1 Common Trace Format (CTF) Specification (v1.8.2)
3 Mathieu Desnoyers, EfficiOS Inc.
5 The goal of the present document is to specify a trace format that suits the
6 needs of the embedded, telecom, high-performance and kernel communities. It is
7 based on the Common Trace Format Requirements (v1.4) document. It is designed to
8 allow traces to be natively generated by the Linux kernel, Linux user-space
9 applications written in C/C++, and hardware components. One major element of
10 CTF is the Trace Stream Description Language (TSDL) which flexibility
11 enables description of various binary trace stream layouts.
13 The latest version of this document can be found at:
15 git tree: git://git.efficios.com/ctf.git
16 gitweb: http://git.efficios.com/?p=ctf.git
18 A reference implementation of a library to read and write this trace format is
19 being implemented within the BabelTrace project, a converter between trace
20 formats. The development tree is available at:
22 git tree: git://git.efficios.com/babeltrace.git
23 gitweb: http://git.efficios.com/?p=babeltrace.git
25 The CE Workgroup of the Linux Foundation, Ericsson, and EfficiOS have
31 1. Preliminary definitions
32 2. High-level representation of a trace
36 4.1.1 Type inheritance
46 4.2.2 Variants (Discriminated/Tagged Unions)
50 5. Event Packet Header
51 5.1 Event Packet Header Description
52 5.2 Event Packet Context Description
55 6.1.1 Type 1 - Few event IDs
56 6.1.2 Type 2 - Many event IDs
57 6.2 Stream Event Context and Event Context
61 7. Trace Stream Description Language (TSDL)
63 7.2 Declaration vs Definition
66 7.3.2 Static and Dynamic Scopes
71 1. Preliminary definitions
73 - Event Trace: An ordered sequence of events.
74 - Event Stream: An ordered sequence of events, containing a subset of the
76 - Event Packet: A sequence of physically contiguous events within an event
78 - Event: This is the basic entry in a trace. (aka: a trace record).
79 - An event identifier (ID) relates to the class (a type) of event within
81 e.g. event: irq_entry.
82 - An event (or event record) relates to a specific instance of an event
84 e.g. event: irq_entry, at time X, on CPU Y
85 - Source Architecture: Architecture writing the trace.
86 - Reader Architecture: Architecture reading the trace.
89 2. High-level representation of a trace
91 A trace is divided into multiple event streams. Each event stream contains a
92 subset of the trace event types.
94 The final output of the trace, after its generation and optional transport over
95 the network, is expected to be either on permanent or temporary storage in a
96 virtual file system. Because each event stream is appended to while a trace is
97 being recorded, each is associated with a distinct set of files for
98 output. Therefore, a stored trace can be represented as a directory
99 containing zero, one or more files per stream.
101 Meta-data description associated with the trace contains information on
102 trace event types expressed in the Trace Stream Description Language
103 (TSDL). This language describes:
107 - Per-trace event header description.
108 - Per-stream event header description.
109 - Per-stream event context description.
111 - Event type to stream mapping.
112 - Event type to name mapping.
113 - Event type to ID mapping.
114 - Event context description.
115 - Event fields description.
120 An event stream can be divided into contiguous event packets of variable
121 size. An event packet can contain a certain amount of padding at the
122 end. The stream header is repeated at the beginning of each event
123 packet. The rationale for the event stream design choices is explained
124 in Appendix B. Stream Header Rationale.
126 The event stream header will therefore be referred to as the "event packet
127 header" throughout the rest of this document.
132 Types are organized as type classes. Each type class belong to either of two
133 kind of types: basic types or compound types.
137 A basic type is a scalar type, as described in this section. It includes
138 integers, GNU/C bitfields, enumerations, and floating point values.
140 4.1.1 Type inheritance
142 Type specifications can be inherited to allow deriving types from a
143 type class. For example, see the uint32_t named type derived from the "integer"
144 type class below ("Integers" section). Types have a precise binary
145 representation in the trace. A type class has methods to read and write these
146 types, but must be derived into a type to be usable in an event field.
150 We define "byte-packed" types as aligned on the byte size, namely 8-bit.
151 We define "bit-packed" types as following on the next bit, as defined by the
154 Each basic type must specify its alignment, in bits. Examples of
155 possible alignments are: bit-packed (align = 1), byte-packed (align =
156 8), or word-aligned (e.g. align = 32 or align = 64). The choice depends
157 on the architecture preference and compactness vs performance trade-offs
158 of the implementation. Architectures providing fast unaligned write
159 byte-packed basic types to save space, aligning each type on byte
160 boundaries (8-bit). Architectures with slow unaligned writes align types
161 on specific alignment values. If no specific alignment is declared for a
162 type, it is assumed to be bit-packed for integers with size not multiple
163 of 8 bits and for gcc bitfields. All other basic types are byte-packed
164 by default. It is however recommended to always specify the alignment
165 explicitly. Alignment values must be power of two. Compound types are
166 aligned as specified in their individual specification.
168 The base offset used for field alignment is the start of the packet
169 containing the field. For instance, a field aligned on 32-bit needs to
170 be at an offset multiple of 32-bit from the start of the packet that
173 TSDL meta-data attribute representation of a specific alignment:
175 align = value; /* value in bits */
179 By default, byte order of a basic type is the byte order described in
180 the trace description. It can be overridden by specifying a
181 "byte_order" attribute for a basic type. Typical use-case is to specify
182 the network byte order (big endian: "be") to save data captured from the
183 network into the trace without conversion.
185 TSDL meta-data representation:
187 byte_order = native OR network OR be OR le; /* network and be are aliases */
189 The "native" keyword selects the byte order described in the trace
190 description. The "network" byte order is an alias for big endian.
192 Even though the trace description section is not per se a type, for sake
193 of clarity, it should be noted that "native" and "network" byte orders
194 are only allowed within type declaration. The byte_order specified in
195 the trace description section only accepts "be" or "le" values.
