1 Common Trace Format (CTF) Specification (pre-v1.8)
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. These subdivisions have a variable size. An event packet can
122 contain a certain amount of padding at the end. The stream header is
123 repeated at the beginning of each event packet. The rationale for the
124 event stream design choices is explained in Appendix B. Stream Header
127 The event stream header will therefore be referred to as the "event packet
128 header" throughout the rest of this document.
133 Types are organized as type classes. Each type class belong to either of two
134 kind of types: basic types or compound types.
138 A basic type is a scalar type, as described in this section. It includes
139 integers, GNU/C bitfields, enumerations, and floating point values.
141 4.1.1 Type inheritance
143 Type specifications can be inherited to allow deriving types from a
144 type class. For example, see the uint32_t named type derived from the "integer"
145 type class below ("Integers" section). Types have a precise binary
146 representation in the trace. A type class has methods to read and write these
147 types, but must be derived into a type to be usable in an event field.
151 We define "byte-packed" types as aligned on the byte size, namely 8-bit.
152 We define "bit-packed" types as following on the next bit, as defined by the
155 Each basic type must specify its alignment, in bits. Examples of
156 possible alignments are: bit-packed (align = 1), byte-packed (align =
157 8), or word-aligned (e.g. align = 32 or align = 64). The choice depends
158 on the architecture preference and compactness vs performance trade-offs
159 of the implementation. Architectures providing fast unaligned write
160 byte-packed basic types to save space, aligning each type on byte
161 boundaries (8-bit). Architectures with slow unaligned writes align types
162 on specific alignment values. If no specific alignment is declared for a
163 type, it is assumed to be bit-packed for integers with size not multiple
164 of 8 bits and for gcc bitfields. All other basic types are byte-packed
165 by default. It is however recommended to always specify the alignment
166 explicitly. Alignment values must be power of two. Compound types are
167 aligned as specified in their individual specification.
169 TSDL meta-data attribute representation of a specific alignment:
171 align = value; /* value in bits */
175 By default, the native endianness of the source architecture the trace is used.
176 Byte order can be overridden for a basic type by specifying a "byte_order"
177 attribute. Typical use-case is to specify the network byte order (big endian:
178 "be") to save data captured from the network into the trace without conversion.
179 If not specified, the byte order is native.
181 TSDL meta-data representation:
183 byte_order = native OR network OR be OR le; /* network and be are aliases */
187 Type size, in bits, for integers and floats is that returned by "sizeof()" in C
188 multiplied by CHAR_BIT.
189 We require the size of "char" and "unsigned char" types (CHAR_BIT) to be fixed
190 to 8 bits for cross-endianness compatibility.
192 TSDL meta-data representation:
194 size = value; (value is in bits)
198 Signed integers are represented in two-complement. Integer alignment,
199 size, signedness and byte ordering are defined in the TSDL meta-data.
200 Integers aligned on byte size (8-bit) and with length multiple of byte
201 size (8-bit) correspond to the C99 standard integers. In addition,
202 integers with alignment and/or size that are _not_ a multiple of the
203 byte size are permitted; these correspond to the C99 standard bitfields,
204 with the added specification that the CTF integer bitfields have a fixed
205 binary representation. A MIT-licensed reference implementation of the
206 CTF portable bitfields is available at:
208 http://git.efficios.com/?p=babeltrace.git;a=blob;f=include/babeltrace/bitfield.h
210 Binary representation of integers:
212 - On little and big endian:
213 - Within a byte, high bits correspond to an integer high bits, and low bits
214 correspond to low bits.
216 - Integer across multiple bytes are placed from the less significant to the
218 - Consecutive integers are placed from lower bits to higher bits (even within
221 - Integer across multiple bytes are placed from the most significant to the
223 - Consecutive integers are placed from higher bits to lower bits (even within
226 This binary representation is derived from the bitfield implementation in GCC
227 for little and big endian. However, contrary to what GCC does, integers can
228 cross units boundaries (no padding is required). Padding can be explicitly
229 added (see 4.1.6 GNU/C bitfields) to follow the GCC layout if needed.
231 TSDL meta-data representation:
234 signed = true OR false; /* default false */
235 byte_order = native OR network OR be OR le; /* default native */
236 size = value; /* value in bits, no default */
237 align = value; /* value in bits */
238 /* based used for pretty-printing output, default: decimal. */
239 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
240 OR octal OR oct OR o OR 8 OR binary OR b OR 2;
241 /* character encoding, default: none */
242 encoding = none or UTF8 or ASCII;
245 Example of type inheritance (creation of a uint32_t named type):
253 Definition of a named 5-bit signed bitfield:
261 The character encoding field can be used to specify that the integer
262 must be printed as a text character when read. e.g.:
272 4.1.6 GNU/C bitfields
274 The GNU/C bitfields follow closely the integer representation, with a
275 particularity on alignment: if a bitfield cannot fit in the current unit, the
276 unit is padded and the bitfield starts at the following unit. The unit size is
277 defined by the size of the type "unit_type".
