1 Common Trace Format (CTF) Specification (v1.8.1)
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
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 TSDL meta-data attribute representation of a specific alignment:
170 align = value; /* value in bits */
174 By default, the native endianness of the source architecture is used.
175 Byte order can be overridden for a basic type by specifying a "byte_order"
176 attribute. Typical use-case is to specify the network byte order (big endian:
177 "be") to save data captured from the network into the trace without conversion.
178 If not specified, the byte order is native.
180 TSDL meta-data representation:
182 byte_order = native OR network OR be OR le; /* network and be are aliases */
186 Type size, in bits, for integers and floats is that returned by "sizeof()" in C
187 multiplied by CHAR_BIT.
188 We require the size of "char" and "unsigned char" types (CHAR_BIT) to be fixed
189 to 8 bits for cross-endianness compatibility.
191 TSDL meta-data representation:
193 size = value; (value is in bits)
197 Signed integers are represented in two-complement. Integer alignment,
198 size, signedness and byte ordering are defined in the TSDL meta-data.
199 Integers aligned on byte size (8-bit) and with length multiple of byte
200 size (8-bit) correspond to the C99 standard integers. In addition,
201 integers with alignment and/or size that are _not_ a multiple of the
202 byte size are permitted; these correspond to the C99 standard bitfields,
203 with the added specification that the CTF integer bitfields have a fixed
204 binary representation. A MIT-licensed reference implementation of the
205 CTF portable bitfields is available at:
207 http://git.efficios.com/?p=babeltrace.git;a=blob;f=include/babeltrace/bitfield.h
209 Binary representation of integers:
211 - On little and big endian:
212 - Within a byte, high bits correspond to an integer high bits, and low bits
213 correspond to low bits.
215 - Integer across multiple bytes are placed from the less significant to the
217 - Consecutive integers are placed from lower bits to higher bits (even within
220 - Integer across multiple bytes are placed from the most significant to the
222 - Consecutive integers are placed from higher bits to lower bits (even within
225 This binary representation is derived from the bitfield implementation in GCC
226 for little and big endian. However, contrary to what GCC does, integers can
227 cross units boundaries (no padding is required). Padding can be explicitly
228 added (see 4.1.6 GNU/C bitfields) to follow the GCC layout if needed.
230 TSDL meta-data representation:
233 signed = true OR false; /* default false */
234 byte_order = native OR network OR be OR le; /* default native */
235 size = value; /* value in bits, no default */
236 align = value; /* value in bits */
237 /* based used for pretty-printing output, default: decimal. */
238 base = decimal OR dec OR OR d OR i OR u OR 10 OR hexadecimal OR hex OR x OR X OR p OR 16
239 OR octal OR oct OR o OR 8 OR binary OR b OR 2;
240 /* character encoding, default: none */
241 encoding = none or UTF8 or ASCII;
244 Example of type inheritance (creation of a uint32_t named type):
252 Definition of a named 5-bit signed bitfield:
260 The character encoding field can be used to specify that the integer
261 must be printed as a text character when read. e.g.:
271 4.1.6 GNU/C bitfields
273 The GNU/C bitfields follow closely the integer representation, with a
274 particularity on alignment: if a bitfield cannot fit in the current unit, the
275 unit is padded and the bitfield starts at the following unit. The unit size is
276 defined by the size of the type "unit_type".
278 TSDL meta-data representation:
282 As an example, the following structure declared in C compiled by GCC:
289 The example structure is aligned on the largest element (short). The second
290 bitfield would be aligned on the next unit boundary, because it would not fit in
295 The floating point values byte ordering is defined in the TSDL meta-data.
297 Floating point values follow the IEEE 754-2008 standard interchange formats.
298 Description of the floating point values include the exponent and mantissa size
299 in bits. Some requirements are imposed on the floating point values:
301 - FLT_RADIX must be 2.
302 - mant_dig is the number of digits represented in the mantissa. It is specified
303 by the ISO C99 standard, section 5.2.4, as FLT_MANT_DIG, DBL_MANT_DIG and
304 LDBL_MANT_DIG as defined by <float.h>.
305 - exp_dig is the number of digits represented in the exponent. Given that
306 mant_dig is one bit more than its actual size in bits (leading 1 is not
307 needed) and also given that the sign bit always takes one bit, exp_dig can be
310 - sizeof(float) * CHAR_BIT - FLT_MANT_DIG
311 - sizeof(double) * CHAR_BIT - DBL_MANT_DIG
312 - sizeof(long double) * CHAR_BIT - LDBL_MANT_DIG
314 TSDL meta-data representation:
319 byte_order = native OR network OR be OR le;
323 Example of type inheritance:
325 typealias floating_point {
326 exp_dig = 8; /* sizeof(float) * CHAR_BIT - FLT_MANT_DIG */
327 mant_dig = 24; /* FLT_MANT_DIG */
332 TODO: define NaN, +inf, -inf behavior.
334 Bit-packed, byte-packed or larger alignments can be used for floating
335 point values, similarly to integers.
