2 RFC: Common Trace Format (CTF) Proposal (pre-v1.7)
4 Mathieu Desnoyers, EfficiOS Inc.
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.
14 The latest version of this document can be found at:
16 git tree: git://git.efficios.com/ctf.git
17 gitweb: http://git.efficios.com/?p=ctf.git
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:
23 git tree: git://git.efficios.com/babeltrace.git
24 gitweb: http://git.efficios.com/?p=babeltrace.git
27 1. Preliminary definitions
29 - Event Trace: An ordered sequence of events.
30 - Event Stream: An ordered sequence of events, containing a subset of the
32 - Event Packet: A sequence of physically contiguous events within an event
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
37 e.g. event: irq_entry.
38 - An event (or event record) relates to a specific instance of an event
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.
45 2. High-level representation of a trace
47 A trace is divided into multiple event streams. Each event stream contains a
48 subset of the trace event types.
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.
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:
62 - Per-trace event header description.
63 - Per-stream event header description.
64 - Per-stream event context description.
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.
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
82 The event stream header will therefore be referred to as the "event packet
83 header" throughout the rest of this document.
88 Types are organized as type classes. Each type class belong to either of two
89 kind of types: basic types or compound types.
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.
96 4.1.1 Type inheritance
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.
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
110 Each basic type must specify its alignment, in bits. Examples of
111 possible alignments are: bit-packed, byte-packed, or word-aligned. The
112 choice depends on the architecture preference and compactness vs
113 performance trade-offs of the implementation. Architectures providing
114 fast unaligned write byte-packed basic types to save space, aligning
115 each type on byte boundaries (8-bit). Architectures with slow unaligned
116 writes align types on specific alignment values. If no specific
117 alignment is declared for a type, it is assumed to be bit-packed for
118 integers with size not multiple of 8 bits and for gcc bitfields. All
119 other types are byte-packed. It is however recommended to always specify
120 the alignment explicitly.
122 TSDL meta-data attribute representation of a specific alignment:
124 align = value; /* value in bits */
128 By default, the native endianness of the source architecture the trace is used.
129 Byte order can be overridden for a basic type by specifying a "byte_order"
130 attribute. Typical use-case is to specify the network byte order (big endian:
131 "be") to save data captured from the network into the trace without conversion.
132 If not specified, the byte order is native.
134 TSDL meta-data representation:
136 byte_order = native OR network OR be OR le; /* network and be are aliases */
140 Type size, in bits, for integers and floats is that returned by "sizeof()" in C
141 multiplied by CHAR_BIT.
142 We require the size of "char" and "unsigned char" types (CHAR_BIT) to be fixed
143 to 8 bits for cross-endianness compatibility.
145 TSDL meta-data representation:
147 size = value; (value is in bits)
151 Signed integers are represented in two-complement. Integer alignment,
152 size, signedness and byte ordering are defined in the TSDL meta-data.
153 Integers aligned on byte size (8-bit) and with length multiple of byte
154 size (8-bit) correspond to the C99 standard integers. In addition,
155 integers with alignment and/or size that are _not_ a multiple of the
156 byte size are permitted; these correspond to the C99 standard bitfields,
157 with the added specification that the CTF integer bitfields have a fixed
158 binary representation. A MIT-licensed reference implementation of the
159 CTF portable bitfields is available at:
161 http://git.efficios.com/?p=babeltrace.git;a=blob;f=include/babeltrace/bitfield.h
163 Binary representation of integers:
165 - On little and big endian:
166 - Within a byte, high bits correspond to an integer high bits, and low bits
167 correspond to low bits.
169 - Integer across multiple bytes are placed from the less significant to the
171 - Consecutive integers are placed from lower bits to higher bits (even within
174 - Integer across multiple bytes are placed from the most significant to the
176 - Consecutive integers are placed from higher bits to lower bits (even within
179 This binary representation is derived from the bitfield implementation in GCC
180 for little and big endian. However, contrary to what GCC does, integers can
181 cross units boundaries (no padding is required). Padding can be explicitly
182 added (see 4.1.6 GNU/C bitfields) to follow the GCC layout if needed.
