1 Common Trace Format (CTF) Specification (v1.8.2)
3 Mathieu Desnoyers, EfficiOS Inc.
5 The goal of the present document is to specify a trace format that suits the
6 needs of the embedded, telecom, high-performance and kernel communities. It is
7 based on the Common Trace Format Requirements (v1.4) document. It is designed to
8 allow traces to be natively generated by the Linux kernel, Linux user-space
9 applications written in C/C++, and hardware components. One major element of
10 CTF is the Trace Stream Description Language (TSDL) which flexibility
11 enables description of various binary trace stream layouts.
13 The latest version of this document can be found at:
15 git tree: git://git.efficios.com/ctf.git
16 gitweb: http://git.efficios.com/?p=ctf.git
18 A reference implementation of a library to read and write this trace format is
19 being implemented within the BabelTrace project, a converter between trace
20 formats. The development tree is available at:
22 git tree: git://git.efficios.com/babeltrace.git
23 gitweb: http://git.efficios.com/?p=babeltrace.git
25 The CE Workgroup of the Linux Foundation, Ericsson, and EfficiOS have
31 1. Preliminary definitions
32 2. High-level representation of a trace
36 4.1.1 Type inheritance
46 4.2.2 Variants (Discriminated/Tagged Unions)
50 5. Event Packet Header
51 5.1 Event Packet Header Description
52 5.2 Event Packet Context Description
55 6.1.1 Type 1 - Few event IDs
56 6.1.2 Type 2 - Many event IDs
61 7. Trace Stream Description Language (TSDL)
63 7.2 Declaration vs Definition
66 7.3.2 Static and Dynamic Scopes
71 1. Preliminary definitions
73 - Event Trace: An ordered sequence of events.
74 - Event Stream: An ordered sequence of events, containing a subset of the
76 - Event Packet: A sequence of physically contiguous events within an event
78 - Event: This is the basic entry in a trace. (aka: a trace record).
79 - An event identifier (ID) relates to the class (a type) of event within
81 e.g. event: irq_entry.
82 - An event (or event record) relates to a specific instance of an event
84 e.g. event: irq_entry, at time X, on CPU Y
85 - Source Architecture: Architecture writing the trace.
86 - Reader Architecture: Architecture reading the trace.
89 2. High-level representation of a trace
91 A trace is divided into multiple event streams. Each event stream contains a
92 subset of the trace event types.
94 The final output of the trace, after its generation and optional transport over
95 the network, is expected to be either on permanent or temporary storage in a
96 virtual file system. Because each event stream is appended to while a trace is
97 being recorded, each is associated with a distinct set of files for
98 output. Therefore, a stored trace can be represented as a directory
99 containing zero, one or more files per stream.
101 Meta-data description associated with the trace contains information on
102 trace event types expressed in the Trace Stream Description Language
103 (TSDL). This language describes:
107 - Per-trace event header description.
108 - Per-stream event header description.
109 - Per-stream event context description.
111 - Event type to stream mapping.
112 - Event type to name mapping.
113 - Event type to ID mapping.
114 - Event context description.
115 - Event fields description.
120 An event stream can be divided into contiguous event packets of variable
121 size. An event packet can contain a certain amount of padding at the
122 end. The stream header is repeated at the beginning of each event
123 packet. The rationale for the event stream design choices is explained
124 in Appendix B. Stream Header Rationale.
126 The event stream header will therefore be referred to as the "event packet
127 header" throughout the rest of this document.
132 Types are organized as type classes. Each type class belong to either of two
133 kind of types: basic types or compound types.
137 A basic type is a scalar type, as described in this section. It includes
138 integers, GNU/C bitfields, enumerations, and floating point values.
140 4.1.1 Type inheritance
142 Type specifications can be inherited to allow deriving types from a
143 type class. For example, see the uint32_t named type derived from the "integer"
144 type class below ("Integers" section). Types have a precise binary
145 representation in the trace. A type class has methods to read and write these
146 types, but must be derived into a type to be usable in an event field.
