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
70 1. Preliminary definitions
72 - Event Trace: An ordered sequence of events.
73 - Event Stream: An ordered sequence of events, containing a subset of the
75 - Event Packet: A sequence of physically contiguous events within an event
77 - Event: This is the basic entry in a trace. (aka: a trace record).
78 - An event identifier (ID) relates to the class (a type) of event within
80 e.g. event: irq_entry.
81 - An event (or event record) relates to a specific instance of an event
83 e.g. event: irq_entry, at time X, on CPU Y
84 - Source Architecture: Architecture writing the trace.
85 - Reader Architecture: Architecture reading the trace.
88 2. High-level representation of a trace
90 A trace is divided into multiple event streams. Each event stream contains a
91 subset of the trace event types.
93 The final output of the trace, after its generation and optional transport over
94 the network, is expected to be either on permanent or temporary storage in a
95 virtual file system. Because each event stream is appended to while a trace is
96 being recorded, each is associated with a separate file for output. Therefore,
97 a stored trace can be represented as a directory containing one file per stream.
99 Meta-data description associated with the trace contains information on
100 trace event types expressed in the Trace Stream Description Language
101 (TSDL). This language describes:
105 - Per-trace event header description.
106 - Per-stream event header description.
107 - Per-stream event context description.
109 - Event type to stream mapping.
110 - Event type to name mapping.
111 - Event type to ID mapping.
112 - Event context description.
113 - Event fields description.
118 An event stream can be divided into contiguous event packets of variable
119 size. These subdivisions have a variable size. An event packet can
120 contain a certain amount of padding at the end. The stream header is
121 repeated at the beginning of each event packet. The rationale for the
122 event stream design choices is explained in Appendix B. Stream Header
125 The event stream header will therefore be referred to as the "event packet
126 header" throughout the rest of this document.
131 Types are organized as type classes. Each type class belong to either of two
132 kind of types: basic types or compound types.
136 A basic type is a scalar type, as described in this section. It includes
137 integers, GNU/C bitfields, enumerations, and floating point values.
139 4.1.1 Type inheritance
141 Type specifications can be inherited to allow deriving types from a
142 type class. For example, see the uint32_t named type derived from the "integer"
143 type class below ("Integers" section). Types have a precise binary
144 representation in the trace. A type class has methods to read and write these
145 types, but must be derived into a type to be usable in an event field.
149 We define "byte-packed" types as aligned on the byte size, namely 8-bit.
150 We define "bit-packed" types as following on the next bit, as defined by the
153 Each basic type must specify its alignment, in bits. Examples of
154 possible alignments are: bit-packed (align = 1), byte-packed (align =
155 8), or word-aligned (e.g. align = 32 or align = 64). The choice depends
156 on the architecture preference and compactness vs performance trade-offs
157 of the implementation. Architectures providing fast unaligned write
158 byte-packed basic types to save space, aligning each type on byte
159 boundaries (8-bit). Architectures with slow unaligned writes align types
160 on specific alignment values. If no specific alignment is declared for a
161 type, it is assumed to be bit-packed for integers with size not multiple
162 of 8 bits and for gcc bitfields. All other basic types are byte-packed
163 by default. It is however recommended to always specify the alignment
164 explicitly. Alignment values must be power of two. Compound types are
165 aligned as specified in their individual specification.
167 TSDL meta-data attribute representation of a specific alignment:
169 align = value; /* value in bits */
173 By default, the native endianness of the source architecture the trace is used.
174 Byte order can be overridden for a basic type by specifying a "byte_order"
175 attribute. Typical use-case is to specify the network byte order (big endian:
176 "be") to save data captured from the network into the trace without conversion.
177 If not specified, the byte order is native.
179 TSDL meta-data representation:
181 byte_order = native OR network OR be OR le; /* network and be are aliases */
185 Type size, in bits, for integers and floats is that returned by "sizeof()" in C
186 multiplied by CHAR_BIT.
187 We require the size of "char" and "unsigned char" types (CHAR_BIT) to be fixed
188 to 8 bits for cross-endianness compatibility.
190 TSDL meta-data representation:
192 size = value; (value is in bits)
196 Signed integers are represented in two-complement. Integer alignment,
197 size, signedness and byte ordering are defined in the TSDL meta-data.
198 Integers aligned on byte size (8-bit) and with length multiple of byte
199 size (8-bit) correspond to the C99 standard integers. In addition,
200 integers with alignment and/or size that are _not_ a multiple of the
201 byte size are permitted; these correspond to the C99 standard bitfields,
202 with the added specification that the CTF integer bitfields have a fixed
203 binary representation. A MIT-licensed reference implementation of the
204 CTF portable bitfields is available at:
206 http://git.efficios.com/?p=babeltrace.git;a=blob;f=include/babeltrace/bitfield.h
208 Binary representation of integers:
210 - On little and big endian:
211 - Within a byte, high bits correspond to an integer high bits, and low bits
212 correspond to low bits.