184 TSDL meta-data representation:
187 signed = true OR false; /* default false */
188 byte_order = native OR network OR be OR le; /* default native */
189 size = value; /* value in bits, no default */
190 align = value; /* value in bits */
193 Example of type inheritance (creation of a uint32_t named type):
201 Definition of a named 5-bit signed bitfield:
209 4.1.6 GNU/C bitfields
211 The GNU/C bitfields follow closely the integer representation, with a
212 particularity on alignment: if a bitfield cannot fit in the current unit, the
213 unit is padded and the bitfield starts at the following unit. The unit size is
214 defined by the size of the type "unit_type".
216 TSDL meta-data representation:
220 As an example, the following structure declared in C compiled by GCC:
227 The example structure is aligned on the largest element (short). The second
228 bitfield would be aligned on the next unit boundary, because it would not fit in
233 The floating point values byte ordering is defined in the TSDL meta-data.
235 Floating point values follow the IEEE 754-2008 standard interchange formats.
236 Description of the floating point values include the exponent and mantissa size
237 in bits. Some requirements are imposed on the floating point values:
239 - FLT_RADIX must be 2.
240 - mant_dig is the number of digits represented in the mantissa. It is specified
241 by the ISO C99 standard, section 5.2.4, as FLT_MANT_DIG, DBL_MANT_DIG and
242 LDBL_MANT_DIG as defined by <float.h>.
243 - exp_dig is the number of digits represented in the exponent. Given that
244 mant_dig is one bit more than its actual size in bits (leading 1 is not
245 needed) and also given that the sign bit always takes one bit, exp_dig can be
248 - sizeof(float) * CHAR_BIT - FLT_MANT_DIG
249 - sizeof(double) * CHAR_BIT - DBL_MANT_DIG
250 - sizeof(long double) * CHAR_BIT - LDBL_MANT_DIG
252 TSDL meta-data representation:
257 byte_order = native OR network OR be OR le;
260 Example of type inheritance:
262 typealias floating_point {
263 exp_dig = 8; /* sizeof(float) * CHAR_BIT - FLT_MANT_DIG */
264 mant_dig = 24; /* FLT_MANT_DIG */
268 TODO: define NaN, +inf, -inf behavior.
272 Enumerations are a mapping between an integer type and a table of strings. The
273 numerical representation of the enumeration follows the integer type specified
274 by the meta-data. The enumeration mapping table is detailed in the enumeration
275 description within the meta-data. The mapping table maps inclusive value
276 ranges (or single values) to strings. Instead of being limited to simple
277 "value -> string" mappings, these enumerations map
278 "[ start_value ... end_value ] -> string", which map inclusive ranges of
279 values to strings. An enumeration from the C language can be represented in
280 this format by having the same start_value and end_value for each element, which
281 is in fact a range of size 1. This single-value range is supported without
282 repeating the start and end values with the value = string declaration.
284 enum name : integer_type {
285 somestring = start_value1 ... end_value1,
286 "other string" = start_value2 ... end_value2,
287 yet_another_string, /* will be assigned to end_value2 + 1 */
288 "some other string" = value,
292 If the values are omitted, the enumeration starts at 0 and increment of 1 for
295 enum name : unsigned int {
303 Overlapping ranges within a single enumeration are implementation defined.
305 A nameless enumeration can be declared as a field type or as part of a typedef:
307 enum : integer_type {
311 Enumerations omitting the container type ": integer_type" use the "int"
312 type (for compatibility with C99). The "int" type must be previously
315 typealias integer { size = 32; align = 32; signed = true } := int;
324 Compound are aggregation of type declarations. Compound types include
325 structures, variant, arrays, sequences, and strings.
329 Structures are aligned on the largest alignment required by basic types
330 contained within the structure. (This follows the ISO/C standard for structures)
332 TSDL meta-data representation of a named structure:
335 field_type field_name;
336 field_type field_name;
343 integer { /* Nameless type */
348 uint64_t second_field_name; /* Named type declared in the meta-data */
351 The fields are placed in a sequence next to each other. They each possess a
352 field name, which is a unique identifier within the structure.