279 TSDL meta-data representation:
283 As an example, the following structure declared in C compiled by GCC:
290 The example structure is aligned on the largest element (short). The second
291 bitfield would be aligned on the next unit boundary, because it would not fit in
296 The floating point values byte ordering is defined in the TSDL meta-data.
298 Floating point values follow the IEEE 754-2008 standard interchange formats.
299 Description of the floating point values include the exponent and mantissa size
300 in bits. Some requirements are imposed on the floating point values:
302 - FLT_RADIX must be 2.
303 - mant_dig is the number of digits represented in the mantissa. It is specified
304 by the ISO C99 standard, section 5.2.4, as FLT_MANT_DIG, DBL_MANT_DIG and
305 LDBL_MANT_DIG as defined by <float.h>.
306 - exp_dig is the number of digits represented in the exponent. Given that
307 mant_dig is one bit more than its actual size in bits (leading 1 is not
308 needed) and also given that the sign bit always takes one bit, exp_dig can be
311 - sizeof(float) * CHAR_BIT - FLT_MANT_DIG
312 - sizeof(double) * CHAR_BIT - DBL_MANT_DIG
313 - sizeof(long double) * CHAR_BIT - LDBL_MANT_DIG
315 TSDL meta-data representation:
320 byte_order = native OR network OR be OR le;
324 Example of type inheritance:
326 typealias floating_point {
327 exp_dig = 8; /* sizeof(float) * CHAR_BIT - FLT_MANT_DIG */
328 mant_dig = 24; /* FLT_MANT_DIG */
333 TODO: define NaN, +inf, -inf behavior.
335 Bit-packed, byte-packed or larger alignments can be used for floating
336 point values, similarly to integers.
340 Enumerations are a mapping between an integer type and a table of strings. The
341 numerical representation of the enumeration follows the integer type specified
342 by the meta-data. The enumeration mapping table is detailed in the enumeration
343 description within the meta-data. The mapping table maps inclusive value
344 ranges (or single values) to strings. Instead of being limited to simple
345 "value -> string" mappings, these enumerations map
346 "[ start_value ... end_value ] -> string", which map inclusive ranges of
347 values to strings. An enumeration from the C language can be represented in
348 this format by having the same start_value and end_value for each element, which
349 is in fact a range of size 1. This single-value range is supported without
350 repeating the start and end values with the value = string declaration.
352 enum name : integer_type {
353 somestring = start_value1 ... end_value1,
354 "other string" = start_value2 ... end_value2,
355 yet_another_string, /* will be assigned to end_value2 + 1 */
356 "some other string" = value,
360 If the values are omitted, the enumeration starts at 0 and increment of 1 for
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.
425 A nameless structure can be declared as a field type or as part of a typedef:
431 Alignment for a structure compound type can be forced to a minimum value
432 by adding an "align" specifier after the declaration of a structure
433 body. This attribute is read as: align(value). The value is specified in
434 bits. The structure will be aligned on the maximum value between this
435 attribute and the alignment required by the basic types contained within
442 4.2.2 Variants (Discriminated/Tagged Unions)
444 A CTF variant is a selection between different types. A CTF variant must
445 always be defined within the scope of a structure or within fields
446 contained within a structure (defined recursively). A "tag" enumeration
447 field must appear in either the same static scope, prior to the variant
448 field (in field declaration order), in an upper static scope , or in an
449 upper dynamic scope (see Section 7.3.2). The type selection is indicated
450 by the mapping from the enumeration value to the string used as variant
451 type selector. The field to use as tag is specified by the "tag_field",
452 specified between "< >" after the "variant" keyword for unnamed
453 variants, and after "variant name" for named variants.
455 The alignment of the variant is the alignment of the type as selected by the tag
456 value for the specific instance of the variant. The alignment of the type
457 containing the variant is independent of the variant alignment. The size of the
458 variant is the size as selected by the tag value for the specific instance of
461 Each variant type selector possess a field name, which is a unique
462 identifier within the variant. The identifier is not allowed to use any
463 reserved keyword (see Section C.1.2). Replacing reserved keywords with
464 underscore-prefixed field names is recommended.
466 A named variant declaration followed by its definition within a structure
477 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
479 variant name <tag_field> v;
482 An unnamed variant definition within a structure is expressed by the following
486 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
488 variant <tag_field> {
496 Example of a named variant within a sequence that refers to a single tag field:
505 enum : uint2_t { a, b, c } choice;
507 variant example <choice> v[seqlen];
510 Example of an unnamed variant:
513 enum : uint2_t { a, b, c, d } choice;
514 /* Unrelated fields can be added between the variant and its tag */
527 Example of an unnamed variant within an array:
530 enum : uint2_t { a, b, c } choice;
538 Example of a variant type definition within a structure, where the defined type
539 is then declared within an array of structures. This variant refers to a tag
540 located in an upper static scope. This example clearly shows that a variant
541 type definition referring to the tag "x" uses the closest preceding field from
542 the static scope of the type definition.