339 Enumerations are a mapping between an integer type and a table of strings. The
340 numerical representation of the enumeration follows the integer type specified
341 by the meta-data. The enumeration mapping table is detailed in the enumeration
342 description within the meta-data. The mapping table maps inclusive value
343 ranges (or single values) to strings. Instead of being limited to simple
344 "value -> string" mappings, these enumerations map
345 "[ start_value ... end_value ] -> string", which map inclusive ranges of
346 values to strings. An enumeration from the C language can be represented in
347 this format by having the same start_value and end_value for each element, which
348 is in fact a range of size 1. This single-value range is supported without
349 repeating the start and end values with the value = string declaration.
351 enum name : integer_type {
352 somestring = start_value1 ... end_value1,
353 "other string" = start_value2 ... end_value2,
354 yet_another_string, /* will be assigned to end_value2 + 1 */
355 "some other string" = value,
359 If the values are omitted, the enumeration starts at 0 and increment of 1 for
360 each entry. An entry with omitted value that follows a range entry takes
361 as value the end_value of the previous range + 1:
363 enum name : unsigned int {
371 Overlapping ranges within a single enumeration are implementation defined.
373 A nameless enumeration can be declared as a field type or as part of a typedef:
375 enum : integer_type {
379 Enumerations omitting the container type ": integer_type" use the "int"
380 type (for compatibility with C99). The "int" type must be previously
383 typealias integer { size = 32; align = 32; signed = true } := int;
392 Compound are aggregation of type declarations. Compound types include
393 structures, variant, arrays, sequences, and strings.
397 Structures are aligned on the largest alignment required by basic types
398 contained within the structure. (This follows the ISO/C standard for structures)
400 TSDL meta-data representation of a named structure:
403 field_type field_name;
404 field_type field_name;
411 integer { /* Nameless type */
416 uint64_t second_field_name; /* Named type declared in the meta-data */
419 The fields are placed in a sequence next to each other. They each
420 possess a field name, which is a unique identifier within the structure.
421 The identifier is not allowed to use any reserved keyword
422 (see Section C.1.2). Replacing reserved keywords with
423 underscore-prefixed field names is recommended. Fields starting with an
424 underscore should have their leading underscore removed by the CTF trace
427 A nameless structure can be declared as a field type or as part of a typedef:
433 Alignment for a structure compound type can be forced to a minimum value
434 by adding an "align" specifier after the declaration of a structure
435 body. This attribute is read as: align(value). The value is specified in
436 bits. The structure will be aligned on the maximum value between this
437 attribute and the alignment required by the basic types contained within
444 4.2.2 Variants (Discriminated/Tagged Unions)
446 A CTF variant is a selection between different types. A CTF variant must
447 always be defined within the scope of a structure or within fields
448 contained within a structure (defined recursively). A "tag" enumeration
449 field must appear in either the same static scope, prior to the variant
450 field (in field declaration order), in an upper static scope , or in an
451 upper dynamic scope (see Section 7.3.2). The type selection is indicated
452 by the mapping from the enumeration value to the string used as variant
453 type selector. The field to use as tag is specified by the "tag_field",
454 specified between "< >" after the "variant" keyword for unnamed
455 variants, and after "variant name" for named variants.
457 The alignment of the variant is the alignment of the type as selected by the tag
458 value for the specific instance of the variant. The alignment of the type
459 containing the variant is independent of the variant alignment. The size of the
460 variant is the size as selected by the tag value for the specific instance of
463 Each variant type selector possess a field name, which is a unique
464 identifier within the variant. The identifier is not allowed to use any
465 reserved keyword (see Section C.1.2). Replacing reserved keywords with
466 underscore-prefixed field names is recommended. Fields starting with an
467 underscore should have their leading underscore removed by the CTF trace
471 A named variant declaration followed by its definition within a structure
482 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
484 variant name <tag_field> v;
487 An unnamed variant definition within a structure is expressed by the following
491 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
493 variant <tag_field> {
501 Example of a named variant within a sequence that refers to a single tag field:
510 enum : uint2_t { a, b, c } choice;
512 variant example <choice> v[seqlen];
515 Example of an unnamed variant:
518 enum : uint2_t { a, b, c, d } choice;
519 /* Unrelated fields can be added between the variant and its tag */
532 Example of an unnamed variant within an array:
535 enum : uint2_t { a, b, c } choice;
543 Example of a variant type definition within a structure, where the defined type
544 is then declared within an array of structures. This variant refers to a tag
545 located in an upper static scope. This example clearly shows that a variant
546 type definition referring to the tag "x" uses the closest preceding field from
547 the static scope of the type definition.
550 enum : uint2_t { a, b, c, d } x;
552 typedef variant <x> { /*
553 * "x" refers to the preceding "x" enumeration in the
554 * static scope of the type definition.