354 A nameless structure can be declared as a field type or as part of a typedef:
360 4.2.2 Variants (Discriminated/Tagged Unions)
362 A CTF variant is a selection between different types. A CTF variant must
363 always be defined within the scope of a structure or within fields
364 contained within a structure (defined recursively). A "tag" enumeration
365 field must appear in either the same lexical scope, prior to the variant
366 field (in field declaration order), in an uppermost lexical scope (see
367 Section 7.3.1), or in an uppermost dynamic scope (see Section 7.3.2).
368 The type selection is indicated by the mapping from the enumeration
369 value to the string used as variant type selector. The field to use as
370 tag is specified by the "tag_field", specified between "< >" after the
371 "variant" keyword for unnamed variants, and after "variant name" for
374 The alignment of the variant is the alignment of the type as selected by the tag
375 value for the specific instance of the variant. The alignment of the type
376 containing the variant is independent of the variant alignment. The size of the
377 variant is the size as selected by the tag value for the specific instance of
380 A named variant declaration followed by its definition within a structure
391 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
393 variant name <tag_field> v;
396 An unnamed variant definition within a structure is expressed by the following
400 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
402 variant <tag_field> {
410 Example of a named variant within a sequence that refers to a single tag field:
419 enum : uint2_t { a, b, c } choice;
420 variant example <choice> v[unsigned int];
423 Example of an unnamed variant:
426 enum : uint2_t { a, b, c, d } choice;
427 /* Unrelated fields can be added between the variant and its tag */
440 Example of an unnamed variant within an array:
443 enum : uint2_t { a, b, c } choice;
451 Example of a variant type definition within a structure, where the defined type
452 is then declared within an array of structures. This variant refers to a tag
453 located in an upper lexical scope. This example clearly shows that a variant
454 type definition referring to the tag "x" uses the closest preceding field from
455 the lexical scope of the type definition.
458 enum : uint2_t { a, b, c, d } x;
460 typedef variant <x> { /*
461 * "x" refers to the preceding "x" enumeration in the
462 * lexical scope of the type definition.
470 enum : int { x, y, z } x; /* This enumeration is not used by "v". */
471 example_variant v; /*
472 * "v" uses the "enum : uint2_t { a, b, c, d }"
480 Arrays are fixed-length. Their length is declared in the type
481 declaration within the meta-data. They contain an array of "inner type"
482 elements, which can refer to any type not containing the type of the
483 array being declared (no circular dependency). The length is the number
484 of elements in an array.
486 TSDL meta-data representation of a named array:
488 typedef elem_type name[length];
490 A nameless array can be declared as a field type within a structure, e.g.:
492 uint8_t field_name[10];
497 Sequences are dynamically-sized arrays. They start with an integer that specify
498 the length of the sequence, followed by an array of "inner type" elements.
499 The length is the number of elements in the sequence.
501 TSDL meta-data representation for a named sequence:
503 typedef elem_type name[length_type];
505 A nameless sequence can be declared as a field type, e.g.:
507 long field_name[int];
509 The length type follows the integer types specifications, and the sequence
510 elements follow the "array" specifications.
514 Strings are an array of bytes of variable size and are terminated by a '\0'
515 "NULL" character. Their encoding is described in the TSDL meta-data. In
516 absence of encoding attribute information, the default encoding is
519 TSDL meta-data representation of a named string type:
522 encoding = UTF8 OR ASCII;
525 A nameless string type can be declared as a field type:
527 string field_name; /* Use default UTF8 encoding */
529 5. Event Packet Header
531 The event packet header consists of two parts: the "event packet header"
532 is the same for all streams of a trace. The second part, the "event
533 packet context", is described on a per-stream basis. Both are described
534 in the TSDL meta-data. The packets are aligned on architecture-page-sized
537 Event packet header (all fields are optional, specified by TSDL meta-data):
539 - Magic number (CTF magic number: 0xC1FC1FC1) specifies that this is a
540 CTF packet. This magic number is optional, but when present, it should
541 come at the very beginning of the packet.
542 - Trace UUID, used to ensure the event packet match the meta-data used.
543 (note: we cannot use a meta-data checksum in every cases instead of a
544 UUID because meta-data can be appended to while tracing is active)
545 This field is optional.
546 - Stream ID, used as reference to stream description in meta-data.
547 This field is optional if there is only one stream description in the
548 meta-data, but becomes required if there are more than one stream in
549 the TSDL meta-data description.
551 Event packet context (all fields are optional, specified by TSDL meta-data):
553 - Event packet content size (in bytes).
554 - Event packet size (in bytes, includes padding).
555 - Event packet content checksum (optional). Checksum excludes the event packet
557 - Per-stream event packet sequence count (to deal with UDP packet loss). The
558 number of significant sequence counter bits should also be present, so
559 wrap-arounds are dealt with correctly.