545 enum : uint2_t { a, b, c, d } x;
547 typedef variant <x> { /*
548 * "x" refers to the preceding "x" enumeration in the
549 * static scope of the type definition.
557 enum : int { x, y, z } x; /* This enumeration is not used by "v". */
558 example_variant v; /*
559 * "v" uses the "enum : uint2_t { a, b, c, d }"
567 Arrays are fixed-length. Their length is declared in the type
568 declaration within the meta-data. They contain an array of "inner type"
569 elements, which can refer to any type not containing the type of the
570 array being declared (no circular dependency). The length is the number
571 of elements in an array.
573 TSDL meta-data representation of a named array:
575 typedef elem_type name[length];
577 A nameless array can be declared as a field type within a structure, e.g.:
579 uint8_t field_name[10];
581 Arrays are always aligned on their element alignment requirement.
585 Sequences are dynamically-sized arrays. They refer to a a "length"
586 unsigned integer field, which must appear in either the same static scope,
587 prior to the sequence field (in field declaration order), in an upper
588 static scope, or in an upper dynamic scope (see Section 7.3.2). This
589 length field represents the number of elements in the sequence. The
590 sequence per se is an array of "inner type" elements.
592 TSDL meta-data representation for a sequence type definition:
595 unsigned int length_field;
596 typedef elem_type typename[length_field];
597 typename seq_field_name;
600 A sequence can also be declared as a field type, e.g.:
603 unsigned int length_field;
604 long seq_field_name[length_field];
607 Multiple sequences can refer to the same length field, and these length
608 fields can be in a different upper dynamic scope:
610 e.g., assuming the stream.event.header defines:
615 event.header := struct {
624 long seq_a[stream.event.header.seq_len];
625 char seq_b[stream.event.header.seq_len];
629 The sequence elements follow the "array" specifications.
633 Strings are an array of bytes of variable size and are terminated by a '\0'
634 "NULL" character. Their encoding is described in the TSDL meta-data. In
635 absence of encoding attribute information, the default encoding is
638 TSDL meta-data representation of a named string type:
641 encoding = UTF8 OR ASCII;
644 A nameless string type can be declared as a field type:
646 string field_name; /* Use default UTF8 encoding */
648 Strings are always aligned on byte size.
650 5. Event Packet Header
652 The event packet header consists of two parts: the "event packet header"
653 is the same for all streams of a trace. The second part, the "event
654 packet context", is described on a per-stream basis. Both are described
655 in the TSDL meta-data. The packets are aligned on architecture-page-sized
658 Event packet header (all fields are optional, specified by TSDL meta-data):
660 - Magic number (CTF magic number: 0xC1FC1FC1) specifies that this is a
661 CTF packet. This magic number is optional, but when present, it should
662 come at the very beginning of the packet.
663 - Trace UUID, used to ensure the event packet match the meta-data used.
664 (note: we cannot use a meta-data checksum in every cases instead of a
665 UUID because meta-data can be appended to while tracing is active)
666 This field is optional.
667 - Stream ID, used as reference to stream description in meta-data.
668 This field is optional if there is only one stream description in the
669 meta-data, but becomes required if there are more than one stream in
670 the TSDL meta-data description.
672 Event packet context (all fields are optional, specified by TSDL meta-data):
674 - Event packet content size (in bits).
675 - Event packet size (in bits, includes padding).
676 - Event packet content checksum. Checksum excludes the event packet
678 - Per-stream event packet sequence count (to deal with UDP packet loss). The
679 number of significant sequence counter bits should also be present, so
680 wrap-arounds are dealt with correctly.
681 - Time-stamp at the beginning and time-stamp at the end of the event packet.
682 Both timestamps are written in the packet header, but sampled respectively
683 while (or before) writing the first event and while (or after) writing the
684 last event in the packet. The inclusive range between these timestamps should
685 include all event timestamps assigned to events contained within the packet.
686 - Events discarded count
687 - Snapshot of a per-stream free-running counter, counting the number of
688 events discarded that were supposed to be written in the stream prior to
689 the first event in the event packet.
690 * Note: producer-consumer buffer full condition should fill the current
691 event packet with padding so we know exactly where events have been
693 - Lossless compression scheme used for the event packet content. Applied
694 directly to raw data. New types of compression can be added in following
695 versions of the format.
696 0: no compression scheme
700 - Cypher used for the event packet content. Applied after compression.
703 - Checksum scheme used for the event packet content. Applied after encryption.
709 5.1 Event Packet Header Description
711 The event packet header layout is indicated by the trace packet.header
712 field. Here is a recommended structure type for the packet header with
713 the fields typically expected (although these fields are each optional):
715 struct event_packet_header {
723 packet.header := struct event_packet_header;
726 If the magic number is not present, tools such as "file" will have no
727 mean to discover the file type.
