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. A MIT-licensed reference implementation of the
218 CTF portable bitfields is available at:
220 http://git.efficios.com/?p=babeltrace.git;a=blob;f=include/babeltrace/bitfield.h
222 Binary representation of integers:
224 - On little and big endian:
225 - Within a byte, high bits correspond to an integer high bits, and low bits
226 correspond to low bits.
228 - Integer across multiple bytes are placed from the less significant to the
230 - Consecutive integers are placed from lower bits to higher bits (even within
233 - Integer across multiple bytes are placed from the most significant to the
235 - Consecutive integers are placed from higher bits to lower bits (even within
238 This binary representation is derived from the bitfield implementation in GCC
239 for little and big endian. However, contrary to what GCC does, integers can
240 cross units boundaries (no padding is required). Padding can be explicitly
241 added (see 4.1.6 GNU/C bitfields) to follow the GCC layout if needed.
243 TSDL meta-data representation:
246 signed = true OR false; /* default false */
247 byte_order = native OR network OR be OR le; /* default native */
248 size = value; /* value in bits, no default */
249 align = value; /* value in bits */
250 /* based used for pretty-printing output, default: decimal. */
251 base = decimal OR dec OR d OR i OR u OR 10 OR hexadecimal OR hex OR x OR X OR p OR 16
252 OR octal OR oct OR o OR 8 OR binary OR b OR 2;
253 /* character encoding, default: none */
254 encoding = none or UTF8 or ASCII;
257 Example of type inheritance (creation of a uint32_t named type):
265 Definition of a named 5-bit signed bitfield:
273 The character encoding field can be used to specify that the integer
274 must be printed as a text character when read. e.g.:
284 4.1.6 GNU/C bitfields
286 The GNU/C bitfields follow closely the integer representation, with a
287 particularity on alignment: if a bitfield cannot fit in the current unit, the
288 unit is padded and the bitfield starts at the following unit. The unit size is
289 defined by the size of the type "unit_type".
291 TSDL meta-data representation:
295 As an example, the following structure declared in C compiled by GCC:
302 The example structure is aligned on the largest element (short). The second
303 bitfield would be aligned on the next unit boundary, because it would not fit in
308 The floating point values byte ordering is defined in the TSDL meta-data.
310 Floating point values follow the IEEE 754-2008 standard interchange formats.
311 Description of the floating point values include the exponent and mantissa size
312 in bits. Some requirements are imposed on the floating point values:
314 - FLT_RADIX must be 2.
315 - mant_dig is the number of digits represented in the mantissa. It is specified
316 by the ISO C99 standard, section 5.2.4, as FLT_MANT_DIG, DBL_MANT_DIG and
317 LDBL_MANT_DIG as defined by <float.h>.
318 - exp_dig is the number of digits represented in the exponent. Given that
319 mant_dig is one bit more than its actual size in bits (leading 1 is not
320 needed) and also given that the sign bit always takes one bit, exp_dig can be
323 - sizeof(float) * CHAR_BIT - FLT_MANT_DIG
324 - sizeof(double) * CHAR_BIT - DBL_MANT_DIG
325 - sizeof(long double) * CHAR_BIT - LDBL_MANT_DIG
327 TSDL meta-data representation:
332 byte_order = native OR network OR be OR le;
336 Example of type inheritance:
338 typealias floating_point {
339 exp_dig = 8; /* sizeof(float) * CHAR_BIT - FLT_MANT_DIG */
340 mant_dig = 24; /* FLT_MANT_DIG */
345 TODO: define NaN, +inf, -inf behavior.
347 Bit-packed, byte-packed or larger alignments can be used for floating
348 point values, similarly to integers.
352 Enumerations are a mapping between an integer type and a table of strings. The
353 numerical representation of the enumeration follows the integer type specified
354 by the meta-data. The enumeration mapping table is detailed in the enumeration
355 description within the meta-data. The mapping table maps inclusive value
356 ranges (or single values) to strings. Instead of being limited to simple
357 "value -> string" mappings, these enumerations map
358 "[ start_value ... end_value ] -> string", which map inclusive ranges of
359 values to strings. An enumeration from the C language can be represented in
360 this format by having the same start_value and end_value for each element, which
361 is in fact a range of size 1. This single-value range is supported without
362 repeating the start and end values with the value = string declaration.
364 enum name : integer_type {
365 somestring = start_value1 ... end_value1,
366 "other string" = start_value2 ... end_value2,
367 yet_another_string, /* will be assigned to end_value2 + 1 */
368 "some other string" = value,
372 If the values are omitted, the enumeration starts at 0 and increment of 1 for
373 each entry. An entry with omitted value that follows a range entry takes
374 as value the end_value of the previous range + 1:
376 enum name : unsigned int {
384 Overlapping ranges within a single enumeration are implementation defined.
386 A nameless enumeration can be declared as a field type or as part of a typedef:
388 enum : integer_type {
392 Enumerations omitting the container type ": integer_type" use the "int"
393 type (for compatibility with C99). The "int" type must be previously
396 typealias integer { size = 32; align = 32; signed = true; } := int;
405 Compound are aggregation of type declarations. Compound types include
406 structures, variant, arrays, sequences, and strings.