214 - Integer across multiple bytes are placed from the less significant to the
216 - Consecutive integers are placed from lower bits to higher bits (even within
219 - Integer across multiple bytes are placed from the most significant to the
221 - Consecutive integers are placed from higher bits to lower bits (even within
224 This binary representation is derived from the bitfield implementation in GCC
225 for little and big endian. However, contrary to what GCC does, integers can
226 cross units boundaries (no padding is required). Padding can be explicitly
227 added (see 4.1.6 GNU/C bitfields) to follow the GCC layout if needed.
229 TSDL meta-data representation:
232 signed = true OR false; /* default false */
233 byte_order = native OR network OR be OR le; /* default native */
234 size = value; /* value in bits, no default */
235 align = value; /* value in bits */
236 /* based used for pretty-printing output, default: decimal. */
237 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
238 OR octal OR oct OR o OR 8 OR binary OR b OR 2;
239 /* character encoding, default: none */
240 encoding = none or UTF8 or ASCII;
243 Example of type inheritance (creation of a uint32_t named type):
251 Definition of a named 5-bit signed bitfield:
259 The character encoding field can be used to specify that the integer
260 must be printed as a text character when read. e.g.:
270 4.1.6 GNU/C bitfields
272 The GNU/C bitfields follow closely the integer representation, with a
273 particularity on alignment: if a bitfield cannot fit in the current unit, the
274 unit is padded and the bitfield starts at the following unit. The unit size is
275 defined by the size of the type "unit_type".
277 TSDL meta-data representation:
281 As an example, the following structure declared in C compiled by GCC:
288 The example structure is aligned on the largest element (short). The second
289 bitfield would be aligned on the next unit boundary, because it would not fit in
294 The floating point values byte ordering is defined in the TSDL meta-data.
296 Floating point values follow the IEEE 754-2008 standard interchange formats.
297 Description of the floating point values include the exponent and mantissa size
298 in bits. Some requirements are imposed on the floating point values:
300 - FLT_RADIX must be 2.
301 - mant_dig is the number of digits represented in the mantissa. It is specified
302 by the ISO C99 standard, section 5.2.4, as FLT_MANT_DIG, DBL_MANT_DIG and
303 LDBL_MANT_DIG as defined by <float.h>.
304 - exp_dig is the number of digits represented in the exponent. Given that
305 mant_dig is one bit more than its actual size in bits (leading 1 is not
306 needed) and also given that the sign bit always takes one bit, exp_dig can be
309 - sizeof(float) * CHAR_BIT - FLT_MANT_DIG
310 - sizeof(double) * CHAR_BIT - DBL_MANT_DIG
311 - sizeof(long double) * CHAR_BIT - LDBL_MANT_DIG
313 TSDL meta-data representation:
318 byte_order = native OR network OR be OR le;
322 Example of type inheritance:
324 typealias floating_point {
325 exp_dig = 8; /* sizeof(float) * CHAR_BIT - FLT_MANT_DIG */
326 mant_dig = 24; /* FLT_MANT_DIG */
331 TODO: define NaN, +inf, -inf behavior.
333 Bit-packed, byte-packed or larger alignments can be used for floating
334 point values, similarly to integers.
338 Enumerations are a mapping between an integer type and a table of strings. The
339 numerical representation of the enumeration follows the integer type specified
340 by the meta-data. The enumeration mapping table is detailed in the enumeration
341 description within the meta-data. The mapping table maps inclusive value
342 ranges (or single values) to strings. Instead of being limited to simple
343 "value -> string" mappings, these enumerations map
344 "[ start_value ... end_value ] -> string", which map inclusive ranges of
345 values to strings. An enumeration from the C language can be represented in
346 this format by having the same start_value and end_value for each element, which
347 is in fact a range of size 1. This single-value range is supported without
348 repeating the start and end values with the value = string declaration.
350 enum name : integer_type {
351 somestring = start_value1 ... end_value1,
352 "other string" = start_value2 ... end_value2,
353 yet_another_string, /* will be assigned to end_value2 + 1 */
354 "some other string" = value,
358 If the values are omitted, the enumeration starts at 0 and increment of 1 for
361 enum name : unsigned int {
369 Overlapping ranges within a single enumeration are implementation defined.
371 A nameless enumeration can be declared as a field type or as part of a typedef:
373 enum : integer_type {
377 Enumerations omitting the container type ": integer_type" use the "int"
378 type (for compatibility with C99). The "int" type must be previously
381 typealias integer { size = 32; align = 32; signed = true } := int;
390 Compound are aggregation of type declarations. Compound types include
391 structures, variant, arrays, sequences, and strings.