150 We define "byte-packed" types as aligned on the byte size, namely 8-bit.
151 We define "bit-packed" types as following on the next bit, as defined by the
154 Each basic type must specify its alignment, in bits. Examples of
155 possible alignments are: bit-packed (align = 1), byte-packed (align =
156 8), or word-aligned (e.g. align = 32 or align = 64). The choice depends
157 on the architecture preference and compactness vs performance trade-offs
158 of the implementation. Architectures providing fast unaligned write
159 byte-packed basic types to save space, aligning each type on byte
160 boundaries (8-bit). Architectures with slow unaligned writes align types
161 on specific alignment values. If no specific alignment is declared for a
162 type, it is assumed to be bit-packed for integers with size not multiple
163 of 8 bits and for gcc bitfields. All other basic types are byte-packed
164 by default. It is however recommended to always specify the alignment
165 explicitly. Alignment values must be power of two. Compound types are
166 aligned as specified in their individual specification.
168 TSDL meta-data attribute representation of a specific alignment:
170 align = value; /* value in bits */
174 By default, byte order of a basic type is the byte order described in
175 the trace description. It can be overridden by specifying a
176 "byte_order" attribute for a basic type. Typical use-case is to specify
177 the network byte order (big endian: "be") to save data captured from the
178 network into the trace without conversion.
180 TSDL meta-data representation:
182 byte_order = native OR network OR be OR le; /* network and be are aliases */
184 The "native" keyword selects the byte order described in the trace
185 description. The "network" byte order is an alias for big endian.
187 Even though the trace description section is not per se a type, for sake
188 of clarity, it should be noted that "native" and "network" byte orders
189 are only allowed within type declaration. The byte_order specified in
190 the trace description section only accepts "be" or "le" values.
194 Type size, in bits, for integers and floats is that returned by "sizeof()" in C
195 multiplied by CHAR_BIT.
196 We require the size of "char" and "unsigned char" types (CHAR_BIT) to be fixed
197 to 8 bits for cross-endianness compatibility.
199 TSDL meta-data representation:
201 size = value; (value is in bits)
205 Signed integers are represented in two-complement. Integer alignment,
206 size, signedness and byte ordering are defined in the TSDL meta-data.
207 Integers aligned on byte size (8-bit) and with length multiple of byte
208 size (8-bit) correspond to the C99 standard integers. In addition,
209 integers with alignment and/or size that are _not_ a multiple of the
210 byte size are permitted; these correspond to the C99 standard bitfields,
211 with the added specification that the CTF integer bitfields have a fixed
212 binary representation. A MIT-licensed reference implementation of the
213 CTF portable bitfields is available at:
215 http://git.efficios.com/?p=babeltrace.git;a=blob;f=include/babeltrace/bitfield.h
217 Binary representation of integers:
219 - On little and big endian:
220 - Within a byte, high bits correspond to an integer high bits, and low bits
221 correspond to low bits.
223 - Integer across multiple bytes are placed from the less significant to the
225 - Consecutive integers are placed from lower bits to higher bits (even within
228 - Integer across multiple bytes are placed from the most significant to the
230 - Consecutive integers are placed from higher bits to lower bits (even within
233 This binary representation is derived from the bitfield implementation in GCC
234 for little and big endian. However, contrary to what GCC does, integers can
235 cross units boundaries (no padding is required). Padding can be explicitly
236 added (see 4.1.6 GNU/C bitfields) to follow the GCC layout if needed.
238 TSDL meta-data representation:
241 signed = true OR false; /* default false */
242 byte_order = native OR network OR be OR le; /* default native */
243 size = value; /* value in bits, no default */
244 align = value; /* value in bits */
245 /* based used for pretty-printing output, default: decimal. */
246 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
247 OR octal OR oct OR o OR 8 OR binary OR b OR 2;
248 /* character encoding, default: none */
249 encoding = none or UTF8 or ASCII;
252 Example of type inheritance (creation of a uint32_t named type):
260 Definition of a named 5-bit signed bitfield:
268 The character encoding field can be used to specify that the integer
269 must be printed as a text character when read. e.g.:
279 4.1.6 GNU/C bitfields
281 The GNU/C bitfields follow closely the integer representation, with a
282 particularity on alignment: if a bitfield cannot fit in the current unit, the
283 unit is padded and the bitfield starts at the following unit. The unit size is
284 defined by the size of the type "unit_type".
286 TSDL meta-data representation:
290 As an example, the following structure declared in C compiled by GCC:
297 The example structure is aligned on the largest element (short). The second
298 bitfield would be aligned on the next unit boundary, because it would not fit in
303 The floating point values byte ordering is defined in the TSDL meta-data.
305 Floating point values follow the IEEE 754-2008 standard interchange formats.
306 Description of the floating point values include the exponent and mantissa size
307 in bits. Some requirements are imposed on the floating point values:
309 - FLT_RADIX must be 2.