562 enum : int { x, y, z } x; /* This enumeration is not used by "v". */
563 example_variant v; /*
564 * "v" uses the "enum : uint2_t { a, b, c, d }"
572 Arrays are fixed-length. Their length is declared in the type
573 declaration within the meta-data. They contain an array of "inner type"
574 elements, which can refer to any type not containing the type of the
575 array being declared (no circular dependency). The length is the number
576 of elements in an array.
578 TSDL meta-data representation of a named array:
580 typedef elem_type name[length];
582 A nameless array can be declared as a field type within a structure, e.g.:
584 uint8_t field_name[10];
586 Arrays are always aligned on their element alignment requirement.
590 Sequences are dynamically-sized arrays. They refer to a a "length"
591 unsigned integer field, which must appear in either the same static scope,
592 prior to the sequence field (in field declaration order), in an upper
593 static scope, or in an upper dynamic scope (see Section 7.3.2). This
594 length field represents the number of elements in the sequence. The
595 sequence per se is an array of "inner type" elements.
597 TSDL meta-data representation for a sequence type definition:
600 unsigned int length_field;
601 typedef elem_type typename[length_field];
602 typename seq_field_name;
605 A sequence can also be declared as a field type, e.g.:
608 unsigned int length_field;
609 long seq_field_name[length_field];
612 Multiple sequences can refer to the same length field, and these length
613 fields can be in a different upper dynamic scope:
615 e.g., assuming the stream.event.header defines:
620 event.header := struct {
629 long seq_a[stream.event.header.seq_len];
630 char seq_b[stream.event.header.seq_len];
634 The sequence elements follow the "array" specifications.
638 Strings are an array of bytes of variable size and are terminated by a '\0'
639 "NULL" character. Their encoding is described in the TSDL meta-data. In
640 absence of encoding attribute information, the default encoding is
643 TSDL meta-data representation of a named string type:
646 encoding = UTF8 OR ASCII;
649 A nameless string type can be declared as a field type:
651 string field_name; /* Use default UTF8 encoding */
653 Strings are always aligned on byte size.
655 5. Event Packet Header
657 The event packet header consists of two parts: the "event packet header"
658 is the same for all streams of a trace. The second part, the "event
659 packet context", is described on a per-stream basis. Both are described
660 in the TSDL meta-data. The packets are aligned on architecture-page-sized
663 Event packet header (all fields are optional, specified by TSDL meta-data):
665 - Magic number (CTF magic number: 0xC1FC1FC1) specifies that this is a
666 CTF packet. This magic number is optional, but when present, it should
667 come at the very beginning of the packet.
668 - Trace UUID, used to ensure the event packet match the meta-data used.
669 (note: we cannot use a meta-data checksum in every cases instead of a
670 UUID because meta-data can be appended to while tracing is active)
671 This field is optional.
672 - Stream ID, used as reference to stream description in meta-data.
673 This field is optional if there is only one stream description in the
674 meta-data, but becomes required if there are more than one stream in
675 the TSDL meta-data description.
677 Event packet context (all fields are optional, specified by TSDL meta-data):
679 - Event packet content size (in bits).
680 - Event packet size (in bits, includes padding).
681 - Event packet content checksum. Checksum excludes the event packet
683 - Per-stream event packet sequence count (to deal with UDP packet loss). The
684 number of significant sequence counter bits should also be present, so
685 wrap-arounds are dealt with correctly.
686 - Time-stamp at the beginning and time-stamp at the end of the event packet.
687 Both timestamps are written in the packet header, but sampled respectively
688 while (or before) writing the first event and while (or after) writing the
689 last event in the packet. The inclusive range between these timestamps should
690 include all event timestamps assigned to events contained within the packet.
691 - Events discarded count
692 - Snapshot of a per-stream free-running counter, counting the number of
693 events discarded that were supposed to be written in the stream after
694 the last event in the event packet.
695 * Note: producer-consumer buffer full condition can fill the current
696 event packet with padding so we know exactly where events have been
697 discarded. However, if the buffer full condition chooses not
698 to fill the current event packet with padding, all we know
699 about the timestamp range in which the events have been
700 discarded is that it is somewhere between the beginning and
701 the end of the packet.
702 - Lossless compression scheme used for the event packet content. Applied
703 directly to raw data. New types of compression can be added in following
704 versions of the format.
705 0: no compression scheme
709 - Cypher used for the event packet content. Applied after compression.
712 - Checksum scheme used for the event packet content. Applied after encryption.
718 5.1 Event Packet Header Description
720 The event packet header layout is indicated by the trace packet.header
721 field. Here is a recommended structure type for the packet header with
722 the fields typically expected (although these fields are each optional):
724 struct event_packet_header {
732 packet.header := struct event_packet_header;
735 If the magic number is not present, tools such as "file" will have no
736 mean to discover the file type.
738 If the uuid is not present, no validation that the meta-data actually
739 corresponds to the stream is performed.
741 If the stream_id packet header field is missing, the trace can only
742 contain a single stream. Its "id" field can be left out, and its events
743 don't need to declare a "stream_id" field.