729 If the uuid is not present, no validation that the meta-data actually
730 corresponds to the stream is performed.
732 If the stream_id packet header field is missing, the trace can only
733 contain a single stream. Its "id" field can be left out, and its events
734 don't need to declare a "stream_id" field.
737 5.2 Event Packet Context Description
739 Event packet context example. These are declared within the stream declaration
740 in the meta-data. All these fields are optional. If the packet size field is
741 missing, the whole stream only contains a single packet. If the content
742 size field is missing, the packet is filled (no padding). The content
743 and packet sizes include all headers.
745 An example event packet context type:
747 struct event_packet_context {
748 uint64_t timestamp_begin;
749 uint64_t timestamp_end;
751 uint32_t stream_packet_count;
752 uint32_t events_discarded;
754 uint32_t/uint16_t content_size;
755 uint32_t/uint16_t packet_size;
756 uint8_t compression_scheme;
757 uint8_t encryption_scheme;
758 uint8_t checksum_scheme;
764 The overall structure of an event is:
766 1 - Stream Packet Context (as specified by the stream meta-data)
767 2 - Event Header (as specified by the stream meta-data)
768 3 - Stream Event Context (as specified by the stream meta-data)
769 4 - Event Context (as specified by the event meta-data)
770 5 - Event Payload (as specified by the event meta-data)
772 This structure defines an implicit dynamic scoping, where variants
773 located in inner structures (those with a higher number in the listing
774 above) can refer to the fields of outer structures (with lower number in
775 the listing above). See Section 7.3 TSDL Scopes for more detail.
779 Event headers can be described within the meta-data. We hereby propose, as an
780 example, two types of events headers. Type 1 accommodates streams with less than
781 31 event IDs. Type 2 accommodates streams with 31 or more event IDs.
783 One major factor can vary between streams: the number of event IDs assigned to
784 a stream. Luckily, this information tends to stay relatively constant (modulo
785 event registration while trace is being recorded), so we can specify different
786 representations for streams containing few event IDs and streams containing
787 many event IDs, so we end up representing the event ID and time-stamp as
788 densely as possible in each case.
790 The header is extended in the rare occasions where the information cannot be
791 represented in the ranges available in the standard event header. They are also
792 used in the rare occasions where the data required for a field could not be
793 collected: the flag corresponding to the missing field within the missing_fields
794 array is then set to 1.
796 Types uintX_t represent an X-bit unsigned integer, as declared with
799 typealias integer { size = X; align = X; signed = false } := uintX_t;
803 typealias integer { size = X; align = 1; signed = false } := uintX_t;
805 6.1.1 Type 1 - Few event IDs
807 - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
809 - Native architecture byte ordering.
810 - For "compact" selection
811 - Fixed size: 32 bits.
812 - For "extended" selection
813 - Size depends on the architecture and variant alignment.
815 struct event_header_1 {
818 * id 31 is reserved to indicate an extended header.
820 enum : uint5_t { compact = 0 ... 30, extended = 31 } id;
826 uint32_t id; /* 32-bit event IDs */
827 uint64_t timestamp; /* 64-bit timestamps */
830 } align(32); /* or align(8) */
833 6.1.2 Type 2 - Many event IDs
835 - Aligned on 16-bit (or 8-bit if byte-packed, depending on the architecture
837 - Native architecture byte ordering.
838 - For "compact" selection
839 - Size depends on the architecture and variant alignment.
840 - For "extended" selection
841 - Size depends on the architecture and variant alignment.
843 struct event_header_2 {
845 * id: range: 0 - 65534.
846 * id 65535 is reserved to indicate an extended header.
848 enum : uint16_t { compact = 0 ... 65534, extended = 65535 } id;
854 uint32_t id; /* 32-bit event IDs */
855 uint64_t timestamp; /* 64-bit timestamps */
858 } align(16); /* or align(8) */
863 The event context contains information relative to the current event.
864 The choice and meaning of this information is specified by the TSDL
865 stream and event meta-data descriptions. The stream context is applied
866 to all events within the stream. The stream context structure follows
867 the event header. The event context is applied to specific events. Its
868 structure follows the stream context structure.
870 An example of stream-level event context is to save the event payload size with
871 each event, or to save the current PID with each event. These are declared
872 within the stream declaration within the meta-data:
876 event.context := struct {
878 uint16_t payload_size;
882 An example of event-specific event context is to declare a bitmap of missing
883 fields, only appended after the stream event context if the extended event
884 header is selected. NR_FIELDS is the number of fields within the event (a
892 uint1_t missing_fields[NR_FIELDS]; /* missing event fields bitmap */
901 An event payload contains fields specific to a given event type. The fields
902 belonging to an event type are described in the event-specific meta-data
903 within a structure type.
907 No padding at the end of the event payload. This differs from the ISO/C standard
908 for structures, but follows the CTF standard for structures. In a trace, even
909 though it makes sense to align the beginning of a structure, it really makes no
910 sense to add padding at the end of the structure, because structures are usually
911 not followed by a structure of the same type.