199 Type size, in bits, for integers and floats is that returned by "sizeof()" in C
200 multiplied by CHAR_BIT.
201 We require the size of "char" and "unsigned char" types (CHAR_BIT) to be fixed
202 to 8 bits for cross-endianness compatibility.
204 TSDL meta-data representation:
206 size = value; (value is in bits)
210 Signed integers are represented in two-complement. Integer alignment,
211 size, signedness and byte ordering are defined in the TSDL meta-data.
212 Integers aligned on byte size (8-bit) and with length multiple of byte
213 size (8-bit) correspond to the C99 standard integers. In addition,
214 integers with alignment and/or size that are _not_ a multiple of the
215 byte size are permitted; these correspond to the C99 standard bitfields,
216 with the added specification that the CTF integer bitfields have a fixed
217 binary representation. Integer size needs to be a positive integer.
218 Integers of size 0 are forbidden. A MIT-licensed reference
219 implementation of the CTF portable bitfields is available at:
221 http://git.efficios.com/?p=babeltrace.git;a=blob;f=include/babeltrace/bitfield.h
223 Binary representation of integers:
225 - On little and big endian:
226 - Within a byte, high bits correspond to an integer high bits, and low bits
227 correspond to low bits.
229 - Integer across multiple bytes are placed from the less significant to the
231 - Consecutive integers are placed from lower bits to higher bits (even within
234 - Integer across multiple bytes are placed from the most significant to the
236 - Consecutive integers are placed from higher bits to lower bits (even within
239 This binary representation is derived from the bitfield implementation in GCC
240 for little and big endian. However, contrary to what GCC does, integers can
241 cross units boundaries (no padding is required). Padding can be explicitly
242 added (see 4.1.6 GNU/C bitfields) to follow the GCC layout if needed.
244 TSDL meta-data representation:
247 signed = true OR false; /* default false */
248 byte_order = native OR network OR be OR le; /* default native */
249 size = value; /* value in bits, no default */
250 align = value; /* value in bits */
251 /* based used for pretty-printing output, default: decimal. */
252 base = decimal OR dec OR d OR i OR u OR 10 OR hexadecimal OR hex OR x OR X OR p OR 16
253 OR octal OR oct OR o OR 8 OR binary OR b OR 2;
254 /* character encoding, default: none */
255 encoding = none or UTF8 or ASCII;
258 Example of type inheritance (creation of a uint32_t named type):
266 Definition of a named 5-bit signed bitfield:
274 The character encoding field can be used to specify that the integer
275 must be printed as a text character when read. e.g.:
285 4.1.6 GNU/C bitfields
287 The GNU/C bitfields follow closely the integer representation, with a
288 particularity on alignment: if a bitfield cannot fit in the current unit, the
289 unit is padded and the bitfield starts at the following unit. The unit size is
290 defined by the size of the type "unit_type".
292 TSDL meta-data representation:
296 As an example, the following structure declared in C compiled by GCC:
303 The example structure is aligned on the largest element (short). The second
304 bitfield would be aligned on the next unit boundary, because it would not fit in
309 The floating point values byte ordering is defined in the TSDL meta-data.
311 Floating point values follow the IEEE 754-2008 standard interchange formats.
312 Description of the floating point values include the exponent and mantissa size
313 in bits. Some requirements are imposed on the floating point values:
315 - FLT_RADIX must be 2.
316 - mant_dig is the number of digits represented in the mantissa. It is specified
317 by the ISO C99 standard, section 5.2.4, as FLT_MANT_DIG, DBL_MANT_DIG and
318 LDBL_MANT_DIG as defined by <float.h>.
319 - exp_dig is the number of digits represented in the exponent. Given that
320 mant_dig is one bit more than its actual size in bits (leading 1 is not
321 needed) and also given that the sign bit always takes one bit, exp_dig can be
324 - sizeof(float) * CHAR_BIT - FLT_MANT_DIG
325 - sizeof(double) * CHAR_BIT - DBL_MANT_DIG
326 - sizeof(long double) * CHAR_BIT - LDBL_MANT_DIG
328 TSDL meta-data representation:
333 byte_order = native OR network OR be OR le;
337 Example of type inheritance:
339 typealias floating_point {
340 exp_dig = 8; /* sizeof(float) * CHAR_BIT - FLT_MANT_DIG */
341 mant_dig = 24; /* FLT_MANT_DIG */
346 TODO: define NaN, +inf, -inf behavior.
348 Bit-packed, byte-packed or larger alignments can be used for floating
349 point values, similarly to integers.
353 Enumerations are a mapping between an integer type and a table of strings. The
354 numerical representation of the enumeration follows the integer type specified
355 by the meta-data. The enumeration mapping table is detailed in the enumeration
356 description within the meta-data. The mapping table maps inclusive value
357 ranges (or single values) to strings. Instead of being limited to simple
358 "value -> string" mappings, these enumerations map
359 "[ start_value ... end_value ] -> string", which map inclusive ranges of
360 values to strings. An enumeration from the C language can be represented in
361 this format by having the same start_value and end_value for each
362 mapping, which is in fact a range of size 1. This single-value range is
363 supported without repeating the start and end values with the value =
364 string declaration. Enumerations need to contain at least one entry.
366 enum name : integer_type {
367 somestring = start_value1 ... end_value1,
368 "other string" = start_value2 ... end_value2,
369 yet_another_string, /* will be assigned to end_value2 + 1 */
370 "some other string" = value,
374 If the values are omitted, the enumeration starts at 0 and increment of 1 for
375 each entry. An entry with omitted value that follows a range entry takes
376 as value the end_value of the previous range + 1:
378 enum name : unsigned int {
386 Overlapping ranges within a single enumeration are implementation defined.
388 A nameless enumeration can be declared as a field type or as part of a typedef:
390 enum : integer_type {
394 Enumerations omitting the container type ": integer_type" use the "int"
395 type (for compatibility with C99). The "int" type must be previously
398 typealias integer { size = 32; align = 32; signed = true; } := int;
407 Compound are aggregation of type declarations. Compound types include
408 structures, variant, arrays, sequences, and strings.