746 5.2 Event Packet Context Description
748 Event packet context example. These are declared within the stream declaration
749 in the meta-data. All these fields are optional. If the packet size field is
750 missing, the whole stream only contains a single packet. If the content
751 size field is missing, the packet is filled (no padding). The content
752 and packet sizes include all headers.
754 An example event packet context type:
756 struct event_packet_context {
757 uint64_t timestamp_begin;
758 uint64_t timestamp_end;
760 uint32_t stream_packet_count;
761 uint32_t events_discarded;
763 uint32_t/uint16_t content_size;
764 uint32_t/uint16_t packet_size;
765 uint8_t compression_scheme;
766 uint8_t encryption_scheme;
767 uint8_t checksum_scheme;
773 The overall structure of an event is:
775 1 - Stream Packet Context (as specified by the stream meta-data)
776 2 - Event Header (as specified by the stream meta-data)
777 3 - Stream Event Context (as specified by the stream meta-data)
778 4 - Event Context (as specified by the event meta-data)
779 5 - Event Payload (as specified by the event meta-data)
781 This structure defines an implicit dynamic scoping, where variants
782 located in inner structures (those with a higher number in the listing
783 above) can refer to the fields of outer structures (with lower number in
784 the listing above). See Section 7.3 TSDL Scopes for more detail.
788 Event headers can be described within the meta-data. We hereby propose, as an
789 example, two types of events headers. Type 1 accommodates streams with less than
790 31 event IDs. Type 2 accommodates streams with 31 or more event IDs.
792 One major factor can vary between streams: the number of event IDs assigned to
793 a stream. Luckily, this information tends to stay relatively constant (modulo
794 event registration while trace is being recorded), so we can specify different
795 representations for streams containing few event IDs and streams containing
796 many event IDs, so we end up representing the event ID and time-stamp as
797 densely as possible in each case.
799 The header is extended in the rare occasions where the information cannot be
800 represented in the ranges available in the standard event header. They are also
801 used in the rare occasions where the data required for a field could not be
802 collected: the flag corresponding to the missing field within the missing_fields
803 array is then set to 1.
805 Types uintX_t represent an X-bit unsigned integer, as declared with
808 typealias integer { size = X; align = X; signed = false } := uintX_t;
812 typealias integer { size = X; align = 1; signed = false } := uintX_t;
814 6.1.1 Type 1 - Few event IDs
816 - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
818 - Native architecture byte ordering.
819 - For "compact" selection
820 - Fixed size: 32 bits.
821 - For "extended" selection
822 - Size depends on the architecture and variant alignment.
824 struct event_header_1 {
827 * id 31 is reserved to indicate an extended header.
829 enum : uint5_t { compact = 0 ... 30, extended = 31 } id;
835 uint32_t id; /* 32-bit event IDs */
836 uint64_t timestamp; /* 64-bit timestamps */
839 } align(32); /* or align(8) */
842 6.1.2 Type 2 - Many event IDs
844 - Aligned on 16-bit (or 8-bit if byte-packed, depending on the architecture
846 - Native architecture byte ordering.
847 - For "compact" selection
848 - Size depends on the architecture and variant alignment.
849 - For "extended" selection
850 - Size depends on the architecture and variant alignment.
852 struct event_header_2 {
854 * id: range: 0 - 65534.
855 * id 65535 is reserved to indicate an extended header.
857 enum : uint16_t { compact = 0 ... 65534, extended = 65535 } id;
863 uint32_t id; /* 32-bit event IDs */
864 uint64_t timestamp; /* 64-bit timestamps */
867 } align(16); /* or align(8) */
872 The event context contains information relative to the current event.
873 The choice and meaning of this information is specified by the TSDL
874 stream and event meta-data descriptions. The stream context is applied
875 to all events within the stream. The stream context structure follows
876 the event header. The event context is applied to specific events. Its
877 structure follows the stream context structure.
879 An example of stream-level event context is to save the event payload size with
880 each event, or to save the current PID with each event. These are declared
881 within the stream declaration within the meta-data:
885 event.context := struct {
887 uint16_t payload_size;
891 An example of event-specific event context is to declare a bitmap of missing
892 fields, only appended after the stream event context if the extended event
893 header is selected. NR_FIELDS is the number of fields within the event (a
901 uint1_t missing_fields[NR_FIELDS]; /* missing event fields bitmap */
910 An event payload contains fields specific to a given event type. The fields
911 belonging to an event type are described in the event-specific meta-data
912 within a structure type.
916 No padding at the end of the event payload. This differs from the ISO/C standard
917 for structures, but follows the CTF standard for structures. In a trace, even
918 though it makes sense to align the beginning of a structure, it really makes no
919 sense to add padding at the end of the structure, because structures are usually
920 not followed by a structure of the same type.
922 This trick can be done by adding a zero-length "end" field at the end of the C
923 structures, and by using the offset of this field rather than using sizeof()
924 when calculating the size of a structure (see Appendix "A. Helper macros").