410 Structures are aligned on the largest alignment required by basic types
411 contained within the structure. (This follows the ISO/C standard for structures)
413 TSDL meta-data representation of a named structure:
416 field_type field_name;
417 field_type field_name;
424 integer { /* Nameless type */
429 uint64_t second_field_name; /* Named type declared in the meta-data */
432 The fields are placed in a sequence next to each other. They each
433 possess a field name, which is a unique identifier within the structure.
434 The identifier is not allowed to use any reserved keyword
435 (see Section C.1.2). Replacing reserved keywords with
436 underscore-prefixed field names is recommended. Fields starting with an
437 underscore should have their leading underscore removed by the CTF trace
440 A nameless structure can be declared as a field type or as part of a typedef:
446 Alignment for a structure compound type can be forced to a minimum value
447 by adding an "align" specifier after the declaration of a structure
448 body. This attribute is read as: align(value). The value is specified in
449 bits. The structure will be aligned on the maximum value between this
450 attribute and the alignment required by the basic types contained within
457 4.2.2 Variants (Discriminated/Tagged Unions)
459 A CTF variant is a selection between different types. A CTF variant must
460 always be defined within the scope of a structure or within fields
461 contained within a structure (defined recursively). A "tag" enumeration
462 field must appear in either the same static scope, prior to the variant
463 field (in field declaration order), in an upper static scope , or in an
464 upper dynamic scope (see Section 7.3.2). The type selection is indicated
465 by the mapping from the enumeration value to the string used as variant
466 type selector. The field to use as tag is specified by the "tag_field",
467 specified between "< >" after the "variant" keyword for unnamed
468 variants, and after "variant name" for named variants.
470 The alignment of the variant is the alignment of the type as selected by
471 the tag value for the specific instance of the variant. The size of the
472 variant is the size as selected by the tag value for the specific
473 instance of the variant.
475 The alignment of the type containing the variant is independent of the
476 variant alignment. For instance, if a structure contains two fields, a
477 32-bit integer, aligned on 32 bits, and a variant, which contains two
478 choices: either a 32-bit field, aligned on 32 bits, or a 64-bit field,
479 aligned on 64 bits, the alignment of the outmost structure will be
480 32-bit (the alignment of its largest field, disregarding the alignment
481 of the variant). The alignment of the variant will depend on the
482 selector: if the variant's 32-bit field is selected, its alignment will
483 be 32-bit, or 64-bit otherwise. It is important to note that variants
484 are specifically tailored for compactness in a stream. Therefore, the
485 relative offsets of compound type fields can vary depending on
486 the offset at which the compound type starts if it contains a variant
487 that itself contains a type with alignment larger than the largest field
488 contained within the compound type. This is caused by the fact that the
489 compound type may contain the enumeration that select the variant's
490 choice, and therefore the alignment to be applied to the compound type
491 cannot be determined before encountering the enumeration.
493 Each variant type selector possess a field name, which is a unique
494 identifier within the variant. The identifier is not allowed to use any
495 reserved keyword (see Section C.1.2). Replacing reserved keywords with
496 underscore-prefixed field names is recommended. Fields starting with an
497 underscore should have their leading underscore removed by the CTF trace
501 A named variant declaration followed by its definition within a structure
512 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
514 variant name <tag_field> v;
517 An unnamed variant definition within a structure is expressed by the following
521 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
523 variant <tag_field> {
531 Example of a named variant within a sequence that refers to a single tag field:
540 enum : uint2_t { a, b, c } choice;
542 variant example <choice> v[seqlen];
545 Example of an unnamed variant:
548 enum : uint2_t { a, b, c, d } choice;
549 /* Unrelated fields can be added between the variant and its tag */
562 Example of an unnamed variant within an array:
565 enum : uint2_t { a, b, c } choice;
573 Example of a variant type definition within a structure, where the defined type
574 is then declared within an array of structures. This variant refers to a tag
575 located in an upper static scope. This example clearly shows that a variant
576 type definition referring to the tag "x" uses the closest preceding field from
577 the static scope of the type definition.
580 enum : uint2_t { a, b, c, d } x;
582 typedef variant <x> { /*
583 * "x" refers to the preceding "x" enumeration in the
584 * static scope of the type definition.
592 enum : int { x, y, z } x; /* This enumeration is not used by "v". */
593 example_variant v; /*
594 * "v" uses the "enum : uint2_t { a, b, c, d }"
602 Arrays are fixed-length. Their length is declared in the type
603 declaration within the meta-data. They contain an array of "inner type"
604 elements, which can refer to any type not containing the type of the
605 array being declared (no circular dependency). The length is the number
606 of elements in an array.