412 Structures are aligned on the largest alignment required by basic types
413 contained within the structure. (This follows the ISO/C standard for structures)
415 TSDL meta-data representation of a named structure:
418 field_type field_name;
419 field_type field_name;
426 integer { /* Nameless type */
431 uint64_t second_field_name; /* Named type declared in the meta-data */
434 The fields are placed in a sequence next to each other. They each
435 possess a field name, which is a unique identifier within the structure.
436 The identifier is not allowed to use any reserved keyword
437 (see Section C.1.2). Replacing reserved keywords with
438 underscore-prefixed field names is recommended. Fields starting with an
439 underscore should have their leading underscore removed by the CTF trace
442 A nameless structure can be declared as a field type or as part of a typedef:
448 Alignment for a structure compound type can be forced to a minimum value
449 by adding an "align" specifier after the declaration of a structure
450 body. This attribute is read as: align(value). The value is specified in
451 bits. The structure will be aligned on the maximum value between this
452 attribute and the alignment required by the basic types contained within
459 4.2.2 Variants (Discriminated/Tagged Unions)
461 A CTF variant is a selection between different types. A CTF variant must
462 always be defined within the scope of a structure or within fields
463 contained within a structure (defined recursively). A "tag" enumeration
464 field must appear in either the same static scope, prior to the variant
465 field (in field declaration order), in an upper static scope, or in an
466 upper dynamic scope (see Section 7.3.2). The type selection is indicated
467 by the mapping from the enumeration value to the string used as variant
468 type selector. The field to use as tag is specified by the "tag_field",
469 specified between "< >" after the "variant" keyword for unnamed
470 variants, and after "variant name" for named variants. It is not
471 required that each enumeration mapping appears as variant type tag field.
472 However, it is required that any enumeration mapping encountered within a
473 stream has a matching variant type tag field.
475 The alignment of the variant is the alignment of the type as selected by
476 the tag value for the specific instance of the variant. The size of the
477 variant is the size as selected by the tag value for the specific
478 instance of the variant.
480 The alignment of the type containing the variant is independent of the
481 variant alignment. For instance, if a structure contains two fields, a
482 32-bit integer, aligned on 32 bits, and a variant, which contains two
483 choices: either a 32-bit field, aligned on 32 bits, or a 64-bit field,
484 aligned on 64 bits, the alignment of the outmost structure will be
485 32-bit (the alignment of its largest field, disregarding the alignment
486 of the variant). The alignment of the variant will depend on the
487 selector: if the variant's 32-bit field is selected, its alignment will
488 be 32-bit, or 64-bit otherwise. It is important to note that variants
489 are specifically tailored for compactness in a stream. Therefore, the
490 relative offsets of compound type fields can vary depending on
491 the offset at which the compound type starts if it contains a variant
492 that itself contains a type with alignment larger than the largest field
493 contained within the compound type. This is caused by the fact that the
494 compound type may contain the enumeration that select the variant's
495 choice, and therefore the alignment to be applied to the compound type
496 cannot be determined before encountering the enumeration.
498 Each variant type selector possess a field name, which is a unique
499 identifier within the variant. The identifier is not allowed to use any
500 reserved keyword (see Section C.1.2). Replacing reserved keywords with
501 underscore-prefixed field names is recommended. Fields starting with an
502 underscore should have their leading underscore removed by the CTF trace
506 A named variant declaration followed by its definition within a structure
517 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
519 variant name <tag_field> v;
522 An unnamed variant definition within a structure is expressed by the following
526 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
528 variant <tag_field> {
536 Example of a named variant within a sequence that refers to a single tag field:
545 enum : uint2_t { a, b, c } choice;
547 variant example <choice> v[seqlen];
550 Example of an unnamed variant:
553 enum : uint2_t { a, b, c, d } choice;
554 /* Unrelated fields can be added between the variant and its tag */
567 Example of an unnamed variant within an array:
570 enum : uint2_t { a, b, c } choice;
578 Example of a variant type definition within a structure, where the defined type
579 is then declared within an array of structures. This variant refers to a tag
580 located in an upper static scope. This example clearly shows that a variant
581 type definition referring to the tag "x" uses the closest preceding field from
582 the static scope of the type definition.
585 enum : uint2_t { a, b, c, d } x;
587 typedef variant <x> { /*
588 * "x" refers to the preceding "x" enumeration in the
589 * static scope of the type definition.
597 enum : int { x, y, z } x; /* This enumeration is not used by "v". */
598 example_variant v; /*
599 * "v" uses the "enum : uint2_t { a, b, c, d }"
607 Arrays are fixed-length. Their length is declared in the type
608 declaration within the meta-data. They contain an array of "inner type"
609 elements, which can refer to any type not containing the type of the
610 array being declared (no circular dependency). The length is the number
611 of elements in an array.
613 TSDL meta-data representation of a named array:
615 typedef elem_type name[length];
617 A nameless array can be declared as a field type within a structure, e.g.:
619 uint8_t field_name[10];
621 Arrays are always aligned on their element alignment requirement.
625 Sequences are dynamically-sized arrays. They refer to a "length"
626 unsigned integer field, which must appear in either the same static scope,
627 prior to the sequence field (in field declaration order), in an upper
628 static scope, or in an upper dynamic scope (see Section 7.3.2). This
629 length field represents the number of elements in the sequence. The
630 sequence per se is an array of "inner type" elements.