928 The event payload is aligned on the largest alignment required by types
929 contained within the payload. (This follows the ISO/C standard for structures)
932 7. Trace Stream Description Language (TSDL)
934 The Trace Stream Description Language (TSDL) allows expression of the
935 binary trace streams layout in a C99-like Domain Specific Language
941 The trace stream layout description is located in the trace meta-data.
942 The meta-data is itself located in a stream identified by its name:
945 The meta-data description can be expressed in two different formats:
946 text-only and packet-based. The text-only description facilitates
947 generation of meta-data and provides a convenient way to enter the
948 meta-data information by hand. The packet-based meta-data provides the
949 CTF stream packet facilities (checksumming, compression, encryption,
950 network-readiness) for meta-data stream generated and transported by a
953 The text-only meta-data file is a plain-text TSDL description. This file
954 must begin with the following characters to identify the file as a CTF
955 TSDL text-based metadata file (without the double-quotes) :
959 It must be followed by a space, and the version of the specification
960 followed by the CTF trace, e.g.:
964 These characters allow automated discovery of file type and CTF
965 specification version. They are interpreted as a the beginning of a
966 comment by the TSDL metadata parser. The comment can be continued to
967 contain extra commented characters before it is closed.
969 The packet-based meta-data is made of "meta-data packets", which each
970 start with a meta-data packet header. The packet-based meta-data
971 description is detected by reading the magic number "0x75D11D57" at the
972 beginning of the file. This magic number is also used to detect the
973 endianness of the architecture by trying to read the CTF magic number
974 and its counterpart in reversed endianness. The events within the
975 meta-data stream have no event header nor event context. Each event only
976 contains a special "sequence" payload, which is a sequence of bits which
977 length is implicitly calculated by using the
978 "trace.packet.header.content_size" field, minus the packet header size.
979 The formatting of this sequence of bits is a plain-text representation
980 of the TSDL description. Each meta-data packet start with a special
981 packet header, specific to the meta-data stream, which contains,
984 struct metadata_packet_header {
985 uint32_t magic; /* 0x75D11D57 */
986 uint8_t uuid[16]; /* Unique Universal Identifier */
987 uint32_t checksum; /* 0 if unused */
988 uint32_t content_size; /* in bits */
989 uint32_t packet_size; /* in bits */
990 uint8_t compression_scheme; /* 0 if unused */
991 uint8_t encryption_scheme; /* 0 if unused */
992 uint8_t checksum_scheme; /* 0 if unused */
993 uint8_t major; /* CTF spec version major number */
994 uint8_t minor; /* CTF spec version minor number */
997 The packet-based meta-data can be converted to a text-only meta-data by
998 concatenating all the strings it contains.
1000 In the textual representation of the meta-data, the text contained
1001 within "/*" and "*/", as well as within "//" and end of line, are
1002 treated as comments. Boolean values can be represented as true, TRUE,
1003 or 1 for true, and false, FALSE, or 0 for false. Within the string-based
1004 meta-data description, the trace UUID is represented as a string of
1005 hexadecimal digits and dashes "-". In the event packet header, the trace
1006 UUID is represented as an array of bytes.
1009 7.2 Declaration vs Definition
1011 A declaration associates a layout to a type, without specifying where
1012 this type is located in the event structure hierarchy (see Section 6).
1013 This therefore includes typedef, typealias, as well as all type
1014 specifiers. In certain circumstances (typedef, structure field and
1015 variant field), a declaration is followed by a declarator, which specify
1016 the newly defined type name (for typedef), or the field name (for
1017 declarations located within structure and variants). Array and sequence,
1018 declared with square brackets ("[" "]"), are part of the declarator,
1019 similarly to C99. The enumeration base type is specified by
1020 ": enum_base", which is part of the type specifier. The variant tag
1021 name, specified between "<" ">", is also part of the type specifier.
1023 A definition associates a type to a location in the event structure
1024 hierarchy (see Section 6). This association is denoted by ":=", as shown
1030 TSDL uses three different types of scoping: a lexical scope is used for
1031 declarations and type definitions, and static and dynamic scopes are
1032 used for variants references to tag fields (with relative and absolute
1033 path lookups) and for sequence references to length fields.
1037 Each of "trace", "env", "stream", "event", "struct" and "variant" have
1038 their own nestable declaration scope, within which types can be declared
1039 using "typedef" and "typealias". A root declaration scope also contains
1040 all declarations located outside of any of the aforementioned
1041 declarations. An inner declaration scope can refer to type declared
1042 within its container lexical scope prior to the inner declaration scope.
1043 Redefinition of a typedef or typealias is not valid, although hiding an
1044 upper scope typedef or typealias is allowed within a sub-scope.
1046 7.3.2 Static and Dynamic Scopes
1048 A local static scope consists in the scope generated by the declaration
1049 of fields within a compound type. A static scope is a local static scope
1050 augmented with the nested sub-static-scopes it contains.
1052 A dynamic scope consists in the static scope augmented with the
1053 implicit event structure definition hierarchy presented at Section 6.
1055 Multiple declarations of the same field name within a local static scope
1056 is not valid. It is however valid to re-use the same field name in
1057 different local scopes.