395 Structures are aligned on the largest alignment required by basic types
396 contained within the structure. (This follows the ISO/C standard for structures)
398 TSDL meta-data representation of a named structure:
401 field_type field_name;
402 field_type field_name;
409 integer { /* Nameless type */
414 uint64_t second_field_name; /* Named type declared in the meta-data */
417 The fields are placed in a sequence next to each other. They each possess a
418 field name, which is a unique identifier within the structure.
420 A nameless structure can be declared as a field type or as part of a typedef:
426 Alignment for a structure compound type can be forced to a minimum value
427 by adding an "align" specifier after the declaration of a structure
428 body. This attribute is read as: align(value). The value is specified in
429 bits. The structure will be aligned on the maximum value between this
430 attribute and the alignment required by the basic types contained within
437 4.2.2 Variants (Discriminated/Tagged Unions)
439 A CTF variant is a selection between different types. A CTF variant must
440 always be defined within the scope of a structure or within fields
441 contained within a structure (defined recursively). A "tag" enumeration
442 field must appear in either the same lexical scope, prior to the variant
443 field (in field declaration order), in an upper lexical scope (see
444 Section 7.3.1), or in an upper dynamic scope (see Section 7.3.2). The
445 type selection is indicated by the mapping from the enumeration value to
446 the string used as variant type selector. The field to use as tag is
447 specified by the "tag_field", specified between "< >" after the
448 "variant" keyword for unnamed variants, and after "variant name" for
451 The alignment of the variant is the alignment of the type as selected by the tag
452 value for the specific instance of the variant. The alignment of the type
453 containing the variant is independent of the variant alignment. The size of the
454 variant is the size as selected by the tag value for the specific instance of
457 A named variant declaration followed by its definition within a structure
468 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
470 variant name <tag_field> v;
473 An unnamed variant definition within a structure is expressed by the following
477 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
479 variant <tag_field> {
487 Example of a named variant within a sequence that refers to a single tag field:
496 enum : uint2_t { a, b, c } choice;
498 variant example <choice> v[seqlen];
501 Example of an unnamed variant:
504 enum : uint2_t { a, b, c, d } choice;
505 /* Unrelated fields can be added between the variant and its tag */
518 Example of an unnamed variant within an array:
521 enum : uint2_t { a, b, c } choice;
529 Example of a variant type definition within a structure, where the defined type
530 is then declared within an array of structures. This variant refers to a tag
531 located in an upper lexical scope. This example clearly shows that a variant
532 type definition referring to the tag "x" uses the closest preceding field from
533 the lexical scope of the type definition.
536 enum : uint2_t { a, b, c, d } x;
538 typedef variant <x> { /*
539 * "x" refers to the preceding "x" enumeration in the
540 * lexical scope of the type definition.
548 enum : int { x, y, z } x; /* This enumeration is not used by "v". */
549 example_variant v; /*
550 * "v" uses the "enum : uint2_t { a, b, c, d }"
558 Arrays are fixed-length. Their length is declared in the type
559 declaration within the meta-data. They contain an array of "inner type"
560 elements, which can refer to any type not containing the type of the
561 array being declared (no circular dependency). The length is the number
562 of elements in an array.
564 TSDL meta-data representation of a named array:
566 typedef elem_type name[length];
568 A nameless array can be declared as a field type within a structure, e.g.:
570 uint8_t field_name[10];
572 Arrays are always aligned on their element alignment requirement.
576 Sequences are dynamically-sized arrays. They refer to a a "length"
577 unsigned integer field, which must appear in either the same lexical scope,
578 prior to the sequence field (in field declaration order), in an upper
579 lexical scope (see Section 7.3.1), or in an upper dynamic scope (see
580 Section 7.3.2). This length field represents the number of elements in
581 the sequence. The sequence per se is an array of "inner type" elements.