632 TSDL meta-data representation for a sequence type definition:
635 unsigned int length_field;
636 typedef elem_type typename[length_field];
637 typename seq_field_name;
640 A sequence can also be declared as a field type, e.g.:
643 unsigned int length_field;
644 long seq_field_name[length_field];
647 Multiple sequences can refer to the same length field, and these length
648 fields can be in a different upper dynamic scope:
650 e.g., assuming the stream.event.header defines:
655 event.header := struct {
664 long seq_a[stream.event.header.seq_len];
665 char seq_b[stream.event.header.seq_len];
669 The sequence elements follow the "array" specifications.
673 Strings are an array of bytes of variable size and are terminated by a '\0'
674 "NULL" character. Their encoding is described in the TSDL meta-data. In
675 absence of encoding attribute information, the default encoding is
678 TSDL meta-data representation of a named string type:
681 encoding = UTF8 OR ASCII;
684 A nameless string type can be declared as a field type:
686 string field_name; /* Use default UTF8 encoding */
688 Strings are always aligned on byte size.
690 5. Event Packet Header
692 The event packet header consists of two parts: the "event packet header"
693 is the same for all streams of a trace. The second part, the "event
694 packet context", is described on a per-stream basis. Both are described
695 in the TSDL meta-data.
697 Event packet header (all fields are optional, specified by TSDL meta-data):
699 - Magic number (CTF magic number: 0xC1FC1FC1) specifies that this is a
700 CTF packet. This magic number is optional, but when present, it should
701 come at the very beginning of the packet.
702 - Trace UUID, used to ensure the event packet match the meta-data used.
703 (note: we cannot use a meta-data checksum in every cases instead of a
704 UUID because meta-data can be appended to while tracing is active)
705 This field is optional.
706 - Stream ID, used as reference to stream description in meta-data.
707 This field is optional if there is only one stream description in the
708 meta-data, but becomes required if there are more than one stream in
709 the TSDL meta-data description.
711 Event packet context (all fields are optional, specified by TSDL meta-data):
713 - Event packet content size (in bits).
714 - Event packet size (in bits, includes padding).
715 - Event packet content checksum. Checksum excludes the event packet
717 - Per-stream event packet sequence count (to deal with UDP packet loss). The
718 number of significant sequence counter bits should also be present, so
719 wrap-arounds are dealt with correctly.
720 - Time-stamp at the beginning and time-stamp at the end of the event packet.
721 Both timestamps are written in the packet header, but sampled respectively
722 while (or before) writing the first event and while (or after) writing the
723 last event in the packet. The inclusive range between these timestamps should
724 include all event timestamps assigned to events contained within the packet.
725 The timestamp at the beginning of an event packet is guaranteed to be
726 below or equal the timestamp at the end of that event packet.
727 The timestamp at the end of an event packet is guaranteed to be below
728 or equal the timestamps at the end of any following packet within the
729 same stream. See Section 8. Clocks for more detail.
730 - Events discarded count
731 - Snapshot of a per-stream free-running counter, counting the number of
732 events discarded that were supposed to be written in the stream after
733 the last event in the event packet.
734 * Note: producer-consumer buffer full condition can fill the current
735 event packet with padding so we know exactly where events have been
736 discarded. However, if the buffer full condition chooses not
737 to fill the current event packet with padding, all we know
738 about the timestamp range in which the events have been
739 discarded is that it is somewhere between the beginning and
740 the end of the packet.
741 - Lossless compression scheme used for the event packet content. Applied
742 directly to raw data. New types of compression can be added in following
743 versions of the format.
744 0: no compression scheme
748 - Cypher used for the event packet content. Applied after compression.
751 - Checksum scheme used for the event packet content. Applied after encryption.
757 5.1 Event Packet Header Description
759 The event packet header layout is indicated by the trace packet.header
760 field. Here is a recommended structure type for the packet header with
761 the fields typically expected (although these fields are each optional):
763 struct event_packet_header {
771 packet.header := struct event_packet_header;
774 If the magic number is not present, tools such as "file" will have no
775 mean to discover the file type.
777 If the uuid is not present, no validation that the meta-data actually
778 corresponds to the stream is performed.
780 If the stream_id packet header field is missing, the trace can only
781 contain a single stream. Its "id" field can be left out, and its events
782 don't need to declare a "stream_id" field.
785 5.2 Event Packet Context Description
787 Event packet context example. These are declared within the stream declaration
788 in the meta-data. All these fields are optional. If the packet size field is
789 missing, the whole stream only contains a single packet. If the content
790 size field is missing, the packet is filled (no padding). The content
791 and packet sizes include all headers.
793 An example event packet context type:
795 struct event_packet_context {
796 uint64_t timestamp_begin;
797 uint64_t timestamp_end;
799 uint32_t stream_packet_count;
800 uint32_t events_discarded;
802 uint64_t/uint32_t/uint16_t content_size;
803 uint64_t/uint32_t/uint16_t packet_size;
804 uint8_t compression_scheme;
805 uint8_t encryption_scheme;
806 uint8_t checksum_scheme;
812 The overall structure of an event is:
814 1 - Event Header (as specified by the stream meta-data)
815 2 - Stream Event Context (as specified by the stream meta-data)
816 3 - Event Context (as specified by the event meta-data)
817 4 - Event Payload (as specified by the event meta-data)
819 This structure defines an implicit dynamic scoping, where variants
820 located in inner structures (those with a higher number in the listing
821 above) can refer to the fields of outer structures (with lower number in
822 the listing above). See Section 7.3 TSDL Scopes for more detail.
824 The total length of an event is defined as the difference between the
825 end of its Event Payload and the end of the previous event's Event
826 Payload. Therefore, it includes the event header alignment padding, and
827 all its fields and their respective alignment padding. Events of length
832 Event headers can be described within the meta-data. We hereby propose, as an
833 example, two types of events headers. Type 1 accommodates streams with less than
834 31 event IDs. Type 2 accommodates streams with 31 or more event IDs.
836 One major factor can vary between streams: the number of event IDs assigned to
837 a stream. Luckily, this information tends to stay relatively constant (modulo
838 event registration while trace is being recorded), so we can specify different
839 representations for streams containing few event IDs and streams containing
840 many event IDs, so we end up representing the event ID and time-stamp as
841 densely as possible in each case.