560 - Time-stamp at the beginning and time-stamp at the end of the event packet.
561 Both timestamps are written in the packet header, but sampled respectively
562 while (or before) writing the first event and while (or after) writing the
563 last event in the packet. The inclusive range between these timestamps should
564 include all event timestamps assigned to events contained within the packet.
565 - Events discarded count
566 - Snapshot of a per-stream free-running counter, counting the number of
567 events discarded that were supposed to be written in the stream prior to
568 the first event in the event packet.
569 * Note: producer-consumer buffer full condition should fill the current
570 event packet with padding so we know exactly where events have been
572 - Lossless compression scheme used for the event packet content. Applied
573 directly to raw data. New types of compression can be added in following
574 versions of the format.
575 0: no compression scheme
579 - Cypher used for the event packet content. Applied after compression.
582 - Checksum scheme used for the event packet content. Applied after encryption.
588 5.1 Event Packet Header Description
590 The event packet header layout is indicated by the trace packet.header
591 field. Here is a recommended structure type for the packet header with
592 the fields typically expected (although these fields are each optional):
594 struct event_packet_header {
596 uint8_t trace_uuid[16];
602 packet.header := struct event_packet_header;
605 If the magic number is not present, tools such as "file" will have no
606 mean to discover the file type.
608 If the trace_uuid is not present, no validation that the meta-data
609 actually corresponds to the stream is performed.
611 If the stream_id packet header field is missing, the trace can only
612 contain a single stream. Its "id" field can be left out, and its events
613 don't need to declare a "stream_id" field.
616 5.2 Event Packet Context Description
618 Event packet context example. These are declared within the stream declaration
619 in the meta-data. All these fields are optional. If the packet size field is
620 missing, the whole stream only contains a single packet. If the content
621 size field is missing, the packet is filled (no padding). The content
622 and packet sizes include all headers.
624 An example event packet context type:
626 struct event_packet_context {
627 uint64_t timestamp_begin;
628 uint64_t timestamp_end;
630 uint32_t stream_packet_count;
631 uint32_t events_discarded;
633 uint32_t/uint16_t content_size;
634 uint32_t/uint16_t packet_size;
635 uint8_t stream_packet_count_bits; /* Significant counter bits */
636 uint8_t compression_scheme;
637 uint8_t encryption_scheme;
638 uint8_t checksum_scheme;
644 The overall structure of an event is:
646 1 - Stream Packet Context (as specified by the stream meta-data)
647 2 - Event Header (as specified by the stream meta-data)
648 3 - Stream Event Context (as specified by the stream meta-data)
649 4 - Event Context (as specified by the event meta-data)
650 5 - Event Payload (as specified by the event meta-data)
652 This structure defines an implicit dynamic scoping, where variants
653 located in inner structures (those with a higher number in the listing
654 above) can refer to the fields of outer structures (with lower number in
655 the listing above). See Section 7.3 TSDL Scopes for more detail.
659 Event headers can be described within the meta-data. We hereby propose, as an
660 example, two types of events headers. Type 1 accommodates streams with less than
661 31 event IDs. Type 2 accommodates streams with 31 or more event IDs.
663 One major factor can vary between streams: the number of event IDs assigned to
664 a stream. Luckily, this information tends to stay relatively constant (modulo
665 event registration while trace is being recorded), so we can specify different
666 representations for streams containing few event IDs and streams containing
667 many event IDs, so we end up representing the event ID and time-stamp as
668 densely as possible in each case.
670 The header is extended in the rare occasions where the information cannot be
671 represented in the ranges available in the standard event header. They are also
672 used in the rare occasions where the data required for a field could not be
673 collected: the flag corresponding to the missing field within the missing_fields
674 array is then set to 1.
676 Types uintX_t represent an X-bit unsigned integer, as declared with
679 typealias integer { size = X; align = X; signed = false } := uintX_t;
683 typealias integer { size = X; align = 1; signed = false } := uintX_t;
685 6.1.1 Type 1 - Few event IDs
687 - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
689 - Native architecture byte ordering.
690 - For "compact" selection
691 - Fixed size: 32 bits.
692 - For "extended" selection
693 - Size depends on the architecture and variant alignment.
695 struct event_header_1 {
698 * id 31 is reserved to indicate an extended header.