583 TSDL meta-data representation for a sequence type definition:
586 unsigned int length_field;
587 typedef elem_type typename[length_field];
588 typename seq_field_name;
591 A sequence can also be declared as a field type, e.g.:
594 unsigned int length_field;
595 long seq_field_name[length_field];
598 Multiple sequences can refer to the same length field, and these length
599 fields can be in a different upper dynamic scope:
601 e.g., assuming the stream.event.header defines:
606 event.header := struct {
615 long seq_a[stream.event.header.seq_len];
616 char seq_b[stream.event.header.seq_len];
620 The sequence elements follow the "array" specifications.
624 Strings are an array of bytes of variable size and are terminated by a '\0'
625 "NULL" character. Their encoding is described in the TSDL meta-data. In
626 absence of encoding attribute information, the default encoding is
629 TSDL meta-data representation of a named string type:
632 encoding = UTF8 OR ASCII;
635 A nameless string type can be declared as a field type:
637 string field_name; /* Use default UTF8 encoding */
639 Strings are always aligned on byte size.
641 5. Event Packet Header
643 The event packet header consists of two parts: the "event packet header"
644 is the same for all streams of a trace. The second part, the "event
645 packet context", is described on a per-stream basis. Both are described
646 in the TSDL meta-data. The packets are aligned on architecture-page-sized
649 Event packet header (all fields are optional, specified by TSDL meta-data):
651 - Magic number (CTF magic number: 0xC1FC1FC1) specifies that this is a
652 CTF packet. This magic number is optional, but when present, it should
653 come at the very beginning of the packet.
654 - Trace UUID, used to ensure the event packet match the meta-data used.
655 (note: we cannot use a meta-data checksum in every cases instead of a
656 UUID because meta-data can be appended to while tracing is active)
657 This field is optional.
658 - Stream ID, used as reference to stream description in meta-data.
659 This field is optional if there is only one stream description in the
660 meta-data, but becomes required if there are more than one stream in
661 the TSDL meta-data description.
663 Event packet context (all fields are optional, specified by TSDL meta-data):
665 - Event packet content size (in bits).
666 - Event packet size (in bits, includes padding).
667 - Event packet content checksum. Checksum excludes the event packet
669 - Per-stream event packet sequence count (to deal with UDP packet loss). The
670 number of significant sequence counter bits should also be present, so
671 wrap-arounds are dealt with correctly.
672 - Time-stamp at the beginning and time-stamp at the end of the event packet.
673 Both timestamps are written in the packet header, but sampled respectively
674 while (or before) writing the first event and while (or after) writing the
675 last event in the packet. The inclusive range between these timestamps should
676 include all event timestamps assigned to events contained within the packet.
677 - Events discarded count
678 - Snapshot of a per-stream free-running counter, counting the number of
679 events discarded that were supposed to be written in the stream prior to
680 the first event in the event packet.
681 * Note: producer-consumer buffer full condition should fill the current
682 event packet with padding so we know exactly where events have been
684 - Lossless compression scheme used for the event packet content. Applied
685 directly to raw data. New types of compression can be added in following
686 versions of the format.
687 0: no compression scheme
691 - Cypher used for the event packet content. Applied after compression.
694 - Checksum scheme used for the event packet content. Applied after encryption.
700 5.1 Event Packet Header Description
702 The event packet header layout is indicated by the trace packet.header
703 field. Here is a recommended structure type for the packet header with
704 the fields typically expected (although these fields are each optional):
706 struct event_packet_header {
714 packet.header := struct event_packet_header;
717 If the magic number is not present, tools such as "file" will have no
718 mean to discover the file type.
720 If the uuid is not present, no validation that the meta-data actually
721 corresponds to the stream is performed.
723 If the stream_id packet header field is missing, the trace can only
724 contain a single stream. Its "id" field can be left out, and its events
725 don't need to declare a "stream_id" field.
728 5.2 Event Packet Context Description
730 Event packet context example. These are declared within the stream declaration
731 in the meta-data. All these fields are optional. If the packet size field is
732 missing, the whole stream only contains a single packet. If the content
733 size field is missing, the packet is filled (no padding). The content
734 and packet sizes include all headers.
736 An example event packet context type:
738 struct event_packet_context {
739 uint64_t timestamp_begin;
740 uint64_t timestamp_end;
742 uint32_t stream_packet_count;
743 uint32_t events_discarded;
745 uint32_t/uint16_t content_size;
746 uint32_t/uint16_t packet_size;
747 uint8_t stream_packet_count_bits; /* Significant counter bits */
748 uint8_t compression_scheme;
749 uint8_t encryption_scheme;
750 uint8_t checksum_scheme;
756 The overall structure of an event is:
758 1 - Stream Packet Context (as specified by the stream meta-data)
759 2 - Event Header (as specified by the stream meta-data)
760 3 - Stream Event Context (as specified by the stream meta-data)
761 4 - Event Context (as specified by the event meta-data)
762 5 - Event Payload (as specified by the event meta-data)
764 This structure defines an implicit dynamic scoping, where variants
765 located in inner structures (those with a higher number in the listing
766 above) can refer to the fields of outer structures (with lower number in
767 the listing above). See Section 7.3 TSDL Scopes for more detail.