843 The header is extended in the rare occasions where the information cannot be
844 represented in the ranges available in the standard event header. They are also
845 used in the rare occasions where the data required for a field could not be
846 collected: the flag corresponding to the missing field within the missing_fields
847 array is then set to 1.
849 Types uintX_t represent an X-bit unsigned integer, as declared with
852 typealias integer { size = X; align = X; signed = false; } := uintX_t;
856 typealias integer { size = X; align = 1; signed = false; } := uintX_t;
858 For more information about timestamp fields, see Section 8. Clocks.
860 6.1.1 Type 1 - Few event IDs
862 - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
864 - Native architecture byte ordering.
865 - For "compact" selection
866 - Fixed size: 32 bits.
867 - For "extended" selection
868 - Size depends on the architecture and variant alignment.
870 struct event_header_1 {
873 * id 31 is reserved to indicate an extended header.
875 enum : uint5_t { compact = 0 ... 30, extended = 31 } id;
881 uint32_t id; /* 32-bit event IDs */
882 uint64_t timestamp; /* 64-bit timestamps */
885 } align(32); /* or align(8) */
888 6.1.2 Type 2 - Many event IDs
890 - Aligned on 16-bit (or 8-bit if byte-packed, depending on the architecture
892 - Native architecture byte ordering.
893 - For "compact" selection
894 - Size depends on the architecture and variant alignment.
895 - For "extended" selection
896 - Size depends on the architecture and variant alignment.
898 struct event_header_2 {
900 * id: range: 0 - 65534.
901 * id 65535 is reserved to indicate an extended header.
903 enum : uint16_t { compact = 0 ... 65534, extended = 65535 } id;
909 uint32_t id; /* 32-bit event IDs */
910 uint64_t timestamp; /* 64-bit timestamps */
913 } align(16); /* or align(8) */
916 6.2 Stream Event Context and Event Context
918 The event context contains information relative to the current event.
919 The choice and meaning of this information is specified by the TSDL
920 stream and event meta-data descriptions. The stream context is applied
921 to all events within the stream. The stream context structure follows
922 the event header. The event context is applied to specific events. Its
923 structure follows the stream context structure.
925 An example of stream-level event context is to save the event payload size with
926 each event, or to save the current PID with each event. These are declared
927 within the stream declaration within the meta-data:
931 event.context := struct {
933 uint16_t payload_size;
937 An example of event-specific event context is to declare a bitmap of missing
938 fields, only appended after the stream event context if the extended event
939 header is selected. NR_FIELDS is the number of fields within the event (a
947 uint1_t missing_fields[NR_FIELDS]; /* missing event fields bitmap */
956 An event payload contains fields specific to a given event type. The fields
957 belonging to an event type are described in the event-specific meta-data
958 within a structure type.
962 No padding at the end of the event payload. This differs from the ISO/C standard
963 for structures, but follows the CTF standard for structures. In a trace, even
964 though it makes sense to align the beginning of a structure, it really makes no
965 sense to add padding at the end of the structure, because structures are usually
966 not followed by a structure of the same type.
968 This trick can be done by adding a zero-length "end" field at the end of the C
969 structures, and by using the offset of this field rather than using sizeof()
970 when calculating the size of a structure (see Appendix "A. Helper macros").
974 The event payload is aligned on the largest alignment required by types
975 contained within the payload. (This follows the ISO/C standard for structures)
978 7. Trace Stream Description Language (TSDL)
980 The Trace Stream Description Language (TSDL) allows expression of the
981 binary trace streams layout in a C99-like Domain Specific Language
987 The trace stream layout description is located in the trace meta-data.
988 The meta-data is itself located in a stream identified by its name:
991 The meta-data description can be expressed in two different formats:
992 text-only and packet-based. The text-only description facilitates
993 generation of meta-data and provides a convenient way to enter the
994 meta-data information by hand. The packet-based meta-data provides the
995 CTF stream packet facilities (checksumming, compression, encryption,
996 network-readiness) for meta-data stream generated and transported by a
999 The text-only meta-data file is a plain-text TSDL description. This file
1000 must begin with the following characters to identify the file as a CTF
1001 TSDL text-based metadata file (without the double-quotes) :
1005 It must be followed by a space, and the version of the specification
1006 followed by the CTF trace, e.g.:
1010 These characters allow automated discovery of file type and CTF
1011 specification version. They are interpreted as a the beginning of a
1012 comment by the TSDL metadata parser. The comment can be continued to
1013 contain extra commented characters before it is closed.
1015 The packet-based meta-data is made of "meta-data packets", which each
1016 start with a meta-data packet header. The packet-based meta-data
1017 description is detected by reading the magic number "0x75D11D57" at the
1018 beginning of the file. This magic number is also used to detect the
1019 endianness of the architecture by trying to read the CTF magic number
1020 and its counterpart in reversed endianness. The events within the
1021 meta-data stream have no event header nor event context. Each event only
1022 contains a special "sequence" payload, which is a sequence of bits which
1023 length is implicitly calculated by using the
1024 "trace.packet.header.content_size" field, minus the packet header size.
1025 The formatting of this sequence of bits is a plain-text representation
1026 of the TSDL description. Each meta-data packet start with a special
1027 packet header, specific to the meta-data stream, which contains,
1030 struct metadata_packet_header {
1031 uint32_t magic; /* 0x75D11D57 */
1032 uint8_t uuid[16]; /* Unique Universal Identifier */
1033 uint32_t checksum; /* 0 if unused */
1034 uint32_t content_size; /* in bits */
1035 uint32_t packet_size; /* in bits */
1036 uint8_t compression_scheme; /* 0 if unused */
1037 uint8_t encryption_scheme; /* 0 if unused */
1038 uint8_t checksum_scheme; /* 0 if unused */
1039 uint8_t major; /* CTF spec version major number */
1040 uint8_t minor; /* CTF spec version minor number */
1043 The packet-based meta-data can be converted to a text-only meta-data by
1044 concatenating all the strings it contains.