913 This trick can be done by adding a zero-length "end" field at the end of the C
914 structures, and by using the offset of this field rather than using sizeof()
915 when calculating the size of a structure (see Appendix "A. Helper macros").
919 The event payload is aligned on the largest alignment required by types
920 contained within the payload. (This follows the ISO/C standard for structures)
923 7. Trace Stream Description Language (TSDL)
925 The Trace Stream Description Language (TSDL) allows expression of the
926 binary trace streams layout in a C99-like Domain Specific Language
932 The trace stream layout description is located in the trace meta-data.
933 The meta-data is itself located in a stream identified by its name:
936 The meta-data description can be expressed in two different formats:
937 text-only and packet-based. The text-only description facilitates
938 generation of meta-data and provides a convenient way to enter the
939 meta-data information by hand. The packet-based meta-data provides the
940 CTF stream packet facilities (checksumming, compression, encryption,
941 network-readiness) for meta-data stream generated and transported by a
944 The text-only meta-data file is a plain-text TSDL description. This file
945 must begin with the following characters to identify the file as a CTF
946 TSDL text-based metadata file (without the double-quotes) :
950 It must be followed by a space, and the version of the specification
951 followed by the CTF trace, e.g.:
955 These characters allow automated discovery of file type and CTF
956 specification version. They are interpreted as a the beginning of a
957 comment by the TSDL metadata parser. The comment can be continued to
958 contain extra commented characters before it is closed.
960 The packet-based meta-data is made of "meta-data packets", which each
961 start with a meta-data packet header. The packet-based meta-data
962 description is detected by reading the magic number "0x75D11D57" at the
963 beginning of the file. This magic number is also used to detect the
964 endianness of the architecture by trying to read the CTF magic number
965 and its counterpart in reversed endianness. The events within the
966 meta-data stream have no event header nor event context. Each event only
967 contains a "sequence" payload, which is a sequence of bits using the
968 "trace.packet.header.content_size" field as a placeholder for its length
969 (the packet header size should be substracted). The formatting of this
970 sequence of bits is a plain-text representation of the TSDL description.
971 Each meta-data packet start with a special packet header, specific to
972 the meta-data stream, which contains, exactly:
974 struct metadata_packet_header {
975 uint32_t magic; /* 0x75D11D57 */
976 uint8_t uuid[16]; /* Unique Universal Identifier */
977 uint32_t checksum; /* 0 if unused */
978 uint32_t content_size; /* in bits */
979 uint32_t packet_size; /* in bits */
980 uint8_t compression_scheme; /* 0 if unused */
981 uint8_t encryption_scheme; /* 0 if unused */
982 uint8_t checksum_scheme; /* 0 if unused */
983 uint8_t major; /* CTF spec version major number */
984 uint8_t minor; /* CTF spec version minor number */
987 The packet-based meta-data can be converted to a text-only meta-data by
988 concatenating all the strings in contains.
990 In the textual representation of the meta-data, the text contained
991 within "/*" and "*/", as well as within "//" and end of line, are
992 treated as comments. Boolean values can be represented as true, TRUE,
993 or 1 for true, and false, FALSE, or 0 for false. Within the string-based
994 meta-data description, the trace UUID is represented as a string of
995 hexadecimal digits and dashes "-". In the event packet header, the trace
996 UUID is represented as an array of bytes.
999 7.2 Declaration vs Definition
1001 A declaration associates a layout to a type, without specifying where
1002 this type is located in the event structure hierarchy (see Section 6).
1003 This therefore includes typedef, typealias, as well as all type
1004 specifiers. In certain circumstances (typedef, structure field and
1005 variant field), a declaration is followed by a declarator, which specify
1006 the newly defined type name (for typedef), or the field name (for
1007 declarations located within structure and variants). Array and sequence,
1008 declared with square brackets ("[" "]"), are part of the declarator,
1009 similarly to C99. The enumeration base type is specified by
1010 ": enum_base", which is part of the type specifier. The variant tag
1011 name, specified between "<" ">", is also part of the type specifier.
1013 A definition associates a type to a location in the event structure
1014 hierarchy (see Section 6). This association is denoted by ":=", as shown
1020 TSDL uses three different types of scoping: a lexical scope is used for
1021 declarations and type definitions, and static and dynamic scopes are
1022 used for variants references to tag fields (with relative and absolute
1023 path lookups) and for sequence references to length fields.
1027 Each of "trace", "stream", "event", "struct" and "variant" have their own
1028 nestable declaration scope, within which types can be declared using "typedef"
1029 and "typealias". A root declaration scope also contains all declarations
1030 located outside of any of the aforementioned declarations. An inner
1031 declaration scope can refer to type declared within its container
1032 lexical scope prior to the inner declaration scope. Redefinition of a
1033 typedef or typealias is not valid, although hiding an upper scope
1034 typedef or typealias is allowed within a sub-scope.