1059 Nested static and dynamic scopes form lookup paths. These are used for
1060 variant tag and sequence length references. They are used at the variant
1061 and sequence definition site to look up the location of the tag field
1062 associated with a variant, and to lookup up the location of the length
1063 field associated with a sequence.
1065 Variants and sequences can refer to a tag field either using a relative
1066 path or an absolute path. The relative path is relative to the scope in
1067 which the variant or sequence performing the lookup is located.
1068 Relative paths are only allowed to lookup within the same static scope,
1069 which includes its nested static scopes. Lookups targeting parent static
1070 scopes need to be performed with an absolute path.
1072 Absolute path lookups use the full path including the dynamic scope
1073 followed by a "." and then the static scope. Therefore, variants (or
1074 sequences) in lower levels in the dynamic scope (e.g. event context) can
1075 refer to a tag (or length) field located in upper levels (e.g. in the
1076 event header) by specifying, in this case, the associated tag with
1077 <stream.event.header.field_name>. This allows, for instance, the event
1078 context to define a variant referring to the "id" field of the event
1081 The dynamic scope prefixes are thus:
1083 - Trace Environment: <env. >,
1084 - Trace Packet Header: <trace.packet.header. >,
1085 - Stream Packet Context: <stream.packet.context. >,
1086 - Event Header: <stream.event.header. >,
1087 - Stream Event Context: <stream.event.context. >,
1088 - Event Context: <event.context. >,
1089 - Event Payload: <event.fields. >.
1092 The target dynamic scope must be specified explicitly when referring to
1093 a field outside of the static scope (absolute scope reference). No
1094 conflict can occur between relative and dynamic paths, because the
1095 keywords "trace", "stream", and "event" are reserved, and thus
1096 not permitted as field names. It is recommended that field names
1097 clashing with CTF and C99 reserved keywords use an underscore prefix to
1098 eliminate the risk of generating a description containing an invalid
1099 field name. Consequently, fields starting with an underscore should have
1100 their leading underscore removed by the CTF trace readers.
1103 The information available in the dynamic scopes can be thought of as the
1104 current tracing context. At trace production, information about the
1105 current context is saved into the specified scope field levels. At trace
1106 consumption, for each event, the current trace context is therefore
1107 readable by accessing the upper dynamic scopes.
1112 The grammar representing the TSDL meta-data is presented in Appendix C.
1113 TSDL Grammar. This section presents a rather lighter reading that
1114 consists in examples of TSDL meta-data, with template values.
1116 The stream "id" can be left out if there is only one stream in the
1117 trace. The event "id" field can be left out if there is only one event
1121 major = value; /* CTF spec version major number */
1122 minor = value; /* CTF spec version minor number */
1123 uuid = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"; /* Trace UUID */
1124 byte_order = be OR le; /* Endianness (required) */
1125 packet.header := struct {
1133 * The "env" (environment) scope contains assignment expressions. The
1134 * field names and content are implementation-defined.
1137 pid = value; /* example */
1138 proc_name = "name"; /* example */
1144 /* Type 1 - Few event IDs; Type 2 - Many event IDs. See section 6.1. */
1145 event.header := event_header_1 OR event_header_2;
1146 event.context := struct {
1149 packet.context := struct {
1155 name = "event_name";
1156 id = value; /* Numeric identifier within the stream */
1157 stream_id = stream_id;
1167 /* More detail on types in section 4. Types */
1172 * Type declarations behave similarly to the C standard.
1175 typedef aliased_type_specifiers new_type_declarators;
1177 /* e.g.: typedef struct example new_type_name[10]; */
1182 * The "typealias" declaration can be used to give a name (including
1183 * pointer declarator specifier) to a type. It should also be used to
1184 * map basic C types (float, int, unsigned long, ...) to a CTF type.
1185 * Typealias is a superset of "typedef": it also allows assignment of a
1186 * simple variable identifier to a type.
1189 typealias type_class {
1191 } := type_specifiers type_declarator;
1195 * typealias integer {
1199 * } := struct page *;
1201 * typealias integer {
1216 enum name : integer_type {
1222 * Unnamed types, contained within compound type fields, typedef or typealias.
1237 enum : integer_type {
1241 typedef type new_type[length];
1244 type field_name[length];
1247 typedef type new_type[length_type];
1250 type field_name[length_type];
1262 integer_type field_name:size; /* GNU/C bitfield */
1272 Clock metadata allows to describe the clock topology of the system, as
1273 well as to detail each clock parameter. In absence of clock description,
1274 it is assumed that all fields named "timestamp" use the same clock
1275 source, which increments once per nanosecond.