1046 In the textual representation of the meta-data, the text contained
1047 within "/*" and "*/", as well as within "//" and end of line, are
1048 treated as comments. Boolean values can be represented as true, TRUE,
1049 or 1 for true, and false, FALSE, or 0 for false. Within the string-based
1050 meta-data description, the trace UUID is represented as a string of
1051 hexadecimal digits and dashes "-". In the event packet header, the trace
1052 UUID is represented as an array of bytes.
1055 7.2 Declaration vs Definition
1057 A declaration associates a layout to a type, without specifying where
1058 this type is located in the event structure hierarchy (see Section 6).
1059 This therefore includes typedef, typealias, as well as all type
1060 specifiers. In certain circumstances (typedef, structure field and
1061 variant field), a declaration is followed by a declarator, which specify
1062 the newly defined type name (for typedef), or the field name (for
1063 declarations located within structure and variants). Array and sequence,
1064 declared with square brackets ("[" "]"), are part of the declarator,
1065 similarly to C99. The enumeration base type is specified by
1066 ": enum_base", which is part of the type specifier. The variant tag
1067 name, specified between "<" ">", is also part of the type specifier.
1069 A definition associates a type to a location in the event structure
1070 hierarchy (see Section 6). This association is denoted by ":=", as shown
1076 TSDL uses three different types of scoping: a lexical scope is used for
1077 declarations and type definitions, and static and dynamic scopes are
1078 used for variants references to tag fields (with relative and absolute
1079 path lookups) and for sequence references to length fields.
1083 Each of "trace", "env", "stream", "event", "struct" and "variant" have
1084 their own nestable declaration scope, within which types can be declared
1085 using "typedef" and "typealias". A root declaration scope also contains
1086 all declarations located outside of any of the aforementioned
1087 declarations. An inner declaration scope can refer to type declared
1088 within its container lexical scope prior to the inner declaration scope.
1089 Redefinition of a typedef or typealias is not valid, although hiding an
1090 upper scope typedef or typealias is allowed within a sub-scope.
1092 7.3.2 Static and Dynamic Scopes
1094 A local static scope consists in the scope generated by the declaration
1095 of fields within a compound type. A static scope is a local static scope
1096 augmented with the nested sub-static-scopes it contains.
1098 A dynamic scope consists in the static scope augmented with the
1099 implicit event structure definition hierarchy presented at Section 6.
1101 Multiple declarations of the same field name within a local static scope
1102 is not valid. It is however valid to re-use the same field name in
1103 different local scopes.
1105 Nested static and dynamic scopes form lookup paths. These are used for
1106 variant tag and sequence length references. They are used at the variant
1107 and sequence definition site to look up the location of the tag field
1108 associated with a variant, and to lookup up the location of the length
1109 field associated with a sequence.
1111 Variants and sequences can refer to a tag field either using a relative
1112 path or an absolute path. The relative path is relative to the scope in
1113 which the variant or sequence performing the lookup is located.
1114 Relative paths are only allowed to lookup within the same static scope,
1115 which includes its nested static scopes. Lookups targeting parent static
1116 scopes need to be performed with an absolute path.
1118 Absolute path lookups use the full path including the dynamic scope
1119 followed by a "." and then the static scope. Therefore, variants (or
1120 sequences) in lower levels in the dynamic scope (e.g. event context) can
1121 refer to a tag (or length) field located in upper levels (e.g. in the
1122 event header) by specifying, in this case, the associated tag with
1123 <stream.event.header.field_name>. This allows, for instance, the event
1124 context to define a variant referring to the "id" field of the event
1127 The dynamic scope prefixes are thus:
1129 - Trace Environment: <env. >,
1130 - Trace Packet Header: <trace.packet.header. >,
1131 - Stream Packet Context: <stream.packet.context. >,
1132 - Event Header: <stream.event.header. >,
1133 - Stream Event Context: <stream.event.context. >,
1134 - Event Context: <event.context. >,
1135 - Event Payload: <event.fields. >.
1138 The target dynamic scope must be specified explicitly when referring to
1139 a field outside of the static scope (absolute scope reference). No
1140 conflict can occur between relative and dynamic paths, because the
1141 keywords "trace", "stream", and "event" are reserved, and thus
1142 not permitted as field names. It is recommended that field names
1143 clashing with CTF and C99 reserved keywords use an underscore prefix to
1144 eliminate the risk of generating a description containing an invalid
1145 field name. Consequently, fields starting with an underscore should have
1146 their leading underscore removed by the CTF trace readers.
1149 The information available in the dynamic scopes can be thought of as the
1150 current tracing context. At trace production, information about the
1151 current context is saved into the specified scope field levels. At trace
1152 consumption, for each event, the current trace context is therefore
1153 readable by accessing the upper dynamic scopes.
1158 The grammar representing the TSDL meta-data is presented in Appendix C.
1159 TSDL Grammar. This section presents a rather lighter reading that
1160 consists in examples of TSDL meta-data, with template values.
1162 The stream "id" can be left out if there is only one stream in the
1163 trace. The event "id" field can be left out if there is only one event
1167 major = value; /* CTF spec version major number */
1168 minor = value; /* CTF spec version minor number */
1169 uuid = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"; /* Trace UUID */
1170 byte_order = be OR le; /* Endianness (required) */
1171 packet.header := struct {
1179 * The "env" (environment) scope contains assignment expressions. The
1180 * field names and content are implementation-defined.
1183 pid = value; /* example */
1184 proc_name = "name"; /* example */
1190 /* Type 1 - Few event IDs; Type 2 - Many event IDs. See section 6.1. */
1191 event.header := event_header_1 OR event_header_2;
1192 event.context := struct {
1195 packet.context := struct {
1201 name = "event_name";
1202 id = value; /* Numeric identifier within the stream */
1203 stream_id = stream_id;
1205 model.emf.uri = "string";
1215 name = "event_name";
1222 /* More detail on types in section 4. Types */
1227 * Type declarations behave similarly to the C standard.