1036 7.3.2 Static and Dynamic Scopes
1038 A local static scope consists in the scope generated by the declaration
1039 of fields within a compound type. A static scope is a local static scope
1040 augmented with the nested sub-static-scopes it contains.
1042 A dynamic scope consists in the static scope augmented with the
1043 implicit event structure definition hierarchy presented at Section 6.
1045 Multiple declarations of the same field name within a local static scope
1046 is not valid. It is however valid to re-use the same field name in
1047 different local scopes.
1049 Nested static and dynamic scopes form lookup paths. These are used for
1050 variant tag and sequence length references. They are used at the variant
1051 and sequence definition site to look up the location of the tag field
1052 associated with a variant, and to lookup up the location of the length
1053 field associated with a sequence.
1055 Variants and sequences can refer to a tag field either using a relative
1056 path or an absolute path. The relative path is relative to the scope in
1057 which the variant or sequence performing the lookup is located.
1058 Relative paths are only allowed to lookup within the same static scope,
1059 which includes its nested static scopes. Lookups targeting parent static
1060 scopes need to be performed with an absolute path.
1062 Absolute path lookups use the full path including the dynamic scope
1063 followed by a "." and then the static scope. Therefore, variants (or
1064 sequences) in lower levels in the dynamic scope (e.g. event context) can
1065 refer to a tag (or length) field located in upper levels (e.g. in the
1066 event header) by specifying, in this case, the associated tag with
1067 <stream.event.header.field_name>. This allows, for instance, the event
1068 context to define a variant referring to the "id" field of the event
1071 The dynamic scope prefixes are thus:
1073 - Trace Packet Header: <trace.packet.header. >,
1074 - Stream Packet Context: <stream.packet.context. >,
1075 - Event Header: <stream.event.header. >,
1076 - Stream Event Context: <stream.event.context. >,
1077 - Event Context: <event.context. >,
1078 - Event Payload: <event.fields. >.
1081 The target dynamic scope must be specified explicitly when referring to
1082 a field outside of the static scope (absolute scope reference). No
1083 conflict can occur between relative and dynamic paths, because the
1084 keywords "trace", "stream", and "event" are reserved, and thus
1085 not permitted as field names. It is recommended that field names
1086 clashing with CTF and C99 reserved keywords use an underscore prefix to
1087 eliminate the risk of generating a description containing an invalid
1090 The information available in the dynamic scopes can be thought of as the
1091 current tracing context. At trace production, information about the
1092 current context is saved into the specified scope field levels. At trace
1093 consumption, for each event, the current trace context is therefore
1094 readable by accessing the upper dynamic scopes.
1099 The grammar representing the TSDL meta-data is presented in Appendix C.
1100 TSDL Grammar. This section presents a rather lighter reading that
1101 consists in examples of TSDL meta-data, with template values.
1103 The stream "id" can be left out if there is only one stream in the
1104 trace. The event "id" field can be left out if there is only one event
1108 major = value; /* Trace format version */
1110 uuid = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"; /* Trace UUID */
1111 byte_order = be OR le; /* Endianness (required) */
1112 packet.header := struct {
1121 /* Type 1 - Few event IDs; Type 2 - Many event IDs. See section 6.1. */
1122 event.header := event_header_1 OR event_header_2;
1123 event.context := struct {
1126 packet.context := struct {
1132 name = "event_name";
1133 id = value; /* Numeric identifier within the stream */
1134 stream_id = stream_id;
1135 loglevel.identifier = "loglevel_identifier";
1136 loglevel.value = value;
1145 /* More detail on types in section 4. Types */
1150 * Type declarations behave similarly to the C standard.
1153 typedef aliased_type_specifiers new_type_declarators;
1155 /* e.g.: typedef struct example new_type_name[10]; */
1160 * The "typealias" declaration can be used to give a name (including
1161 * pointer declarator specifier) to a type. It should also be used to
1162 * map basic C types (float, int, unsigned long, ...) to a CTF type.
1163 * Typealias is a superset of "typedef": it also allows assignment of a
1164 * simple variable identifier to a type.
1167 typealias type_class {
1169 } := type_specifiers type_declarator;
1173 * typealias integer {
1177 * } := struct page *;
1179 * typealias integer {
1194 enum name : integer_type {
1200 * Unnamed types, contained within compound type fields, typedef or typealias.
1215 enum : integer_type {
1219 typedef type new_type[length];
1222 type field_name[length];
1225 typedef type new_type[length_type];
1228 type field_name[length_type];
1240 integer_type field_name:size; /* GNU/C bitfield */
1250 Clock metadata allows to describe the clock topology of the system, as
1251 well as to detail each clock parameter. In absence of clock description,
1252 it is assumed that all fields named "timestamp" use the same clock
1253 source, which increments once per nanosecond.