608 TSDL meta-data representation of a named array:
610 typedef elem_type name[length];
612 A nameless array can be declared as a field type within a structure, e.g.:
614 uint8_t field_name[10];
616 Arrays are always aligned on their element alignment requirement.
620 Sequences are dynamically-sized arrays. They refer to a "length"
621 unsigned integer field, which must appear in either the same static scope,
622 prior to the sequence field (in field declaration order), in an upper
623 static scope, or in an upper dynamic scope (see Section 7.3.2). This
624 length field represents the number of elements in the sequence. The
625 sequence per se is an array of "inner type" elements.
627 TSDL meta-data representation for a sequence type definition:
630 unsigned int length_field;
631 typedef elem_type typename[length_field];
632 typename seq_field_name;
635 A sequence can also be declared as a field type, e.g.:
638 unsigned int length_field;
639 long seq_field_name[length_field];
642 Multiple sequences can refer to the same length field, and these length
643 fields can be in a different upper dynamic scope:
645 e.g., assuming the stream.event.header defines:
650 event.header := struct {
659 long seq_a[stream.event.header.seq_len];
660 char seq_b[stream.event.header.seq_len];
664 The sequence elements follow the "array" specifications.
668 Strings are an array of bytes of variable size and are terminated by a '\0'
669 "NULL" character. Their encoding is described in the TSDL meta-data. In
670 absence of encoding attribute information, the default encoding is
673 TSDL meta-data representation of a named string type:
676 encoding = UTF8 OR ASCII;
679 A nameless string type can be declared as a field type:
681 string field_name; /* Use default UTF8 encoding */
683 Strings are always aligned on byte size.
685 5. Event Packet Header
687 The event packet header consists of two parts: the "event packet header"
688 is the same for all streams of a trace. The second part, the "event
689 packet context", is described on a per-stream basis. Both are described
690 in the TSDL meta-data.
692 Event packet header (all fields are optional, specified by TSDL meta-data):
694 - Magic number (CTF magic number: 0xC1FC1FC1) specifies that this is a
695 CTF packet. This magic number is optional, but when present, it should
696 come at the very beginning of the packet.
697 - Trace UUID, used to ensure the event packet match the meta-data used.
698 (note: we cannot use a meta-data checksum in every cases instead of a
699 UUID because meta-data can be appended to while tracing is active)
700 This field is optional.
701 - Stream ID, used as reference to stream description in meta-data.
702 This field is optional if there is only one stream description in the
703 meta-data, but becomes required if there are more than one stream in
704 the TSDL meta-data description.
706 Event packet context (all fields are optional, specified by TSDL meta-data):
708 - Event packet content size (in bits).
709 - Event packet size (in bits, includes padding).
710 - Event packet content checksum. Checksum excludes the event packet
712 - Per-stream event packet sequence count (to deal with UDP packet loss). The
713 number of significant sequence counter bits should also be present, so
714 wrap-arounds are dealt with correctly.
715 - Time-stamp at the beginning and time-stamp at the end of the event packet.
716 Both timestamps are written in the packet header, but sampled respectively
717 while (or before) writing the first event and while (or after) writing the
718 last event in the packet. The inclusive range between these timestamps should
719 include all event timestamps assigned to events contained within the packet.
720 The timestamp at the beginning of an event packet is guaranteed to be
721 below or equal the timestamp at the end of that event packet.
722 The timestamp at the end of an event packet is guaranteed to be below
723 or equal the timestamps at the end of any following packet within the
724 same stream. See Section 8. Clocks for more detail.
725 - Events discarded count
726 - Snapshot of a per-stream free-running counter, counting the number of
727 events discarded that were supposed to be written in the stream after
728 the last event in the event packet.
729 * Note: producer-consumer buffer full condition can fill the current
730 event packet with padding so we know exactly where events have been
731 discarded. However, if the buffer full condition chooses not
732 to fill the current event packet with padding, all we know
733 about the timestamp range in which the events have been
734 discarded is that it is somewhere between the beginning and
735 the end of the packet.
736 - Lossless compression scheme used for the event packet content. Applied
737 directly to raw data. New types of compression can be added in following
738 versions of the format.
739 0: no compression scheme
743 - Cypher used for the event packet content. Applied after compression.
746 - Checksum scheme used for the event packet content. Applied after encryption.
752 5.1 Event Packet Header Description
754 The event packet header layout is indicated by the trace packet.header
755 field. Here is a recommended structure type for the packet header with
756 the fields typically expected (although these fields are each optional):
758 struct event_packet_header {
766 packet.header := struct event_packet_header;
769 If the magic number is not present, tools such as "file" will have no
770 mean to discover the file type.
772 If the uuid is not present, no validation that the meta-data actually
773 corresponds to the stream is performed.
775 If the stream_id packet header field is missing, the trace can only
776 contain a single stream. Its "id" field can be left out, and its events
777 don't need to declare a "stream_id" field.
780 5.2 Event Packet Context Description
782 Event packet context example. These are declared within the stream declaration
783 in the meta-data. All these fields are optional. If the packet size field is
784 missing, the whole stream only contains a single packet. If the content
785 size field is missing, the packet is filled (no padding). The content
786 and packet sizes include all headers.