1230 typedef aliased_type_specifiers new_type_declarators;
1232 /* e.g.: typedef struct example new_type_name[10]; */
1237 * The "typealias" declaration can be used to give a name (including
1238 * pointer declarator specifier) to a type. It should also be used to
1239 * map basic C types (float, int, unsigned long, ...) to a CTF type.
1240 * Typealias is a superset of "typedef": it also allows assignment of a
1241 * simple variable identifier to a type.
1244 typealias type_class {
1246 } := type_specifiers type_declarator;
1250 * typealias integer {
1254 * } := struct page *;
1256 * typealias integer {
1271 enum name : integer_type {
1277 * Unnamed types, contained within compound type fields, typedef or typealias.
1292 enum : integer_type {
1296 typedef type new_type[length];
1299 type field_name[length];
1302 typedef type new_type[length_type];
1305 type field_name[length_type];
1317 integer_type field_name:size; /* GNU/C bitfield */
1327 Clock metadata allows to describe the clock topology of the system, as
1328 well as to detail each clock parameter. In absence of clock description,
1329 it is assumed that all fields named "timestamp" use the same clock
1330 source, which increments once per nanosecond.
1332 Describing a clock and how it is used by streams is threefold: first,
1333 the clock and clock topology should be described in a "clock"
1334 description block, e.g.:
1337 name = cycle_counter_sync;
1338 uuid = "62189bee-96dc-11e0-91a8-cfa3d89f3923";
1339 description = "Cycle counter synchronized across CPUs";
1340 freq = 1000000000; /* frequency, in Hz */
1341 /* precision in seconds is: 1000 * (1/freq) */
1344 * clock value offset from Epoch is:
1345 * offset_s + (offset * (1/freq))
1347 offset_s = 1326476837;
1352 The mandatory "name" field specifies the name of the clock identifier,
1353 which can later be used as a reference. The optional field "uuid" is the
1354 unique identifier of the clock. It can be used to correlate different
1355 traces that use the same clock. An optional textual description string
1356 can be added with the "description" field. The "freq" field is the
1357 initial frequency of the clock, in Hz. If the "freq" field is not
1358 present, the frequency is assumed to be 1000000000 (providing clock
1359 increment of 1 ns). The optional "precision" field details the
1360 uncertainty on the clock measurements, in (1/freq) units. The "offset_s"
1361 and "offset" fields indicate the offset from POSIX.1 Epoch, 1970-01-01
1362 00:00:00 +0000 (UTC), to the zero of value of the clock. The "offset_s"
1363 field is in seconds. The "offset" field is in (1/freq) units. If any of
1364 the "offset_s" or "offset" field is not present, it is assigned the 0
1365 value. The field "absolute" is TRUE if the clock is a global reference
1366 across different clock uuid (e.g. NTP time). Otherwise, "absolute" is
1367 FALSE, and the clock can be considered as synchronized only with other
1368 clocks that have the same uuid.
1371 Secondly, a reference to this clock should be added within an integer
1375 size = 64; align = 1; signed = false;
1376 map = clock.cycle_counter_sync.value;
1379 Thirdly, stream declarations can reference the clock they use as a
1382 struct packet_context {
1383 uint64_ccnt_t ccnt_begin;
1384 uint64_ccnt_t ccnt_end;
1390 event.header := struct {
1391 uint64_ccnt_t timestamp;
1394 packet.context := struct packet_context;
1397 For a N-bit integer type referring to a clock, if the integer overflows
1398 compared to the N low order bits of the clock prior value found in the
1399 same stream, then it is assumed that one, and only one, overflow
1400 occurred. It is therefore important that events encoding time on a small
1401 number of bits happen frequently enough to detect when more than one
1402 N-bit overflow occurs.
1404 In a packet context, clock field names ending with "_begin" and "_end"
1405 have a special meaning: this refers to the time-stamps at, respectively,
1406 the beginning and the end of each packet.
1411 The two following macros keep track of the size of a GNU/C structure without
1412 padding at the end by placing HEADER_END as the last field. A one byte end field
1413 is used for C90 compatibility (C99 flexible arrays could be used here). Note
1414 that this does not affect the effective structure size, which should always be
1415 calculated with the header_sizeof() helper.
1417 #define HEADER_END char end_field
1418 #define header_sizeof(type) offsetof(typeof(type), end_field)
1421 B. Stream Header Rationale
1423 An event stream is divided in contiguous event packets of variable size. These
1424 subdivisions allow the trace analyzer to perform a fast binary search by time
1425 within the stream (typically requiring to index only the event packet headers)
1426 without reading the whole stream. These subdivisions have a variable size to
1427 eliminate the need to transfer the event packet padding when partially filled
1428 event packets must be sent when streaming a trace for live viewing/analysis.
1429 An event packet can contain a certain amount of padding at the end. Dividing
1430 streams into event packets is also useful for network streaming over UDP and
1431 flight recorder mode tracing (a whole event packet can be swapped out of the
1432 buffer atomically for reading).
1434 The stream header is repeated at the beginning of each event packet to allow
1435 flexibility in terms of:
1437 - streaming support,
1438 - allowing arbitrary buffers to be discarded without making the trace
1440 - allow UDP packet loss handling by either dealing with missing event packet
1441 or asking for re-transmission.
1442 - transparently support flight recorder mode,
1443 - transparently support crash dump.
1449 * Common Trace Format (CTF) Trace Stream Description Language (TSDL) Grammar.
1451 * Inspired from the C99 grammar:
1452 * http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf (Annex A)
1453 * and c++1x grammar (draft)
1454 * http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3291.pdf (Annex A)
1456 * Specialized for CTF needs by including only constant and declarations from
1457 * C99 (excluding function declarations), and by adding support for variants,
1458 * sequences and CTF-specific specifiers. Enumeration container types
1459 * semantic is inspired from c++1x enum-base.