1255 Describing a clock and how it is used by streams is threefold: first,
1256 the clock and clock topology should be described in a "clock"
1257 description block, e.g.:
1259 typealias integer { size = 32; align = 32; signed = true } := uint32_t;
1261 enum clocks : uint32_t {
1265 clock[cycle_counter] {
1266 uuid = "62189bee-96dc-11e0-91a8-cfa3d89f3923";
1267 description = "Local CPU cycle counter";
1268 freq = 1000000000; /* frequency, in Hz */
1269 /* precision in seconds is: 1000 * (1/freq) */
1271 /* clock value offset from Epoch is: offset * (1/freq) */
1272 offset = 1326476837897235420;
1275 The optional field "uuid" is the unique identifier of the clock. It can
1276 be used to correlate different traces that use the same clock. An
1277 optional textual description string can be added with the "description"
1278 field. The "freq" field is the initial frequency of the clock, in Hz. If
1279 the "freq" field is not present, the frequency is assumed to be
1280 1000000000 (providing clock increment of 1 ns). The optional "precision"
1281 field details the uncertainty on the clock measurements, in (1/freq)
1282 units. The "offset" field indicates the offset from POSIX.1 Epoch,
1283 1970-01-01 00:00:00 +0000 (UTC), to the zero of value of the clock, in
1284 (1/freq) units. If the "offset" field is not present, it is assigned the
1287 Secondly, a reference to this clock should be added within an integer
1291 size = 64; align = 1; signed = false;
1292 map = clock[cycle_counter].value;
1295 Thirdly, stream declarations can reference the clock they use as a
1298 struct packet_context {
1299 uint64_ccnt_t ccnt_begin;
1300 uint64_ccnt_t ccnt_end;
1306 event.header := struct {
1307 uint64_ccnt_t timestamp;
1310 packet.context := struct packet_context;
1313 For a N-bit integer type referring to a clock, if the integer overflows
1314 compared to the N low order bits of the clock prior value, then it is
1315 assumed that one, and only one, overflow occurred. It is therefore
1316 important that events encoding time on a small number of bits happen
1317 frequently enough to detect when more than one N-bit overflow occurs.
1319 In a packet context, clock field names ending with "_begin" and "_end"
1320 have a special meaning: this refers to the time-stamps at, respectively,
1321 the beginning and the end of each packet.
1326 The two following macros keep track of the size of a GNU/C structure without
1327 padding at the end by placing HEADER_END as the last field. A one byte end field
1328 is used for C90 compatibility (C99 flexible arrays could be used here). Note
1329 that this does not affect the effective structure size, which should always be
1330 calculated with the header_sizeof() helper.
1332 #define HEADER_END char end_field
1333 #define header_sizeof(type) offsetof(typeof(type), end_field)
1336 B. Stream Header Rationale
1338 An event stream is divided in contiguous event packets of variable size. These
1339 subdivisions allow the trace analyzer to perform a fast binary search by time
1340 within the stream (typically requiring to index only the event packet headers)
1341 without reading the whole stream. These subdivisions have a variable size to
1342 eliminate the need to transfer the event packet padding when partially filled
1343 event packets must be sent when streaming a trace for live viewing/analysis.
1344 An event packet can contain a certain amount of padding at the end. Dividing
1345 streams into event packets is also useful for network streaming over UDP and
1346 flight recorder mode tracing (a whole event packet can be swapped out of the
1347 buffer atomically for reading).
1349 The stream header is repeated at the beginning of each event packet to allow
1350 flexibility in terms of:
1352 - streaming support,
1353 - allowing arbitrary buffers to be discarded without making the trace
1355 - allow UDP packet loss handling by either dealing with missing event packet
1356 or asking for re-transmission.
1357 - transparently support flight recorder mode,
1358 - transparently support crash dump.
1364 * Common Trace Format (CTF) Trace Stream Description Language (TSDL) Grammar.
1366 * Inspired from the C99 grammar:
1367 * http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf (Annex A)
1368 * and c++1x grammar (draft)
1369 * http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3291.pdf (Annex A)
1371 * Specialized for CTF needs by including only constant and declarations from
1372 * C99 (excluding function declarations), and by adding support for variants,
1373 * sequences and CTF-specific specifiers. Enumeration container types
1374 * semantic is inspired from c++1x enum-base.
1379 1.1) Lexical elements
1424 identifier identifier-nondigit
1427 identifier-nondigit:
1429 universal-character-name
1430 any other implementation-defined characters
1434 [a-zA-Z] /* regular expression */
1437 [0-9] /* regular expression */
1439 1.4) Universal character names
1441 universal-character-name:
1443 \U hex-quad hex-quad
1446 hexadecimal-digit hexadecimal-digit hexadecimal-digit hexadecimal-digit
1452 enumeration-constant
1456 decimal-constant integer-suffix-opt
1457 octal-constant integer-suffix-opt
1458 hexadecimal-constant integer-suffix-opt
1462 decimal-constant digit
1466 octal-constant octal-digit
1468 hexadecimal-constant:
1469 hexadecimal-prefix hexadecimal-digit
1470 hexadecimal-constant hexadecimal-digit
1480 unsigned-suffix long-suffix-opt
1481 unsigned-suffix long-long-suffix
1482 long-suffix unsigned-suffix-opt
1483 long-long-suffix unsigned-suffix-opt
1497 enumeration-constant:
1503 L' c-char-sequence '
1507 c-char-sequence c-char
1510 any member of source charset except single-quote ('), backslash
1511 (\), or new-line character.