700 enum : uint5_t { compact = 0 ... 30, extended = 31 } id;
706 uint32_t id; /* 32-bit event IDs */
707 uint64_t timestamp; /* 64-bit timestamps */
713 6.1.2 Type 2 - Many event IDs
715 - Aligned on 16-bit (or 8-bit if byte-packed, depending on the architecture
717 - Native architecture byte ordering.
718 - For "compact" selection
719 - Size depends on the architecture and variant alignment.
720 - For "extended" selection
721 - Size depends on the architecture and variant alignment.
723 struct event_header_2 {
725 * id: range: 0 - 65534.
726 * id 65535 is reserved to indicate an extended header.
728 enum : uint16_t { compact = 0 ... 65534, extended = 65535 } id;
734 uint32_t id; /* 32-bit event IDs */
735 uint64_t timestamp; /* 64-bit timestamps */
743 The event context contains information relative to the current event.
744 The choice and meaning of this information is specified by the TSDL
745 stream and event meta-data descriptions. The stream context is applied
746 to all events within the stream. The stream context structure follows
747 the event header. The event context is applied to specific events. Its
748 structure follows the stream context structure.
750 An example of stream-level event context is to save the event payload size with
751 each event, or to save the current PID with each event. These are declared
752 within the stream declaration within the meta-data:
756 event.context := struct {
758 uint16_t payload_size;
762 An example of event-specific event context is to declare a bitmap of missing
763 fields, only appended after the stream event context if the extended event
764 header is selected. NR_FIELDS is the number of fields within the event (a
772 uint1_t missing_fields[NR_FIELDS]; /* missing event fields bitmap */
781 An event payload contains fields specific to a given event type. The fields
782 belonging to an event type are described in the event-specific meta-data
783 within a structure type.
787 No padding at the end of the event payload. This differs from the ISO/C standard
788 for structures, but follows the CTF standard for structures. In a trace, even
789 though it makes sense to align the beginning of a structure, it really makes no
790 sense to add padding at the end of the structure, because structures are usually
791 not followed by a structure of the same type.
793 This trick can be done by adding a zero-length "end" field at the end of the C
794 structures, and by using the offset of this field rather than using sizeof()
795 when calculating the size of a structure (see Appendix "A. Helper macros").
799 The event payload is aligned on the largest alignment required by types
800 contained within the payload. (This follows the ISO/C standard for structures)
803 7. Trace Stream Description Language (TSDL)
805 The Trace Stream Description Language (TSDL) allows expression of the
806 binary trace streams layout in a C99-like Domain Specific Language
812 The trace stream layout description is located in the trace meta-data.
813 The meta-data is itself located in a stream identified by its name:
816 The meta-data description can be expressed in two different formats:
817 text-only and packet-based. The text-only description facilitates
818 generation of meta-data and provides a convenient way to enter the
819 meta-data information by hand. The packet-based meta-data provides the
820 CTF stream packet facilities (checksumming, compression, encryption,
821 network-readiness) for meta-data stream generated and transported by a
824 The text-only meta-data file is a plain text TSDL description.
826 The packet-based meta-data is made of "meta-data packets", which each
827 start with a meta-data packet header. The packet-based meta-data
828 description is detected by reading the magic number "0x75D11D57" at the
829 beginning of the file. This magic number is also used to detect the
830 endianness of the architecture by trying to read the CTF magic number
831 and its counterpart in reversed endianness. The events within the
832 meta-data stream have no event header nor event context. Each event only
833 contains a "string" payload. Each meta-data packet start with a special
834 packet header, specific to the meta-data stream, which contains,
837 struct metadata_packet_header {
838 uint32_t magic; /* 0x75D11D57 */
839 uint8_t trace_uuid[16]; /* Unique Universal Identifier */
840 uint32_t checksum; /* 0 if unused */
841 uint32_t content_size; /* in bits */
842 uint32_t packet_size; /* in bits */
843 uint8_t compression_scheme; /* 0 if unused */
844 uint8_t encryption_scheme; /* 0 if unused */
845 uint8_t checksum_scheme; /* 0 if unused */
848 The packet-based meta-data can be converted to a text-only meta-data by
849 concatenating all the strings in contains.