1464 1.1) Lexical elements
1511 identifier identifier-nondigit
1514 identifier-nondigit:
1516 universal-character-name
1517 any other implementation-defined characters
1521 [a-zA-Z] /* regular expression */
1524 [0-9] /* regular expression */
1526 1.4) Universal character names
1528 universal-character-name:
1530 \U hex-quad hex-quad
1533 hexadecimal-digit hexadecimal-digit hexadecimal-digit hexadecimal-digit
1539 enumeration-constant
1543 decimal-constant integer-suffix-opt
1544 octal-constant integer-suffix-opt
1545 hexadecimal-constant integer-suffix-opt
1549 decimal-constant digit
1553 octal-constant octal-digit
1555 hexadecimal-constant:
1556 hexadecimal-prefix hexadecimal-digit
1557 hexadecimal-constant hexadecimal-digit
1567 unsigned-suffix long-suffix-opt
1568 unsigned-suffix long-long-suffix
1569 long-suffix unsigned-suffix-opt
1570 long-long-suffix unsigned-suffix-opt
1584 enumeration-constant:
1590 L' c-char-sequence '
1594 c-char-sequence c-char
1597 any member of source charset except single-quote ('), backslash
1598 (\), or new-line character.
1602 simple-escape-sequence
1603 octal-escape-sequence
1604 hexadecimal-escape-sequence
1605 universal-character-name
1607 simple-escape-sequence: one of
1608 \' \" \? \\ \a \b \f \n \r \t \v
1610 octal-escape-sequence:
1612 \ octal-digit octal-digit
1613 \ octal-digit octal-digit octal-digit
1615 hexadecimal-escape-sequence:
1616 \x hexadecimal-digit
1617 hexadecimal-escape-sequence hexadecimal-digit
1619 1.6) String literals
1622 " s-char-sequence-opt "
1623 L" s-char-sequence-opt "
1627 s-char-sequence s-char
1630 any member of source charset except double-quote ("), backslash
1631 (\), or new-line character.
1637 [ ] ( ) { } . -> * + - < > : ; ... = ,
1640 2) Phrase structure grammar
1646 ( unary-expression )
1650 postfix-expression [ unary-expression ]
1651 postfix-expression . identifier
1652 postfix-expressoin -> identifier
1656 unary-operator postfix-expression
1658 unary-operator: one of
1661 assignment-operator:
1664 type-assignment-operator:
1667 constant-expression-range:
1668 unary-expression ... unary-expression
1673 declaration-specifiers declarator-list-opt ;
1676 declaration-specifiers:
1677 storage-class-specifier declaration-specifiers-opt
1678 type-specifier declaration-specifiers-opt
1679 type-qualifier declaration-specifiers-opt
1683 declarator-list , declarator
1685 abstract-declarator-list:
1687 abstract-declarator-list , abstract-declarator
1689 storage-class-specifier:
1712 align ( unary-expression )
1715 struct identifier-opt { struct-or-variant-declaration-list-opt } align-attribute-opt
1716 struct identifier align-attribute-opt
1718 struct-or-variant-declaration-list:
1719 struct-or-variant-declaration
1720 struct-or-variant-declaration-list struct-or-variant-declaration
1722 struct-or-variant-declaration:
1723 specifier-qualifier-list struct-or-variant-declarator-list ;
1724 declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list ;
1725 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list ;
1726 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list ;
1728 specifier-qualifier-list:
1729 type-specifier specifier-qualifier-list-opt
1730 type-qualifier specifier-qualifier-list-opt
1732 struct-or-variant-declarator-list:
1733 struct-or-variant-declarator
1734 struct-or-variant-declarator-list , struct-or-variant-declarator
1736 struct-or-variant-declarator:
1738 declarator-opt : unary-expression
1741 variant identifier-opt variant-tag-opt { struct-or-variant-declaration-list }
1742 variant identifier variant-tag
1745 < unary-expression >
1748 enum identifier-opt { enumerator-list }
1749 enum identifier-opt { enumerator-list , }
1751 enum identifier-opt : declaration-specifiers { enumerator-list }
1752 enum identifier-opt : declaration-specifiers { enumerator-list , }
1756 enumerator-list , enumerator
1759 enumeration-constant
1760 enumeration-constant assignment-operator unary-expression
1761 enumeration-constant assignment-operator constant-expression-range
1767 pointer-opt direct-declarator
1772 direct-declarator [ unary-expression ]
1774 abstract-declarator:
1775 pointer-opt direct-abstract-declarator
1777 direct-abstract-declarator:
1779 ( abstract-declarator )
1780 direct-abstract-declarator [ unary-expression ]
1781 direct-abstract-declarator [ ]
1784 * type-qualifier-list-opt
1785 * type-qualifier-list-opt pointer
1787 type-qualifier-list:
1789 type-qualifier-list type-qualifier
1794 2.3) CTF-specific declarations
1797 clock { ctf-assignment-expression-list-opt }
1798 event { ctf-assignment-expression-list-opt }
1799 stream { ctf-assignment-expression-list-opt }
1800 env { ctf-assignment-expression-list-opt }
1801 trace { ctf-assignment-expression-list-opt }
1802 callsite { ctf-assignment-expression-list-opt }
1803 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1804 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list
1807 floating_point { ctf-assignment-expression-list-opt }
1808 integer { ctf-assignment-expression-list-opt }
1809 string { ctf-assignment-expression-list-opt }
1812 ctf-assignment-expression-list:
1813 ctf-assignment-expression ;
1814 ctf-assignment-expression-list ctf-assignment-expression ;
1816 ctf-assignment-expression:
1817 unary-expression assignment-operator unary-expression
1818 unary-expression type-assignment-operator type-specifier
1819 declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list
1820 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1821 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list