788 An example event packet context type:
790 struct event_packet_context {
791 uint64_t timestamp_begin;
792 uint64_t timestamp_end;
794 uint32_t stream_packet_count;
795 uint32_t events_discarded;
797 uint64_t/uint32_t/uint16_t content_size;
798 uint64_t/uint32_t/uint16_t packet_size;
799 uint8_t compression_scheme;
800 uint8_t encryption_scheme;
801 uint8_t checksum_scheme;
807 The overall structure of an event is:
809 1 - Event Header (as specified by the stream meta-data)
810 2 - Stream Event Context (as specified by the stream meta-data)
811 3 - Event Context (as specified by the event meta-data)
812 4 - Event Payload (as specified by the event meta-data)
814 This structure defines an implicit dynamic scoping, where variants
815 located in inner structures (those with a higher number in the listing
816 above) can refer to the fields of outer structures (with lower number in
817 the listing above). See Section 7.3 TSDL Scopes for more detail.
821 Event headers can be described within the meta-data. We hereby propose, as an
822 example, two types of events headers. Type 1 accommodates streams with less than
823 31 event IDs. Type 2 accommodates streams with 31 or more event IDs.
825 One major factor can vary between streams: the number of event IDs assigned to
826 a stream. Luckily, this information tends to stay relatively constant (modulo
827 event registration while trace is being recorded), so we can specify different
828 representations for streams containing few event IDs and streams containing
829 many event IDs, so we end up representing the event ID and time-stamp as
830 densely as possible in each case.
832 The header is extended in the rare occasions where the information cannot be
833 represented in the ranges available in the standard event header. They are also
834 used in the rare occasions where the data required for a field could not be
835 collected: the flag corresponding to the missing field within the missing_fields
836 array is then set to 1.
838 Types uintX_t represent an X-bit unsigned integer, as declared with
841 typealias integer { size = X; align = X; signed = false; } := uintX_t;
845 typealias integer { size = X; align = 1; signed = false; } := uintX_t;
847 For more information about timestamp fields, see Section 8. Clocks.
849 6.1.1 Type 1 - Few event IDs
851 - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
853 - Native architecture byte ordering.
854 - For "compact" selection
855 - Fixed size: 32 bits.
856 - For "extended" selection
857 - Size depends on the architecture and variant alignment.
859 struct event_header_1 {
862 * id 31 is reserved to indicate an extended header.
864 enum : uint5_t { compact = 0 ... 30, extended = 31 } id;
870 uint32_t id; /* 32-bit event IDs */
871 uint64_t timestamp; /* 64-bit timestamps */
874 } align(32); /* or align(8) */
877 6.1.2 Type 2 - Many event IDs
879 - Aligned on 16-bit (or 8-bit if byte-packed, depending on the architecture
881 - Native architecture byte ordering.
882 - For "compact" selection
883 - Size depends on the architecture and variant alignment.
884 - For "extended" selection
885 - Size depends on the architecture and variant alignment.
887 struct event_header_2 {
889 * id: range: 0 - 65534.
890 * id 65535 is reserved to indicate an extended header.
892 enum : uint16_t { compact = 0 ... 65534, extended = 65535 } id;
898 uint32_t id; /* 32-bit event IDs */
899 uint64_t timestamp; /* 64-bit timestamps */
902 } align(16); /* or align(8) */
905 6.2 Stream Event Context and Event Context
907 The event context contains information relative to the current event.
908 The choice and meaning of this information is specified by the TSDL
909 stream and event meta-data descriptions. The stream context is applied
910 to all events within the stream. The stream context structure follows
911 the event header. The event context is applied to specific events. Its
912 structure follows the stream context structure.
914 An example of stream-level event context is to save the event payload size with
915 each event, or to save the current PID with each event. These are declared
916 within the stream declaration within the meta-data:
920 event.context := struct {
922 uint16_t payload_size;
926 An example of event-specific event context is to declare a bitmap of missing
927 fields, only appended after the stream event context if the extended event
928 header is selected. NR_FIELDS is the number of fields within the event (a
936 uint1_t missing_fields[NR_FIELDS]; /* missing event fields bitmap */
945 An event payload contains fields specific to a given event type. The fields
946 belonging to an event type are described in the event-specific meta-data
947 within a structure type.
951 No padding at the end of the event payload. This differs from the ISO/C standard
952 for structures, but follows the CTF standard for structures. In a trace, even
953 though it makes sense to align the beginning of a structure, it really makes no
954 sense to add padding at the end of the structure, because structures are usually
955 not followed by a structure of the same type.
957 This trick can be done by adding a zero-length "end" field at the end of the C
958 structures, and by using the offset of this field rather than using sizeof()
959 when calculating the size of a structure (see Appendix "A. Helper macros").