851 In the textual representation of the meta-data, the text contained
852 within "/*" and "*/", as well as within "//" and end of line, are
853 treated as comments. Boolean values can be represented as true, TRUE,
854 or 1 for true, and false, FALSE, or 0 for false. Within the string-based
855 meta-data description, the trace UUID is represented as a string of
856 hexadecimal digits and dashes "-". In the event packet header, the trace
857 UUID is represented as an array of bytes.
860 7.2 Declaration vs Definition
862 A declaration associates a layout to a type, without specifying where
863 this type is located in the event structure hierarchy (see Section 6).
864 This therefore includes typedef, typealias, as well as all type
865 specifiers. In certain circumstances (typedef, structure field and
866 variant field), a declaration is followed by a declarator, which specify
867 the newly defined type name (for typedef), or the field name (for
868 declarations located within structure and variants). Array and sequence,
869 declared with square brackets ("[" "]"), are part of the declarator,
870 similarly to C99. The enumeration base type is specified by
871 ": enum_base", which is part of the type specifier. The variant tag
872 name, specified between "<" ">", is also part of the type specifier.
874 A definition associates a type to a location in the event structure
875 hierarchy (see Section 6). This association is denoted by ":=", as shown
881 TSDL uses two different types of scoping: a lexical scope is used for
882 declarations and type definitions, and a dynamic scope is used for
883 variants references to tag fields.
887 Each of "trace", "stream", "event", "struct" and "variant" have their own
888 nestable declaration scope, within which types can be declared using "typedef"
889 and "typealias". A root declaration scope also contains all declarations
890 located outside of any of the aforementioned declarations. An inner
891 declaration scope can refer to type declared within its container
892 lexical scope prior to the inner declaration scope. Redefinition of a
893 typedef or typealias is not valid, although hiding an upper scope
894 typedef or typealias is allowed within a sub-scope.
898 A dynamic scope consists in the lexical scope augmented with the
899 implicit event structure definition hierarchy presented at Section 6.
900 The dynamic scope is only used for variant tag definitions. It is used
901 at definition time to look up the location of the tag field associated
904 Therefore, variants in lower levels in the dynamic scope (e.g. event
905 context) can refer to a tag field located in upper levels (e.g. in the
906 event header) by specifying, in this case, the associated tag with
907 <header.field_name>. This allows, for instance, the event context to
908 define a variant referring to the "id" field of the event header as
911 The target dynamic scope must be specified explicitly when referring to
912 a field outside of the local static scope. The dynamic scope prefixes
915 - Trace Packet Header: <trace.packet.header. >,
916 - Stream Packet Context: <stream.packet.context. >,
917 - Event Header: <stream.event.header. >,
918 - Stream Event Context: <stream.event.context. >,
919 - Event Context: <event.context. >,
920 - Event Payload: <event.fields. >.
922 Multiple declarations of the same field name within a single scope is
923 not valid. It is however valid to re-use the same field name in
924 different scopes. There is no possible conflict, because the dynamic
925 scope must be specified when a variant refers to a tag field located in
926 a different dynamic scope.
928 The information available in the dynamic scopes can be thought of as the
929 current tracing context. At trace production, information about the
930 current context is saved into the specified scope field levels. At trace
931 consumption, for each event, the current trace context is therefore
932 readable by accessing the upper dynamic scopes.
937 The grammar representing the TSDL meta-data is presented in Appendix C.
938 TSDL Grammar. This section presents a rather lighter reading that
939 consists in examples of TSDL meta-data, with template values.
941 The stream "id" can be left out if there is only one stream in the
942 trace. The event "id" field can be left out if there is only one event
946 major = value; /* Trace format version */
948 uuid = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"; /* Trace UUID */
949 byte_order = be OR le; /* Endianness (required) */
950 packet.header := struct {
952 uint8_t trace_uuid[16];
959 /* Type 1 - Few event IDs; Type 2 - Many event IDs. See section 6.1. */
960 event.header := event_header_1 OR event_header_2;
961 event.context := struct {
964 packet.context := struct {
971 id = value; /* Numeric identifier within the stream */
981 /* More detail on types in section 4. Types */
986 * Type declarations behave similarly to the C standard.
989 typedef aliased_type_specifiers new_type_declarators;
991 /* e.g.: typedef struct example new_type_name[10]; */
996 * The "typealias" declaration can be used to give a name (including
997 * pointer declarator specifier) to a type. It should also be used to
998 * map basic C types (float, int, unsigned long, ...) to a CTF type.