771 Event headers can be described within the meta-data. We hereby propose, as an
772 example, two types of events headers. Type 1 accommodates streams with less than
773 31 event IDs. Type 2 accommodates streams with 31 or more event IDs.
775 One major factor can vary between streams: the number of event IDs assigned to
776 a stream. Luckily, this information tends to stay relatively constant (modulo
777 event registration while trace is being recorded), so we can specify different
778 representations for streams containing few event IDs and streams containing
779 many event IDs, so we end up representing the event ID and time-stamp as
780 densely as possible in each case.
782 The header is extended in the rare occasions where the information cannot be
783 represented in the ranges available in the standard event header. They are also
784 used in the rare occasions where the data required for a field could not be
785 collected: the flag corresponding to the missing field within the missing_fields
786 array is then set to 1.
788 Types uintX_t represent an X-bit unsigned integer, as declared with
791 typealias integer { size = X; align = X; signed = false } := uintX_t;
795 typealias integer { size = X; align = 1; signed = false } := uintX_t;
797 6.1.1 Type 1 - Few event IDs
799 - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
801 - Native architecture byte ordering.
802 - For "compact" selection
803 - Fixed size: 32 bits.
804 - For "extended" selection
805 - Size depends on the architecture and variant alignment.
807 struct event_header_1 {
810 * id 31 is reserved to indicate an extended header.
812 enum : uint5_t { compact = 0 ... 30, extended = 31 } id;
818 uint32_t id; /* 32-bit event IDs */
819 uint64_t timestamp; /* 64-bit timestamps */
822 } align(32); /* or align(8) */
825 6.1.2 Type 2 - Many event IDs
827 - Aligned on 16-bit (or 8-bit if byte-packed, depending on the architecture
829 - Native architecture byte ordering.
830 - For "compact" selection
831 - Size depends on the architecture and variant alignment.
832 - For "extended" selection
833 - Size depends on the architecture and variant alignment.
835 struct event_header_2 {
837 * id: range: 0 - 65534.
838 * id 65535 is reserved to indicate an extended header.
840 enum : uint16_t { compact = 0 ... 65534, extended = 65535 } id;
846 uint32_t id; /* 32-bit event IDs */
847 uint64_t timestamp; /* 64-bit timestamps */
850 } align(16); /* or align(8) */
855 The event context contains information relative to the current event.
856 The choice and meaning of this information is specified by the TSDL
857 stream and event meta-data descriptions. The stream context is applied
858 to all events within the stream. The stream context structure follows
859 the event header. The event context is applied to specific events. Its
860 structure follows the stream context structure.
862 An example of stream-level event context is to save the event payload size with
863 each event, or to save the current PID with each event. These are declared
864 within the stream declaration within the meta-data:
868 event.context := struct {
870 uint16_t payload_size;
874 An example of event-specific event context is to declare a bitmap of missing
875 fields, only appended after the stream event context if the extended event
876 header is selected. NR_FIELDS is the number of fields within the event (a
884 uint1_t missing_fields[NR_FIELDS]; /* missing event fields bitmap */
893 An event payload contains fields specific to a given event type. The fields
894 belonging to an event type are described in the event-specific meta-data
895 within a structure type.
899 No padding at the end of the event payload. This differs from the ISO/C standard
900 for structures, but follows the CTF standard for structures. In a trace, even
901 though it makes sense to align the beginning of a structure, it really makes no
902 sense to add padding at the end of the structure, because structures are usually
903 not followed by a structure of the same type.
905 This trick can be done by adding a zero-length "end" field at the end of the C
906 structures, and by using the offset of this field rather than using sizeof()
907 when calculating the size of a structure (see Appendix "A. Helper macros").
911 The event payload is aligned on the largest alignment required by types
912 contained within the payload. (This follows the ISO/C standard for structures)
915 7. Trace Stream Description Language (TSDL)
917 The Trace Stream Description Language (TSDL) allows expression of the
918 binary trace streams layout in a C99-like Domain Specific Language
924 The trace stream layout description is located in the trace meta-data.
925 The meta-data is itself located in a stream identified by its name:
928 The meta-data description can be expressed in two different formats:
929 text-only and packet-based. The text-only description facilitates
930 generation of meta-data and provides a convenient way to enter the
931 meta-data information by hand. The packet-based meta-data provides the
932 CTF stream packet facilities (checksumming, compression, encryption,
933 network-readiness) for meta-data stream generated and transported by a
936 The text-only meta-data file is a plain text TSDL description.
938 The packet-based meta-data is made of "meta-data packets", which each
939 start with a meta-data packet header. The packet-based meta-data
940 description is detected by reading the magic number "0x75D11D57" at the
941 beginning of the file. This magic number is also used to detect the
942 endianness of the architecture by trying to read the CTF magic number
943 and its counterpart in reversed endianness. The events within the
944 meta-data stream have no event header nor event context. Each event only
945 contains a "sequence" payload, which is a sequence of bits using the
946 "trace.packet.header.content_size" field as a placeholder for its length
947 (the packet header size should be substracted). The formatting of this
948 sequence of bits is a plain-text representation of the TSDL description.