963 The event payload is aligned on the largest alignment required by types
964 contained within the payload. (This follows the ISO/C standard for structures)
967 7. Trace Stream Description Language (TSDL)
969 The Trace Stream Description Language (TSDL) allows expression of the
970 binary trace streams layout in a C99-like Domain Specific Language
976 The trace stream layout description is located in the trace meta-data.
977 The meta-data is itself located in a stream identified by its name:
980 The meta-data description can be expressed in two different formats:
981 text-only and packet-based. The text-only description facilitates
982 generation of meta-data and provides a convenient way to enter the
983 meta-data information by hand. The packet-based meta-data provides the
984 CTF stream packet facilities (checksumming, compression, encryption,
985 network-readiness) for meta-data stream generated and transported by a
988 The text-only meta-data file is a plain-text TSDL description. This file
989 must begin with the following characters to identify the file as a CTF
990 TSDL text-based metadata file (without the double-quotes) :
994 It must be followed by a space, and the version of the specification
995 followed by the CTF trace, e.g.:
999 These characters allow automated discovery of file type and CTF
1000 specification version. They are interpreted as a the beginning of a
1001 comment by the TSDL metadata parser. The comment can be continued to
1002 contain extra commented characters before it is closed.
1004 The packet-based meta-data is made of "meta-data packets", which each
1005 start with a meta-data packet header. The packet-based meta-data
1006 description is detected by reading the magic number "0x75D11D57" at the
1007 beginning of the file. This magic number is also used to detect the
1008 endianness of the architecture by trying to read the CTF magic number
1009 and its counterpart in reversed endianness. The events within the
1010 meta-data stream have no event header nor event context. Each event only
1011 contains a special "sequence" payload, which is a sequence of bits which
1012 length is implicitly calculated by using the
1013 "trace.packet.header.content_size" field, minus the packet header size.
1014 The formatting of this sequence of bits is a plain-text representation
1015 of the TSDL description. Each meta-data packet start with a special
1016 packet header, specific to the meta-data stream, which contains,
1019 struct metadata_packet_header {
1020 uint32_t magic; /* 0x75D11D57 */
1021 uint8_t uuid[16]; /* Unique Universal Identifier */
1022 uint32_t checksum; /* 0 if unused */
1023 uint32_t content_size; /* in bits */
1024 uint32_t packet_size; /* in bits */
1025 uint8_t compression_scheme; /* 0 if unused */
1026 uint8_t encryption_scheme; /* 0 if unused */
1027 uint8_t checksum_scheme; /* 0 if unused */
1028 uint8_t major; /* CTF spec version major number */
1029 uint8_t minor; /* CTF spec version minor number */
1032 The packet-based meta-data can be converted to a text-only meta-data by
1033 concatenating all the strings it contains.
1035 In the textual representation of the meta-data, the text contained
1036 within "/*" and "*/", as well as within "//" and end of line, are
1037 treated as comments. Boolean values can be represented as true, TRUE,
1038 or 1 for true, and false, FALSE, or 0 for false. Within the string-based
1039 meta-data description, the trace UUID is represented as a string of
1040 hexadecimal digits and dashes "-". In the event packet header, the trace
1041 UUID is represented as an array of bytes.
1044 7.2 Declaration vs Definition
1046 A declaration associates a layout to a type, without specifying where
1047 this type is located in the event structure hierarchy (see Section 6).
1048 This therefore includes typedef, typealias, as well as all type
1049 specifiers. In certain circumstances (typedef, structure field and
1050 variant field), a declaration is followed by a declarator, which specify
1051 the newly defined type name (for typedef), or the field name (for
1052 declarations located within structure and variants). Array and sequence,
1053 declared with square brackets ("[" "]"), are part of the declarator,
1054 similarly to C99. The enumeration base type is specified by
1055 ": enum_base", which is part of the type specifier. The variant tag
1056 name, specified between "<" ">", is also part of the type specifier.
1058 A definition associates a type to a location in the event structure
1059 hierarchy (see Section 6). This association is denoted by ":=", as shown
1065 TSDL uses three different types of scoping: a lexical scope is used for
1066 declarations and type definitions, and static and dynamic scopes are
1067 used for variants references to tag fields (with relative and absolute
1068 path lookups) and for sequence references to length fields.
1072 Each of "trace", "env", "stream", "event", "struct" and "variant" have
1073 their own nestable declaration scope, within which types can be declared
1074 using "typedef" and "typealias". A root declaration scope also contains
1075 all declarations located outside of any of the aforementioned
1076 declarations. An inner declaration scope can refer to type declared
1077 within its container lexical scope prior to the inner declaration scope.
1078 Redefinition of a typedef or typealias is not valid, although hiding an
1079 upper scope typedef or typealias is allowed within a sub-scope.
1081 7.3.2 Static and Dynamic Scopes
1083 A local static scope consists in the scope generated by the declaration
1084 of fields within a compound type. A static scope is a local static scope
1085 augmented with the nested sub-static-scopes it contains.
1087 A dynamic scope consists in the static scope augmented with the
1088 implicit event structure definition hierarchy presented at Section 6.
1090 Multiple declarations of the same field name within a local static scope
1091 is not valid. It is however valid to re-use the same field name in
1092 different local scopes.
