - Native architecture byte ordering.
551
552type event_header_1_ext {
553 parent = struct;
554 fields = {
3bf79539
MD
555 uint32_t id; /* 32-bit event IDs */
556 uint64_t timestamp; /* 64-bit timestamps */
557 type {
558 parent = array;
559 length = NR_FIELDS; /* Number of fields within the event */
560 elem_type = uint1_t; /* 1-bit bitfield */
561 } missing_fields;
5ba9f198
MD
562 };
563};
564
5ba9f198
MD
565
5666.1.3 Type 2 - Many event IDs
567
568 - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
569 preference).
570 - Fixed size: 48 bits.
571 - Native architecture byte ordering.
572
573type event_header_2 {
574 parent = struct;
575 fields = {
3bf79539
MD
576 uint32_t timestamp;
577 uint16_t id; /*
5ba9f198
MD
578 * id: range: 0 - 65534.
579 * id 65535 is reserved to indicate a following
580 * extended header.
581 */
582 };
583};
584
585The end of a type 2 header is aligned on a 16-bit boundary (or 8-bit if
586byte-packed).
587
588
5896.1.4 Extended Type 2 Event Header
590
591 - Follows struct event_header_2, which alignment end on a 16-bit boundary, so
3bf79539 592 we need to align on 64-bit integer architecture alignment (or 8-bit if
5ba9f198 593 byte-packed).
3bf79539 594 - Variable size (depends on the number of fields per event).
5ba9f198
MD
595 - Native architecture byte ordering.
596
597type event_header_2_ext {
598 parent = struct;
599 fields = {
3bf79539
MD
600 uint64_t timestamp; /* 64-bit timestamps */
601 uint32_t id; /* 32-bit event IDs */
602 type {
603 parent = array;
604 length = NR_FIELDS; /* Number of fields within the event */
605 elem_type = uint1_t; /* 1-bit bitfield */
606 } missing_fields;
5ba9f198
MD
607 };
608};
609
5ba9f198
MD
610
6116.2 Event Context
612
613The event context contains information relative to the current event. The choice
3bf79539 614and meaning of this information is specified by the metadata "stream"
5ba9f198 615information. For this trace format, event context is usually empty, except when
3bf79539 616the metadata "stream" information specifies otherwise by declaring a non-empty
5ba9f198
MD
617structure for the event context. An example of event context is to save the
618event payload size with each event, or to save the current PID with each event.
3bf79539 619These are declared within the stream declaration within the metadata.
5ba9f198 620
3bf79539 621An example event context type:
5ba9f198 622
3bf79539
MD
623 type event_context {
624 parent = struct;
625 fields = {
626 uint pid;
627 uint16_t payload_size;
628 };
629 };
5ba9f198
MD
630
631
6326.3 Event Payload
633
634An event payload contains fields specific to a given event type. The fields
635belonging to an event type are described in the event-specific metadata
636within a structure type.
637
6386.3.1 Padding
639
640No padding at the end of the event payload. This differs from the ISO/C standard
641for structures, but follows the CTF standard for structures. In a trace, even
642though it makes sense to align the beginning of a structure, it really makes no
643sense to add padding at the end of the structure, because structures are usually
644not followed by a structure of the same type.
645
646This trick can be done by adding a zero-length "end" field at the end of the C
647structures, and by using the offset of this field rather than using sizeof()
3bf79539 648when calculating the size of a structure (see Appendix "A. Helper macros").
5ba9f198
MD
649
6506.3.2 Alignment
651
652The event payload is aligned on the largest alignment required by types
653contained within the payload. (This follows the ISO/C standard for structures)
654
655
656
6577. Metadata
658
3bf79539
MD
659The meta-data is located in a stream named "metadata". It is made of "event
660packets", which each start with an event packet header. The event type within
661the metadata stream have no event header nor event context. Each event only
5ba9f198 662contains a null-terminated "string" payload, which is a metadata description
3bf79539
MD
663entry. The events are packed one next to another. Each event packet start with
664an event packet header, which contains, amongst other fields, the magic number
665and trace UUID.
5ba9f198
MD
666
667The metadata can be parsed by reading through the metadata strings, skipping
3bf79539 668newlines and null-characters. Type names may contain spaces.
5ba9f198
MD
669
670trace {
671 major = value; /* Trace format version */
672 minor = value;
3bf79539
MD
673 uuid = value; /* Trace UUID */
674 word_size = value;
675};
5ba9f198 676
3bf79539
MD
677stream {
678 id = stream_id;
5ba9f198 679 event {
3bf79539
MD
680 /* Type 1 - Few event IDs; Type 2 - Many event IDs. See section 6.1. */
681 header_type = event_header_1 OR event_header_2;
682 /*
683 * Extended event header type. Only present if specified in event header
684 * on a per-event basis.
685 */
686 header_type_ext = event_header_1_ext OR event_header_2_ext;
687 context_type = type inheriting from "struct" abstract type;
688 };
689 packet {
690 context_type = type inheriting from "struct" abstract type;
691 };
692};
5ba9f198
MD
693
694event {
3bf79539
MD
695 name = eventname;
696 id = value; /* Numeric identifier within the stream */
697 stream = stream_id;
698 fields = type inheriting from "struct" abstract type;
699};
5ba9f198
MD
700
701/* More detail on types in section 4. Types */
702
703/* Named types */
3bf79539
MD
704type {
705 name = typename;
706 ...
707};
5ba9f198 708
3bf79539 709/* Unnamed types, contained within compound type fields or type assignments. */
5ba9f198 710type {
3bf79539
MD
711 ...
712};
713
5ba9f198 714
3bf79539
MD
715Structure types used for fields and context_type implicitly inherit from
716"struct" and require no "type" identifier before the braces. E.g.:
717
718 fields = {
719 typeA name1;
720 typeB name2;
721 ...
722 };
723
724
725A. Helper macros
5ba9f198
MD
726
727The two following macros keep track of the size of a GNU/C structure without
728padding at the end by placing HEADER_END as the last field. A one byte end field
729is used for C90 compatibility (C99 flexible arrays could be used here). Note
730that this does not affect the effective structure size, which should always be
731calculated with the header_sizeof() helper.
732
733#define HEADER_END char end_field
734#define header_sizeof(type) offsetof(typeof(type), end_field)
3bf79539
MD
735
736
737B. Stream Header Rationale
738
739An event stream is divided in contiguous event packets of variable size. These
740subdivisions allow the trace analyzer to perform a fast binary search by time
741within the stream (typically requiring to index only the event packet headers)
742without reading the whole stream. These subdivisions have a variable size to
743eliminate the need to transfer the event packet padding when partially filled
744event packets must be sent when streaming a trace for live viewing/analysis.
745An event packet can contain a certain amount of padding at the end. Dividing
746streams into event packets is also useful for network streaming over UDP and
747flight recorder mode tracing (a whole event packet can be swapped out of the
748buffer atomically for reading).
749
750The stream header is repeated at the beginning of each event packet to allow
751flexibility in terms of:
752
753 - streaming support,
754 - allowing arbitrary buffers to be discarded without making the trace
755 unreadable,
756 - allow UDP packet loss handling by either dealing with missing event packet
757 or asking for re-transmission.
758 - transparently support flight recorder mode,
759 - transparently support crash dump.
760
761The event stream header will therefore be referred to as the "event packet
762header" throughout the rest of this document.
This page took 0.083836 seconds and 4 git commands to generate.