949 Each meta-data packet start with a special packet header, specific to
950 the meta-data stream, which contains, exactly:
952 struct metadata_packet_header {
953 uint32_t magic; /* 0x75D11D57 */
954 uint8_t uuid[16]; /* Unique Universal Identifier */
955 uint32_t checksum; /* 0 if unused */
956 uint32_t content_size; /* in bits */
957 uint32_t packet_size; /* in bits */
958 uint8_t compression_scheme; /* 0 if unused */
959 uint8_t encryption_scheme; /* 0 if unused */
960 uint8_t checksum_scheme; /* 0 if unused */
963 The packet-based meta-data can be converted to a text-only meta-data by
964 concatenating all the strings in contains.
966 In the textual representation of the meta-data, the text contained
967 within "/*" and "*/", as well as within "//" and end of line, are
968 treated as comments. Boolean values can be represented as true, TRUE,
969 or 1 for true, and false, FALSE, or 0 for false. Within the string-based
970 meta-data description, the trace UUID is represented as a string of
971 hexadecimal digits and dashes "-". In the event packet header, the trace
972 UUID is represented as an array of bytes.
975 7.2 Declaration vs Definition
977 A declaration associates a layout to a type, without specifying where
978 this type is located in the event structure hierarchy (see Section 6).
979 This therefore includes typedef, typealias, as well as all type
980 specifiers. In certain circumstances (typedef, structure field and
981 variant field), a declaration is followed by a declarator, which specify
982 the newly defined type name (for typedef), or the field name (for
983 declarations located within structure and variants). Array and sequence,
984 declared with square brackets ("[" "]"), are part of the declarator,
985 similarly to C99. The enumeration base type is specified by
986 ": enum_base", which is part of the type specifier. The variant tag
987 name, specified between "<" ">", is also part of the type specifier.
989 A definition associates a type to a location in the event structure
990 hierarchy (see Section 6). This association is denoted by ":=", as shown
996 TSDL uses two different types of scoping: a lexical scope is used for
997 declarations and type definitions, and a dynamic scope is used for
998 variants references to tag fields and for sequence references to length
1003 Each of "trace", "stream", "event", "struct" and "variant" have their own
1004 nestable declaration scope, within which types can be declared using "typedef"
1005 and "typealias". A root declaration scope also contains all declarations
1006 located outside of any of the aforementioned declarations. An inner
1007 declaration scope can refer to type declared within its container
1008 lexical scope prior to the inner declaration scope. Redefinition of a
1009 typedef or typealias is not valid, although hiding an upper scope
1010 typedef or typealias is allowed within a sub-scope.
1014 A dynamic scope consists in the lexical scope augmented with the
1015 implicit event structure definition hierarchy presented at Section 6.
1016 The dynamic scope is used for variant tag and sequence length
1017 definitions. It is used at definition time to look up the location of
1018 the tag field associated with a variant, and to lookup up the location
1019 of the length field associated with a sequence.
1021 Therefore, variants (or sequences) in lower levels in the dynamic scope
1022 (e.g. event context) can refer to a tag (or length) field located in
1023 upper levels (e.g. in the event header) by specifying, in this case, the
1024 associated tag with <header.field_name>. This allows, for instance, the
1025 event context to define a variant referring to the "id" field of the
1026 event header as selector.
1028 The target dynamic scope must be specified explicitly when referring to
1029 a field outside of the local static scope. The dynamic scope prefixes
1032 - Trace Packet Header: <trace.packet.header. >,
1033 - Stream Packet Context: <stream.packet.context. >,
1034 - Event Header: <stream.event.header. >,
1035 - Stream Event Context: <stream.event.context. >,
1036 - Event Context: <event.context. >,
1037 - Event Payload: <event.fields. >.
1039 Multiple declarations of the same field name within a single scope is
1040 not valid. It is however valid to re-use the same field name in
1041 different scopes. There is no possible conflict, because the dynamic
1042 scope must be specified when a variant refers to a tag field located in
1043 a different dynamic scope.
1045 The information available in the dynamic scopes can be thought of as the
1046 current tracing context. At trace production, information about the
1047 current context is saved into the specified scope field levels. At trace
1048 consumption, for each event, the current trace context is therefore
1049 readable by accessing the upper dynamic scopes.