1094 Nested static and dynamic scopes form lookup paths. These are used for
1095 variant tag and sequence length references. They are used at the variant
1096 and sequence definition site to look up the location of the tag field
1097 associated with a variant, and to lookup up the location of the length
1098 field associated with a sequence.
1100 Variants and sequences can refer to a tag field either using a relative
1101 path or an absolute path. The relative path is relative to the scope in
1102 which the variant or sequence performing the lookup is located.
1103 Relative paths are only allowed to lookup within the same static scope,
1104 which includes its nested static scopes. Lookups targeting parent static
1105 scopes need to be performed with an absolute path.
1107 Absolute path lookups use the full path including the dynamic scope
1108 followed by a "." and then the static scope. Therefore, variants (or
1109 sequences) in lower levels in the dynamic scope (e.g. event context) can
1110 refer to a tag (or length) field located in upper levels (e.g. in the
1111 event header) by specifying, in this case, the associated tag with
1112 <stream.event.header.field_name>. This allows, for instance, the event
1113 context to define a variant referring to the "id" field of the event
1116 The dynamic scope prefixes are thus:
1118 - Trace Environment: <env. >,
1119 - Trace Packet Header: <trace.packet.header. >,
1120 - Stream Packet Context: <stream.packet.context. >,
1121 - Event Header: <stream.event.header. >,
1122 - Stream Event Context: <stream.event.context. >,
1123 - Event Context: <event.context. >,
1124 - Event Payload: <event.fields. >.
1127 The target dynamic scope must be specified explicitly when referring to
1128 a field outside of the static scope (absolute scope reference). No
1129 conflict can occur between relative and dynamic paths, because the
1130 keywords "trace", "stream", and "event" are reserved, and thus
1131 not permitted as field names. It is recommended that field names
1132 clashing with CTF and C99 reserved keywords use an underscore prefix to
1133 eliminate the risk of generating a description containing an invalid
1134 field name. Consequently, fields starting with an underscore should have
1135 their leading underscore removed by the CTF trace readers.
1138 The information available in the dynamic scopes can be thought of as the
1139 current tracing context. At trace production, information about the
1140 current context is saved into the specified scope field levels. At trace
1141 consumption, for each event, the current trace context is therefore
1142 readable by accessing the upper dynamic scopes.
1147 The grammar representing the TSDL meta-data is presented in Appendix C.
1148 TSDL Grammar. This section presents a rather lighter reading that
1149 consists in examples of TSDL meta-data, with template values.
1151 The stream "id" can be left out if there is only one stream in the
1152 trace. The event "id" field can be left out if there is only one event
1156 major = value; /* CTF spec version major number */
1157 minor = value; /* CTF spec version minor number */
1158 uuid = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"; /* Trace UUID */
1159 byte_order = be OR le; /* Endianness (required) */
1160 packet.header := struct {
1168 * The "env" (environment) scope contains assignment expressions. The
1169 * field names and content are implementation-defined.
1172 pid = value; /* example */
1173 proc_name = "name"; /* example */
1179 /* Type 1 - Few event IDs; Type 2 - Many event IDs. See section 6.1. */
1180 event.header := event_header_1 OR event_header_2;
1181 event.context := struct {
1184 packet.context := struct {
1190 name = "event_name";
1191 id = value; /* Numeric identifier within the stream */
1192 stream_id = stream_id;
1194 model.emf.uri = "string";
1204 name = "event_name";
1211 /* More detail on types in section 4. Types */
1216 * Type declarations behave similarly to the C standard.
1219 typedef aliased_type_specifiers new_type_declarators;
1221 /* e.g.: typedef struct example new_type_name[10]; */
1226 * The "typealias" declaration can be used to give a name (including
1227 * pointer declarator specifier) to a type. It should also be used to
1228 * map basic C types (float, int, unsigned long, ...) to a CTF type.
1229 * Typealias is a superset of "typedef": it also allows assignment of a
1230 * simple variable identifier to a type.
1233 typealias type_class {
1235 } := type_specifiers type_declarator;
1239 * typealias integer {
1243 * } := struct page *;
1245 * typealias integer {
1260 enum name : integer_type {
1266 * Unnamed types, contained within compound type fields, typedef or typealias.
1281 enum : integer_type {
1285 typedef type new_type[length];
1288 type field_name[length];
1291 typedef type new_type[length_type];
1294 type field_name[length_type];
1306 integer_type field_name:size; /* GNU/C bitfield */
1316 Clock metadata allows to describe the clock topology of the system, as
1317 well as to detail each clock parameter. In absence of clock description,
1318 it is assumed that all fields named "timestamp" use the same clock
1319 source, which increments once per nanosecond.