999 * Typealias is a superset of "typedef": it also allows assignment of a
1000 * simple variable identifier to a type.
1003 typealias type_class {
1005 } := type_specifiers type_declarator;
1009 * typealias integer {
1013 * } := struct page *;
1015 * typealias integer {
1030 enum name : integer_type {
1036 * Unnamed types, contained within compound type fields, typedef or typealias.
1047 enum : integer_type {
1051 typedef type new_type[length];
1054 type field_name[length];
1057 typedef type new_type[length_type];
1060 type field_name[length_type];
1072 integer_type field_name:size; /* GNU/C bitfield */
1082 The two following macros keep track of the size of a GNU/C structure without
1083 padding at the end by placing HEADER_END as the last field. A one byte end field
1084 is used for C90 compatibility (C99 flexible arrays could be used here). Note
1085 that this does not affect the effective structure size, which should always be
1086 calculated with the header_sizeof() helper.
1088 #define HEADER_END char end_field
1089 #define header_sizeof(type) offsetof(typeof(type), end_field)
1092 B. Stream Header Rationale
1094 An event stream is divided in contiguous event packets of variable size. These
1095 subdivisions allow the trace analyzer to perform a fast binary search by time
1096 within the stream (typically requiring to index only the event packet headers)
1097 without reading the whole stream. These subdivisions have a variable size to
1098 eliminate the need to transfer the event packet padding when partially filled
1099 event packets must be sent when streaming a trace for live viewing/analysis.
1100 An event packet can contain a certain amount of padding at the end. Dividing
1101 streams into event packets is also useful for network streaming over UDP and
1102 flight recorder mode tracing (a whole event packet can be swapped out of the
1103 buffer atomically for reading).
1105 The stream header is repeated at the beginning of each event packet to allow
1106 flexibility in terms of:
1108 - streaming support,
1109 - allowing arbitrary buffers to be discarded without making the trace
1111 - allow UDP packet loss handling by either dealing with missing event packet
1112 or asking for re-transmission.
1113 - transparently support flight recorder mode,
1114 - transparently support crash dump.
1120 * Common Trace Format (CTF) Trace Stream Description Language (TSDL) Grammar.
1122 * Inspired from the C99 grammar:
1123 * http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf (Annex A)
1124 * and c++1x grammar (draft)
1125 * http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3291.pdf (Annex A)
1127 * Specialized for CTF needs by including only constant and declarations from
1128 * C99 (excluding function declarations), and by adding support for variants,
1129 * sequences and CTF-specific specifiers. Enumeration container types
1130 * semantic is inspired from c++1x enum-base.
1135 1.1) Lexical elements
1178 identifier identifier-nondigit
1181 identifier-nondigit:
1183 universal-character-name
1184 any other implementation-defined characters
1188 [a-zA-Z] /* regular expression */
1191 [0-9] /* regular expression */
1193 1.4) Universal character names
1195 universal-character-name:
1197 \U hex-quad hex-quad
1200 hexadecimal-digit hexadecimal-digit hexadecimal-digit hexadecimal-digit
1206 enumeration-constant
1210 decimal-constant integer-suffix-opt
1211 octal-constant integer-suffix-opt
1212 hexadecimal-constant integer-suffix-opt
1216 decimal-constant digit
1220 octal-constant octal-digit
1222 hexadecimal-constant:
1223 hexadecimal-prefix hexadecimal-digit
1224 hexadecimal-constant hexadecimal-digit
1234 unsigned-suffix long-suffix-opt
1235 unsigned-suffix long-long-suffix
1236 long-suffix unsigned-suffix-opt
1237 long-long-suffix unsigned-suffix-opt
1251 hexadecimal-digit-sequence:
1253 hexadecimal-digit-sequence hexadecimal-digit
1255 enumeration-constant:
1261 L' c-char-sequence '
1265 c-char-sequence c-char
1268 any member of source charset except single-quote ('), backslash
1269 (\), or new-line character.