1054 The grammar representing the TSDL meta-data is presented in Appendix C.
1055 TSDL Grammar. This section presents a rather lighter reading that
1056 consists in examples of TSDL meta-data, with template values.
1058 The stream "id" can be left out if there is only one stream in the
1059 trace. The event "id" field can be left out if there is only one event
1063 major = value; /* Trace format version */
1065 uuid = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"; /* Trace UUID */
1066 byte_order = be OR le; /* Endianness (required) */
1067 packet.header := struct {
1076 /* Type 1 - Few event IDs; Type 2 - Many event IDs. See section 6.1. */
1077 event.header := event_header_1 OR event_header_2;
1078 event.context := struct {
1081 packet.context := struct {
1088 id = value; /* Numeric identifier within the stream */
1089 stream_id = stream_id;
1098 /* More detail on types in section 4. Types */
1103 * Type declarations behave similarly to the C standard.
1106 typedef aliased_type_specifiers new_type_declarators;
1108 /* e.g.: typedef struct example new_type_name[10]; */
1113 * The "typealias" declaration can be used to give a name (including
1114 * pointer declarator specifier) to a type. It should also be used to
1115 * map basic C types (float, int, unsigned long, ...) to a CTF type.
1116 * Typealias is a superset of "typedef": it also allows assignment of a
1117 * simple variable identifier to a type.
1120 typealias type_class {
1122 } := type_specifiers type_declarator;
1126 * typealias integer {
1130 * } := struct page *;
1132 * typealias integer {
1147 enum name : integer_type {
1153 * Unnamed types, contained within compound type fields, typedef or typealias.
1168 enum : integer_type {
1172 typedef type new_type[length];
1175 type field_name[length];
1178 typedef type new_type[length_type];
1181 type field_name[length_type];
1193 integer_type field_name:size; /* GNU/C bitfield */
1203 The two following macros keep track of the size of a GNU/C structure without
1204 padding at the end by placing HEADER_END as the last field. A one byte end field
1205 is used for C90 compatibility (C99 flexible arrays could be used here). Note
1206 that this does not affect the effective structure size, which should always be
1207 calculated with the header_sizeof() helper.
1209 #define HEADER_END char end_field
1210 #define header_sizeof(type) offsetof(typeof(type), end_field)
1213 B. Stream Header Rationale
1215 An event stream is divided in contiguous event packets of variable size. These
1216 subdivisions allow the trace analyzer to perform a fast binary search by time
1217 within the stream (typically requiring to index only the event packet headers)
1218 without reading the whole stream. These subdivisions have a variable size to
1219 eliminate the need to transfer the event packet padding when partially filled
1220 event packets must be sent when streaming a trace for live viewing/analysis.
1221 An event packet can contain a certain amount of padding at the end. Dividing
1222 streams into event packets is also useful for network streaming over UDP and
1223 flight recorder mode tracing (a whole event packet can be swapped out of the
1224 buffer atomically for reading).
1226 The stream header is repeated at the beginning of each event packet to allow
1227 flexibility in terms of:
1229 - streaming support,
1230 - allowing arbitrary buffers to be discarded without making the trace
1232 - allow UDP packet loss handling by either dealing with missing event packet
1233 or asking for re-transmission.
1234 - transparently support flight recorder mode,
1235 - transparently support crash dump.
1241 * Common Trace Format (CTF) Trace Stream Description Language (TSDL) Grammar.
1243 * Inspired from the C99 grammar:
1244 * http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf (Annex A)
1245 * and c++1x grammar (draft)
1246 * http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3291.pdf (Annex A)
1248 * Specialized for CTF needs by including only constant and declarations from
1249 * C99 (excluding function declarations), and by adding support for variants,
1250 * sequences and CTF-specific specifiers. Enumeration container types
1251 * semantic is inspired from c++1x enum-base.
1256 1.1) Lexical elements
1300 identifier identifier-nondigit
1303 identifier-nondigit:
1305 universal-character-name
1306 any other implementation-defined characters
1310 [a-zA-Z] /* regular expression */
1313 [0-9] /* regular expression */
1315 1.4) Universal character names
1317 universal-character-name:
1319 \U hex-quad hex-quad
1322 hexadecimal-digit hexadecimal-digit hexadecimal-digit hexadecimal-digit
1328 enumeration-constant
1332 decimal-constant integer-suffix-opt
1333 octal-constant integer-suffix-opt
1334 hexadecimal-constant integer-suffix-opt
1338 decimal-constant digit
1342 octal-constant octal-digit
1344 hexadecimal-constant:
1345 hexadecimal-prefix hexadecimal-digit
1346 hexadecimal-constant hexadecimal-digit
1356 unsigned-suffix long-suffix-opt
1357 unsigned-suffix long-long-suffix
1358 long-suffix unsigned-suffix-opt
1359 long-long-suffix unsigned-suffix-opt
1373 enumeration-constant:
1379 L' c-char-sequence '
1383 c-char-sequence c-char
1386 any member of source charset except single-quote ('), backslash
1387 (\), or new-line character.