1321 Describing a clock and how it is used by streams is threefold: first,
1322 the clock and clock topology should be described in a "clock"
1323 description block, e.g.:
1326 name = cycle_counter_sync;
1327 uuid = "62189bee-96dc-11e0-91a8-cfa3d89f3923";
1328 description = "Cycle counter synchronized across CPUs";
1329 freq = 1000000000; /* frequency, in Hz */
1330 /* precision in seconds is: 1000 * (1/freq) */
1333 * clock value offset from Epoch is:
1334 * offset_s + (offset * (1/freq))
1336 offset_s = 1326476837;
1341 The mandatory "name" field specifies the name of the clock identifier,
1342 which can later be used as a reference. The optional field "uuid" is the
1343 unique identifier of the clock. It can be used to correlate different
1344 traces that use the same clock. An optional textual description string
1345 can be added with the "description" field. The "freq" field is the
1346 initial frequency of the clock, in Hz. If the "freq" field is not
1347 present, the frequency is assumed to be 1000000000 (providing clock
1348 increment of 1 ns). The optional "precision" field details the
1349 uncertainty on the clock measurements, in (1/freq) units. The "offset_s"
1350 and "offset" fields indicate the offset from POSIX.1 Epoch, 1970-01-01
1351 00:00:00 +0000 (UTC), to the zero of value of the clock. The "offset_s"
1352 field is in seconds. The "offset" field is in (1/freq) units. If any of
1353 the "offset_s" or "offset" field is not present, it is assigned the 0
1354 value. The field "absolute" is TRUE if the clock is a global reference
1355 across different clock uuid (e.g. NTP time). Otherwise, "absolute" is
1356 FALSE, and the clock can be considered as synchronized only with other
1357 clocks that have the same uuid.
1360 Secondly, a reference to this clock should be added within an integer
1364 size = 64; align = 1; signed = false;
1365 map = clock.cycle_counter_sync.value;
1368 Thirdly, stream declarations can reference the clock they use as a
1371 struct packet_context {
1372 uint64_ccnt_t ccnt_begin;
1373 uint64_ccnt_t ccnt_end;
1379 event.header := struct {
1380 uint64_ccnt_t timestamp;
1383 packet.context := struct packet_context;
1386 For a N-bit integer type referring to a clock, if the integer overflows
1387 compared to the N low order bits of the clock prior value found in the
1388 same stream, then it is assumed that one, and only one, overflow
1389 occurred. It is therefore important that events encoding time on a small
1390 number of bits happen frequently enough to detect when more than one
1391 N-bit overflow occurs.
1393 In a packet context, clock field names ending with "_begin" and "_end"
1394 have a special meaning: this refers to the time-stamps at, respectively,
1395 the beginning and the end of each packet.
1400 The two following macros keep track of the size of a GNU/C structure without
1401 padding at the end by placing HEADER_END as the last field. A one byte end field
1402 is used for C90 compatibility (C99 flexible arrays could be used here). Note
1403 that this does not affect the effective structure size, which should always be
1404 calculated with the header_sizeof() helper.
1406 #define HEADER_END char end_field
1407 #define header_sizeof(type) offsetof(typeof(type), end_field)
1410 B. Stream Header Rationale
1412 An event stream is divided in contiguous event packets of variable size. These
1413 subdivisions allow the trace analyzer to perform a fast binary search by time
1414 within the stream (typically requiring to index only the event packet headers)
1415 without reading the whole stream. These subdivisions have a variable size to
1416 eliminate the need to transfer the event packet padding when partially filled
1417 event packets must be sent when streaming a trace for live viewing/analysis.
1418 An event packet can contain a certain amount of padding at the end. Dividing
1419 streams into event packets is also useful for network streaming over UDP and
1420 flight recorder mode tracing (a whole event packet can be swapped out of the
1421 buffer atomically for reading).
1423 The stream header is repeated at the beginning of each event packet to allow
1424 flexibility in terms of:
1426 - streaming support,
1427 - allowing arbitrary buffers to be discarded without making the trace
1429 - allow UDP packet loss handling by either dealing with missing event packet
1430 or asking for re-transmission.
1431 - transparently support flight recorder mode,
1432 - transparently support crash dump.
1438 * Common Trace Format (CTF) Trace Stream Description Language (TSDL) Grammar.
1440 * Inspired from the C99 grammar:
1441 * http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf (Annex A)
1442 * and c++1x grammar (draft)
1443 * http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3291.pdf (Annex A)
1445 * Specialized for CTF needs by including only constant and declarations from
1446 * C99 (excluding function declarations), and by adding support for variants,
1447 * sequences and CTF-specific specifiers. Enumeration container types
1448 * semantic is inspired from c++1x enum-base.
1453 1.1) Lexical elements
1500 identifier identifier-nondigit
1503 identifier-nondigit:
1505 universal-character-name
1506 any other implementation-defined characters
1510 [a-zA-Z] /* regular expression */
1513 [0-9] /* regular expression */
1515 1.4) Universal character names
1517 universal-character-name:
1519 \U hex-quad hex-quad
1522 hexadecimal-digit hexadecimal-digit hexadecimal-digit hexadecimal-digit
1528 enumeration-constant
1532 decimal-constant integer-suffix-opt
1533 octal-constant integer-suffix-opt
1534 hexadecimal-constant integer-suffix-opt
1538 decimal-constant digit
1542 octal-constant octal-digit
1544 hexadecimal-constant:
1545 hexadecimal-prefix hexadecimal-digit
1546 hexadecimal-constant hexadecimal-digit
1556 unsigned-suffix long-suffix-opt
1557 unsigned-suffix long-long-suffix
1558 long-suffix unsigned-suffix-opt
1559 long-long-suffix unsigned-suffix-opt
1573 enumeration-constant:
1579 L' c-char-sequence '
1583 c-char-sequence c-char
1586 any member of source charset except single-quote ('), backslash
1587 (\), or new-line character.