1277 Describing a clock and how it is used by streams is threefold: first,
1278 the clock and clock topology should be described in a "clock"
1279 description block, e.g.:
1282 name = cycle_counter_sync;
1283 uuid = "62189bee-96dc-11e0-91a8-cfa3d89f3923";
1284 description = "Cycle counter synchronized across CPUs";
1285 freq = 1000000000; /* frequency, in Hz */
1286 /* precision in seconds is: 1000 * (1/freq) */
1289 * clock value offset from Epoch is:
1290 * offset_s + (offset * (1/freq))
1292 offset_s = 1326476837;
1297 The mandatory "name" field specifies the name of the clock identifier,
1298 which can later be used as a reference. The optional field "uuid" is the
1299 unique identifier of the clock. It can be used to correlate different
1300 traces that use the same clock. An optional textual description string
1301 can be added with the "description" field. The "freq" field is the
1302 initial frequency of the clock, in Hz. If the "freq" field is not
1303 present, the frequency is assumed to be 1000000000 (providing clock
1304 increment of 1 ns). The optional "precision" field details the
1305 uncertainty on the clock measurements, in (1/freq) units. The "offset_s"
1306 and "offset" fields indicate the offset from POSIX.1 Epoch, 1970-01-01
1307 00:00:00 +0000 (UTC), to the zero of value of the clock. The "offset_s"
1308 field is in seconds. The "offset" field is in (1/freq) units. If any of
1309 the "offset_s" or "offset" field is not present, it is assigned the 0
1310 value. The field "absolute" is TRUE if the clock is a global reference
1311 across different clock uuid (e.g. NTP time). Otherwise, "absolute" is
1312 FALSE, and the clock can be considered as synchronized only with other
1313 clocks that have the same uuid.
1316 Secondly, a reference to this clock should be added within an integer
1320 size = 64; align = 1; signed = false;
1321 map = clock.cycle_counter_sync.value;
1324 Thirdly, stream declarations can reference the clock they use as a
1327 struct packet_context {
1328 uint64_ccnt_t ccnt_begin;
1329 uint64_ccnt_t ccnt_end;
1335 event.header := struct {
1336 uint64_ccnt_t timestamp;
1339 packet.context := struct packet_context;
1342 For a N-bit integer type referring to a clock, if the integer overflows
1343 compared to the N low order bits of the clock prior value, then it is
1344 assumed that one, and only one, overflow occurred. It is therefore
1345 important that events encoding time on a small number of bits happen
1346 frequently enough to detect when more than one N-bit overflow occurs.
1348 In a packet context, clock field names ending with "_begin" and "_end"
1349 have a special meaning: this refers to the time-stamps at, respectively,
1350 the beginning and the end of each packet.
1355 The two following macros keep track of the size of a GNU/C structure without
1356 padding at the end by placing HEADER_END as the last field. A one byte end field
1357 is used for C90 compatibility (C99 flexible arrays could be used here). Note
1358 that this does not affect the effective structure size, which should always be
1359 calculated with the header_sizeof() helper.
1361 #define HEADER_END char end_field
1362 #define header_sizeof(type) offsetof(typeof(type), end_field)
1365 B. Stream Header Rationale
1367 An event stream is divided in contiguous event packets of variable size. These
1368 subdivisions allow the trace analyzer to perform a fast binary search by time
1369 within the stream (typically requiring to index only the event packet headers)
1370 without reading the whole stream. These subdivisions have a variable size to
1371 eliminate the need to transfer the event packet padding when partially filled
1372 event packets must be sent when streaming a trace for live viewing/analysis.
1373 An event packet can contain a certain amount of padding at the end. Dividing
1374 streams into event packets is also useful for network streaming over UDP and
1375 flight recorder mode tracing (a whole event packet can be swapped out of the
1376 buffer atomically for reading).
1378 The stream header is repeated at the beginning of each event packet to allow
1379 flexibility in terms of:
1381 - streaming support,
1382 - allowing arbitrary buffers to be discarded without making the trace
1384 - allow UDP packet loss handling by either dealing with missing event packet
1385 or asking for re-transmission.
1386 - transparently support flight recorder mode,
1387 - transparently support crash dump.
1393 * Common Trace Format (CTF) Trace Stream Description Language (TSDL) Grammar.
1395 * Inspired from the C99 grammar:
1396 * http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf (Annex A)
1397 * and c++1x grammar (draft)
1398 * http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3291.pdf (Annex A)
1400 * Specialized for CTF needs by including only constant and declarations from
1401 * C99 (excluding function declarations), and by adding support for variants,
1402 * sequences and CTF-specific specifiers. Enumeration container types
1403 * semantic is inspired from c++1x enum-base.
1408 1.1) Lexical elements
1454 identifier identifier-nondigit
1457 identifier-nondigit:
1459 universal-character-name
1460 any other implementation-defined characters
1464 [a-zA-Z] /* regular expression */
1467 [0-9] /* regular expression */
1469 1.4) Universal character names
1471 universal-character-name:
1473 \U hex-quad hex-quad
1476 hexadecimal-digit hexadecimal-digit hexadecimal-digit hexadecimal-digit
1482 enumeration-constant
1486 decimal-constant integer-suffix-opt
1487 octal-constant integer-suffix-opt
1488 hexadecimal-constant integer-suffix-opt
1492 decimal-constant digit
1496 octal-constant octal-digit
1498 hexadecimal-constant:
1499 hexadecimal-prefix hexadecimal-digit
1500 hexadecimal-constant hexadecimal-digit
1510 unsigned-suffix long-suffix-opt
1511 unsigned-suffix long-long-suffix
1512 long-suffix unsigned-suffix-opt
1513 long-long-suffix unsigned-suffix-opt
1527 enumeration-constant:
1533 L' c-char-sequence '
1537 c-char-sequence c-char
1540 any member of source charset except single-quote ('), backslash
1541 (\), or new-line character.