1273 simple-escape-sequence
1274 octal-escape-sequence
1275 hexadecimal-escape-sequence
1276 universal-character-name
1278 simple-escape-sequence: one of
1279 \' \" \? \\ \a \b \f \n \r \t \v
1281 octal-escape-sequence:
1283 \ octal-digit octal-digit
1284 \ octal-digit octal-digit octal-digit
1286 hexadecimal-escape-sequence:
1287 \x hexadecimal-digit
1288 hexadecimal-escape-sequence hexadecimal-digit
1290 1.6) String literals
1293 " s-char-sequence-opt "
1294 L" s-char-sequence-opt "
1298 s-char-sequence s-char
1301 any member of source charset except double-quote ("), backslash
1302 (\), or new-line character.
1308 [ ] ( ) { } . -> * + - < > : ; ... = ,
1311 2) Phrase structure grammar
1317 ( unary-expression )
1321 postfix-expression [ unary-expression ]
1322 postfix-expression . identifier
1323 postfix-expressoin -> identifier
1327 unary-operator postfix-expression
1329 unary-operator: one of
1332 assignment-operator:
1335 type-assignment-operator:
1338 constant-expression:
1341 constant-expression-range:
1342 constant-expression ... constant-expression
1347 declaration-specifiers declarator-list-opt ;
1350 declaration-specifiers:
1351 storage-class-specifier declaration-specifiers-opt
1352 type-specifier declaration-specifiers-opt
1353 type-qualifier declaration-specifiers-opt
1357 declarator-list , declarator
1359 abstract-declarator-list:
1361 abstract-declarator-list , abstract-declarator
1363 storage-class-specifier:
1386 struct identifier-opt { struct-or-variant-declaration-list-opt }
1389 struct-or-variant-declaration-list:
1390 struct-or-variant-declaration
1391 struct-or-variant-declaration-list struct-or-variant-declaration
1393 struct-or-variant-declaration:
1394 specifier-qualifier-list struct-or-variant-declarator-list ;
1395 declaration-specifiers storage-class-specifier declaration-specifiers declarator-list ;
1396 typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list ;
1397 typealias declaration-specifiers abstract-declarator-list := declarator-list ;
1399 specifier-qualifier-list:
1400 type-specifier specifier-qualifier-list-opt
1401 type-qualifier specifier-qualifier-list-opt
1403 struct-or-variant-declarator-list:
1404 struct-or-variant-declarator
1405 struct-or-variant-declarator-list , struct-or-variant-declarator
1407 struct-or-variant-declarator:
1409 declarator-opt : constant-expression
1412 variant identifier-opt variant-tag-opt { struct-or-variant-declaration-list }
1413 variant identifier variant-tag
1419 enum identifier-opt { enumerator-list }
1420 enum identifier-opt { enumerator-list , }
1422 enum identifier-opt : declaration-specifiers { enumerator-list }
1423 enum identifier-opt : declaration-specifiers { enumerator-list , }
1427 enumerator-list , enumerator
1430 enumeration-constant
1431 enumeration-constant = constant-expression
1432 enumeration-constant = constant-expression-range
1438 pointer-opt direct-declarator
1443 direct-declarator [ type-specifier ]
1444 direct-declarator [ constant-expression ]
1446 abstract-declarator:
1447 pointer-opt direct-abstract-declarator
1449 direct-abstract-declarator:
1451 ( abstract-declarator )
1452 direct-abstract-declarator [ type-specifier ]
1453 direct-abstract-declarator [ constant-expression ]
1454 direct-abstract-declarator [ ]
1457 * type-qualifier-list-opt
1458 * type-qualifier-list-opt pointer
1460 type-qualifier-list:
1462 type-qualifier-list type-qualifier
1467 2.3) CTF-specific declarations
1470 event { ctf-assignment-expression-list-opt }
1471 stream { ctf-assignment-expression-list-opt }
1472 trace { ctf-assignment-expression-list-opt }
1473 typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list ;
1474 typealias declaration-specifiers abstract-declarator-list := declarator-list ;
1477 floating_point { ctf-assignment-expression-list-opt }
1478 integer { ctf-assignment-expression-list-opt }
1479 string { ctf-assignment-expression-list-opt }
1481 ctf-assignment-expression-list:
1482 ctf-assignment-expression
1483 ctf-assignment-expression-list ; ctf-assignment-expression
1485 ctf-assignment-expression:
1486 unary-expression assignment-operator unary-expression
1487 unary-expression type-assignment-operator type-specifier
1488 declaration-specifiers storage-class-specifier declaration-specifiers declarator-list
1489 typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list
1490 typealias declaration-specifiers abstract-declarator-list := declarator-list