1591 simple-escape-sequence
1592 octal-escape-sequence
1593 hexadecimal-escape-sequence
1594 universal-character-name
1596 simple-escape-sequence: one of
1597 \' \" \? \\ \a \b \f \n \r \t \v
1599 octal-escape-sequence:
1601 \ octal-digit octal-digit
1602 \ octal-digit octal-digit octal-digit
1604 hexadecimal-escape-sequence:
1605 \x hexadecimal-digit
1606 hexadecimal-escape-sequence hexadecimal-digit
1608 1.6) String literals
1611 " s-char-sequence-opt "
1612 L" s-char-sequence-opt "
1616 s-char-sequence s-char
1619 any member of source charset except double-quote ("), backslash
1620 (\), or new-line character.
1626 [ ] ( ) { } . -> * + - < > : ; ... = ,
1629 2) Phrase structure grammar
1635 ( unary-expression )
1639 postfix-expression [ unary-expression ]
1640 postfix-expression . identifier
1641 postfix-expressoin -> identifier
1645 unary-operator postfix-expression
1647 unary-operator: one of
1650 assignment-operator:
1653 type-assignment-operator:
1656 constant-expression-range:
1657 unary-expression ... unary-expression
1662 declaration-specifiers declarator-list-opt ;
1665 declaration-specifiers:
1666 storage-class-specifier declaration-specifiers-opt
1667 type-specifier declaration-specifiers-opt
1668 type-qualifier declaration-specifiers-opt
1672 declarator-list , declarator
1674 abstract-declarator-list:
1676 abstract-declarator-list , abstract-declarator
1678 storage-class-specifier:
1701 align ( unary-expression )
1704 struct identifier-opt { struct-or-variant-declaration-list-opt } align-attribute-opt
1705 struct identifier align-attribute-opt
1707 struct-or-variant-declaration-list:
1708 struct-or-variant-declaration
1709 struct-or-variant-declaration-list struct-or-variant-declaration
1711 struct-or-variant-declaration:
1712 specifier-qualifier-list struct-or-variant-declarator-list ;
1713 declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list ;
1714 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list ;
1715 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list ;
1717 specifier-qualifier-list:
1718 type-specifier specifier-qualifier-list-opt
1719 type-qualifier specifier-qualifier-list-opt
1721 struct-or-variant-declarator-list:
1722 struct-or-variant-declarator
1723 struct-or-variant-declarator-list , struct-or-variant-declarator
1725 struct-or-variant-declarator:
1727 declarator-opt : unary-expression
1730 variant identifier-opt variant-tag-opt { struct-or-variant-declaration-list }
1731 variant identifier variant-tag
1734 < unary-expression >
1737 enum identifier-opt { enumerator-list }
1738 enum identifier-opt { enumerator-list , }
1740 enum identifier-opt : declaration-specifiers { enumerator-list }
1741 enum identifier-opt : declaration-specifiers { enumerator-list , }
1745 enumerator-list , enumerator
1748 enumeration-constant
1749 enumeration-constant assignment-operator unary-expression
1750 enumeration-constant assignment-operator constant-expression-range
1756 pointer-opt direct-declarator
1761 direct-declarator [ unary-expression ]
1763 abstract-declarator:
1764 pointer-opt direct-abstract-declarator
1766 direct-abstract-declarator:
1768 ( abstract-declarator )
1769 direct-abstract-declarator [ unary-expression ]
1770 direct-abstract-declarator [ ]
1773 * type-qualifier-list-opt
1774 * type-qualifier-list-opt pointer
1776 type-qualifier-list:
1778 type-qualifier-list type-qualifier
1783 2.3) CTF-specific declarations
1786 clock { ctf-assignment-expression-list-opt }
1787 event { ctf-assignment-expression-list-opt }
1788 stream { ctf-assignment-expression-list-opt }
1789 env { ctf-assignment-expression-list-opt }
1790 trace { ctf-assignment-expression-list-opt }
1791 callsite { ctf-assignment-expression-list-opt }
1792 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1793 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list
1796 floating_point { ctf-assignment-expression-list-opt }
1797 integer { ctf-assignment-expression-list-opt }
1798 string { ctf-assignment-expression-list-opt }
1801 ctf-assignment-expression-list:
1802 ctf-assignment-expression ;
1803 ctf-assignment-expression-list ctf-assignment-expression ;
1805 ctf-assignment-expression:
1806 unary-expression assignment-operator unary-expression
1807 unary-expression type-assignment-operator type-specifier
1808 declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list
1809 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1810 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list