310 - mant_dig is the number of digits represented in the mantissa. It is specified
311 by the ISO C99 standard, section 5.2.4, as FLT_MANT_DIG, DBL_MANT_DIG and
312 LDBL_MANT_DIG as defined by <float.h>.
313 - exp_dig is the number of digits represented in the exponent. Given that
314 mant_dig is one bit more than its actual size in bits (leading 1 is not
315 needed) and also given that the sign bit always takes one bit, exp_dig can be
318 - sizeof(float) * CHAR_BIT - FLT_MANT_DIG
319 - sizeof(double) * CHAR_BIT - DBL_MANT_DIG
320 - sizeof(long double) * CHAR_BIT - LDBL_MANT_DIG
322 TSDL meta-data representation:
327 byte_order = native OR network OR be OR le;
331 Example of type inheritance:
333 typealias floating_point {
334 exp_dig = 8; /* sizeof(float) * CHAR_BIT - FLT_MANT_DIG */
335 mant_dig = 24; /* FLT_MANT_DIG */
340 TODO: define NaN, +inf, -inf behavior.
342 Bit-packed, byte-packed or larger alignments can be used for floating
343 point values, similarly to integers.
347 Enumerations are a mapping between an integer type and a table of strings. The
348 numerical representation of the enumeration follows the integer type specified
349 by the meta-data. The enumeration mapping table is detailed in the enumeration
350 description within the meta-data. The mapping table maps inclusive value
351 ranges (or single values) to strings. Instead of being limited to simple
352 "value -> string" mappings, these enumerations map
353 "[ start_value ... end_value ] -> string", which map inclusive ranges of
354 values to strings. An enumeration from the C language can be represented in
355 this format by having the same start_value and end_value for each element, which
356 is in fact a range of size 1. This single-value range is supported without
357 repeating the start and end values with the value = string declaration.
359 enum name : integer_type {
360 somestring = start_value1 ... end_value1,
361 "other string" = start_value2 ... end_value2,
362 yet_another_string, /* will be assigned to end_value2 + 1 */
363 "some other string" = value,
367 If the values are omitted, the enumeration starts at 0 and increment of 1 for
368 each entry. An entry with omitted value that follows a range entry takes
369 as value the end_value of the previous range + 1:
371 enum name : unsigned int {
379 Overlapping ranges within a single enumeration are implementation defined.
381 A nameless enumeration can be declared as a field type or as part of a typedef:
383 enum : integer_type {
387 Enumerations omitting the container type ": integer_type" use the "int"
388 type (for compatibility with C99). The "int" type must be previously
391 typealias integer { size = 32; align = 32; signed = true } := int;
400 Compound are aggregation of type declarations. Compound types include
401 structures, variant, arrays, sequences, and strings.
405 Structures are aligned on the largest alignment required by basic types
406 contained within the structure. (This follows the ISO/C standard for structures)
408 TSDL meta-data representation of a named structure:
411 field_type field_name;
412 field_type field_name;
419 integer { /* Nameless type */
424 uint64_t second_field_name; /* Named type declared in the meta-data */
427 The fields are placed in a sequence next to each other. They each
428 possess a field name, which is a unique identifier within the structure.
429 The identifier is not allowed to use any reserved keyword
430 (see Section C.1.2). Replacing reserved keywords with
431 underscore-prefixed field names is recommended. Fields starting with an
432 underscore should have their leading underscore removed by the CTF trace
435 A nameless structure can be declared as a field type or as part of a typedef:
441 Alignment for a structure compound type can be forced to a minimum value
442 by adding an "align" specifier after the declaration of a structure
443 body. This attribute is read as: align(value). The value is specified in
444 bits. The structure will be aligned on the maximum value between this
445 attribute and the alignment required by the basic types contained within
452 4.2.2 Variants (Discriminated/Tagged Unions)
454 A CTF variant is a selection between different types. A CTF variant must
455 always be defined within the scope of a structure or within fields
456 contained within a structure (defined recursively). A "tag" enumeration
457 field must appear in either the same static scope, prior to the variant
458 field (in field declaration order), in an upper static scope , or in an
459 upper dynamic scope (see Section 7.3.2). The type selection is indicated
460 by the mapping from the enumeration value to the string used as variant
461 type selector. The field to use as tag is specified by the "tag_field",
462 specified between "< >" after the "variant" keyword for unnamed
463 variants, and after "variant name" for named variants.