1545 simple-escape-sequence
1546 octal-escape-sequence
1547 hexadecimal-escape-sequence
1548 universal-character-name
1550 simple-escape-sequence: one of
1551 \' \" \? \\ \a \b \f \n \r \t \v
1553 octal-escape-sequence:
1555 \ octal-digit octal-digit
1556 \ octal-digit octal-digit octal-digit
1558 hexadecimal-escape-sequence:
1559 \x hexadecimal-digit
1560 hexadecimal-escape-sequence hexadecimal-digit
1562 1.6) String literals
1565 " s-char-sequence-opt "
1566 L" s-char-sequence-opt "
1570 s-char-sequence s-char
1573 any member of source charset except double-quote ("), backslash
1574 (\), or new-line character.
1580 [ ] ( ) { } . -> * + - < > : ; ... = ,
1583 2) Phrase structure grammar
1589 ( unary-expression )
1593 postfix-expression [ unary-expression ]
1594 postfix-expression . identifier
1595 postfix-expressoin -> identifier
1599 unary-operator postfix-expression
1601 unary-operator: one of
1604 assignment-operator:
1607 type-assignment-operator:
1610 constant-expression-range:
1611 unary-expression ... unary-expression
1616 declaration-specifiers declarator-list-opt ;
1619 declaration-specifiers:
1620 storage-class-specifier declaration-specifiers-opt
1621 type-specifier declaration-specifiers-opt
1622 type-qualifier declaration-specifiers-opt
1626 declarator-list , declarator
1628 abstract-declarator-list:
1630 abstract-declarator-list , abstract-declarator
1632 storage-class-specifier:
1655 align ( unary-expression )
1658 struct identifier-opt { struct-or-variant-declaration-list-opt } align-attribute-opt
1659 struct identifier align-attribute-opt
1661 struct-or-variant-declaration-list:
1662 struct-or-variant-declaration
1663 struct-or-variant-declaration-list struct-or-variant-declaration
1665 struct-or-variant-declaration:
1666 specifier-qualifier-list struct-or-variant-declarator-list ;
1667 declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list ;
1668 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list ;
1669 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list ;
1671 specifier-qualifier-list:
1672 type-specifier specifier-qualifier-list-opt
1673 type-qualifier specifier-qualifier-list-opt
1675 struct-or-variant-declarator-list:
1676 struct-or-variant-declarator
1677 struct-or-variant-declarator-list , struct-or-variant-declarator
1679 struct-or-variant-declarator:
1681 declarator-opt : unary-expression
1684 variant identifier-opt variant-tag-opt { struct-or-variant-declaration-list }
1685 variant identifier variant-tag
1688 < unary-expression >
1691 enum identifier-opt { enumerator-list }
1692 enum identifier-opt { enumerator-list , }
1694 enum identifier-opt : declaration-specifiers { enumerator-list }
1695 enum identifier-opt : declaration-specifiers { enumerator-list , }
1699 enumerator-list , enumerator
1702 enumeration-constant
1703 enumeration-constant assignment-operator unary-expression
1704 enumeration-constant assignment-operator constant-expression-range
1710 pointer-opt direct-declarator
1715 direct-declarator [ unary-expression ]
1717 abstract-declarator:
1718 pointer-opt direct-abstract-declarator
1720 direct-abstract-declarator:
1722 ( abstract-declarator )
1723 direct-abstract-declarator [ unary-expression ]
1724 direct-abstract-declarator [ ]
1727 * type-qualifier-list-opt
1728 * type-qualifier-list-opt pointer
1730 type-qualifier-list:
1732 type-qualifier-list type-qualifier
1737 2.3) CTF-specific declarations
1740 clock { ctf-assignment-expression-list-opt }
1741 event { ctf-assignment-expression-list-opt }
1742 stream { ctf-assignment-expression-list-opt }
1743 env { ctf-assignment-expression-list-opt }
1744 trace { ctf-assignment-expression-list-opt }
1745 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1746 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list
1749 floating_point { ctf-assignment-expression-list-opt }
1750 integer { ctf-assignment-expression-list-opt }
1751 string { ctf-assignment-expression-list-opt }
1754 ctf-assignment-expression-list:
1755 ctf-assignment-expression ;
1756 ctf-assignment-expression-list ctf-assignment-expression ;
1758 ctf-assignment-expression:
1759 unary-expression assignment-operator unary-expression
1760 unary-expression type-assignment-operator type-specifier
1761 declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list
1762 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1763 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list