1391 simple-escape-sequence
1392 octal-escape-sequence
1393 hexadecimal-escape-sequence
1394 universal-character-name
1396 simple-escape-sequence: one of
1397 \' \" \? \\ \a \b \f \n \r \t \v
1399 octal-escape-sequence:
1401 \ octal-digit octal-digit
1402 \ octal-digit octal-digit octal-digit
1404 hexadecimal-escape-sequence:
1405 \x hexadecimal-digit
1406 hexadecimal-escape-sequence hexadecimal-digit
1408 1.6) String literals
1411 " s-char-sequence-opt "
1412 L" s-char-sequence-opt "
1416 s-char-sequence s-char
1419 any member of source charset except double-quote ("), backslash
1420 (\), or new-line character.
1426 [ ] ( ) { } . -> * + - < > : ; ... = ,
1429 2) Phrase structure grammar
1435 ( unary-expression )
1439 postfix-expression [ unary-expression ]
1440 postfix-expression . identifier
1441 postfix-expressoin -> identifier
1445 unary-operator postfix-expression
1447 unary-operator: one of
1450 assignment-operator:
1453 type-assignment-operator:
1456 constant-expression-range:
1457 unary-expression ... unary-expression
1462 declaration-specifiers declarator-list-opt ;
1465 declaration-specifiers:
1466 storage-class-specifier declaration-specifiers-opt
1467 type-specifier declaration-specifiers-opt
1468 type-qualifier declaration-specifiers-opt
1472 declarator-list , declarator
1474 abstract-declarator-list:
1476 abstract-declarator-list , abstract-declarator
1478 storage-class-specifier:
1501 align ( unary-expression )
1504 struct identifier-opt { struct-or-variant-declaration-list-opt } align-attribute-opt
1505 struct identifier align-attribute-opt
1507 struct-or-variant-declaration-list:
1508 struct-or-variant-declaration
1509 struct-or-variant-declaration-list struct-or-variant-declaration
1511 struct-or-variant-declaration:
1512 specifier-qualifier-list struct-or-variant-declarator-list ;
1513 declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list ;
1514 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list ;
1515 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list ;
1517 specifier-qualifier-list:
1518 type-specifier specifier-qualifier-list-opt
1519 type-qualifier specifier-qualifier-list-opt
1521 struct-or-variant-declarator-list:
1522 struct-or-variant-declarator
1523 struct-or-variant-declarator-list , struct-or-variant-declarator
1525 struct-or-variant-declarator:
1527 declarator-opt : unary-expression
1530 variant identifier-opt variant-tag-opt { struct-or-variant-declaration-list }
1531 variant identifier variant-tag
1537 enum identifier-opt { enumerator-list }
1538 enum identifier-opt { enumerator-list , }
1540 enum identifier-opt : declaration-specifiers { enumerator-list }
1541 enum identifier-opt : declaration-specifiers { enumerator-list , }
1545 enumerator-list , enumerator
1548 enumeration-constant
1549 enumeration-constant assignment-operator unary-expression
1550 enumeration-constant assignment-operator constant-expression-range
1556 pointer-opt direct-declarator
1561 direct-declarator [ unary-expression ]
1563 abstract-declarator:
1564 pointer-opt direct-abstract-declarator
1566 direct-abstract-declarator:
1568 ( abstract-declarator )
1569 direct-abstract-declarator [ unary-expression ]
1570 direct-abstract-declarator [ ]
1573 * type-qualifier-list-opt
1574 * type-qualifier-list-opt pointer
1576 type-qualifier-list:
1578 type-qualifier-list type-qualifier
1583 2.3) CTF-specific declarations
1586 event { ctf-assignment-expression-list-opt }
1587 stream { ctf-assignment-expression-list-opt }
1588 trace { ctf-assignment-expression-list-opt }
1589 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1590 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list
1593 floating_point { ctf-assignment-expression-list-opt }
1594 integer { ctf-assignment-expression-list-opt }
1595 string { ctf-assignment-expression-list-opt }
1598 ctf-assignment-expression-list:
1599 ctf-assignment-expression ;
1600 ctf-assignment-expression-list ctf-assignment-expression ;
1602 ctf-assignment-expression:
1603 unary-expression assignment-operator unary-expression
1604 unary-expression type-assignment-operator type-specifier
1605 declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list
1606 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1607 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list