1515 simple-escape-sequence
1516 octal-escape-sequence
1517 hexadecimal-escape-sequence
1518 universal-character-name
1520 simple-escape-sequence: one of
1521 \' \" \? \\ \a \b \f \n \r \t \v
1523 octal-escape-sequence:
1525 \ octal-digit octal-digit
1526 \ octal-digit octal-digit octal-digit
1528 hexadecimal-escape-sequence:
1529 \x hexadecimal-digit
1530 hexadecimal-escape-sequence hexadecimal-digit
1532 1.6) String literals
1535 " s-char-sequence-opt "
1536 L" s-char-sequence-opt "
1540 s-char-sequence s-char
1543 any member of source charset except double-quote ("), backslash
1544 (\), or new-line character.
1550 [ ] ( ) { } . -> * + - < > : ; ... = ,
1553 2) Phrase structure grammar
1559 ( unary-expression )
1563 postfix-expression [ unary-expression ]
1564 postfix-expression . identifier
1565 postfix-expressoin -> identifier
1569 unary-operator postfix-expression
1571 unary-operator: one of
1574 assignment-operator:
1577 type-assignment-operator:
1580 constant-expression-range:
1581 unary-expression ... unary-expression
1586 declaration-specifiers declarator-list-opt ;
1589 declaration-specifiers:
1590 storage-class-specifier declaration-specifiers-opt
1591 type-specifier declaration-specifiers-opt
1592 type-qualifier declaration-specifiers-opt
1596 declarator-list , declarator
1598 abstract-declarator-list:
1600 abstract-declarator-list , abstract-declarator
1602 storage-class-specifier:
1625 align ( unary-expression )
1628 struct identifier-opt { struct-or-variant-declaration-list-opt } align-attribute-opt
1629 struct identifier align-attribute-opt
1631 struct-or-variant-declaration-list:
1632 struct-or-variant-declaration
1633 struct-or-variant-declaration-list struct-or-variant-declaration
1635 struct-or-variant-declaration:
1636 specifier-qualifier-list struct-or-variant-declarator-list ;
1637 declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list ;
1638 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list ;
1639 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list ;
1641 specifier-qualifier-list:
1642 type-specifier specifier-qualifier-list-opt
1643 type-qualifier specifier-qualifier-list-opt
1645 struct-or-variant-declarator-list:
1646 struct-or-variant-declarator
1647 struct-or-variant-declarator-list , struct-or-variant-declarator
1649 struct-or-variant-declarator:
1651 declarator-opt : unary-expression
1654 variant identifier-opt variant-tag-opt { struct-or-variant-declaration-list }
1655 variant identifier variant-tag
1658 < unary-expression >
1661 enum identifier-opt { enumerator-list }
1662 enum identifier-opt { enumerator-list , }
1664 enum identifier-opt : declaration-specifiers { enumerator-list }
1665 enum identifier-opt : declaration-specifiers { enumerator-list , }
1669 enumerator-list , enumerator
1672 enumeration-constant
1673 enumeration-constant assignment-operator unary-expression
1674 enumeration-constant assignment-operator constant-expression-range
1680 pointer-opt direct-declarator
1685 direct-declarator [ unary-expression ]
1687 abstract-declarator:
1688 pointer-opt direct-abstract-declarator
1690 direct-abstract-declarator:
1692 ( abstract-declarator )
1693 direct-abstract-declarator [ unary-expression ]
1694 direct-abstract-declarator [ ]
1697 * type-qualifier-list-opt
1698 * type-qualifier-list-opt pointer
1700 type-qualifier-list:
1702 type-qualifier-list type-qualifier
1707 2.3) CTF-specific declarations
1710 event { ctf-assignment-expression-list-opt }
1711 stream { ctf-assignment-expression-list-opt }
1712 trace { ctf-assignment-expression-list-opt }
1713 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1714 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list
1717 floating_point { ctf-assignment-expression-list-opt }
1718 integer { ctf-assignment-expression-list-opt }
1719 string { ctf-assignment-expression-list-opt }
1722 ctf-assignment-expression-list:
1723 ctf-assignment-expression ;
1724 ctf-assignment-expression-list ctf-assignment-expression ;
1726 ctf-assignment-expression:
1727 unary-expression assignment-operator unary-expression
1728 unary-expression type-assignment-operator type-specifier
1729 declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list
1730 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1731 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list