465 The alignment of the variant is the alignment of the type as selected by
466 the tag value for the specific instance of the variant. The size of the
467 variant is the size as selected by the tag value for the specific
468 instance of the variant.
470 The alignment of the type containing the variant is independent of the
471 variant alignment. For instance, if a structure contains two fields, a
472 32-bit integer, aligned on 32 bits, and a variant, which contains two
473 choices: either a 32-bit field, aligned on 32 bits, or a 64-bit field,
474 aligned on 64 bits, the alignment of the outmost structure will be
475 32-bit (the alignment of its largest field, disregarding the alignment
476 of the variant). The alignment of the variant will depend on the
477 selector: if the variant's 32-bit field is selected, its alignment will
478 be 32-bit, or 64-bit otherwise. It is important to note that variants
479 are specifically tailored for compactness in a stream. Therefore, the
480 relative offsets of compound type fields can vary depending on
481 the offset at which the compound type starts if it contains a variant
482 that itself contains a type with alignment larger than the largest field
483 contained within the compound type. This is caused by the fact that the
484 compound type may contain the enumeration that select the variant's
485 choice, and therefore the alignment to be applied to the compound type
486 cannot be determined before encountering the enumeration.
488 Each variant type selector possess a field name, which is a unique
489 identifier within the variant. The identifier is not allowed to use any
490 reserved keyword (see Section C.1.2). Replacing reserved keywords with
491 underscore-prefixed field names is recommended. Fields starting with an
492 underscore should have their leading underscore removed by the CTF trace
496 A named variant declaration followed by its definition within a structure
507 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
509 variant name <tag_field> v;
512 An unnamed variant definition within a structure is expressed by the following
516 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
518 variant <tag_field> {
526 Example of a named variant within a sequence that refers to a single tag field:
535 enum : uint2_t { a, b, c } choice;
537 variant example <choice> v[seqlen];
540 Example of an unnamed variant:
543 enum : uint2_t { a, b, c, d } choice;
544 /* Unrelated fields can be added between the variant and its tag */
557 Example of an unnamed variant within an array:
560 enum : uint2_t { a, b, c } choice;
568 Example of a variant type definition within a structure, where the defined type
569 is then declared within an array of structures. This variant refers to a tag
570 located in an upper static scope. This example clearly shows that a variant
571 type definition referring to the tag "x" uses the closest preceding field from
572 the static scope of the type definition.
575 enum : uint2_t { a, b, c, d } x;
577 typedef variant <x> { /*
578 * "x" refers to the preceding "x" enumeration in the
579 * static scope of the type definition.
587 enum : int { x, y, z } x; /* This enumeration is not used by "v". */
588 example_variant v; /*
589 * "v" uses the "enum : uint2_t { a, b, c, d }"
597 Arrays are fixed-length. Their length is declared in the type
598 declaration within the meta-data. They contain an array of "inner type"
599 elements, which can refer to any type not containing the type of the
600 array being declared (no circular dependency). The length is the number
601 of elements in an array.
603 TSDL meta-data representation of a named array:
605 typedef elem_type name[length];
607 A nameless array can be declared as a field type within a structure, e.g.:
609 uint8_t field_name[10];
611 Arrays are always aligned on their element alignment requirement.
615 Sequences are dynamically-sized arrays. They refer to a "length"
616 unsigned integer field, which must appear in either the same static scope,
617 prior to the sequence field (in field declaration order), in an upper
618 static scope, or in an upper dynamic scope (see Section 7.3.2). This
619 length field represents the number of elements in the sequence. The
620 sequence per se is an array of "inner type" elements.
622 TSDL meta-data representation for a sequence type definition:
625 unsigned int length_field;
626 typedef elem_type typename[length_field];
627 typename seq_field_name;
630 A sequence can also be declared as a field type, e.g.:
633 unsigned int length_field;
634 long seq_field_name[length_field];
637 Multiple sequences can refer to the same length field, and these length
638 fields can be in a different upper dynamic scope:
640 e.g., assuming the stream.event.header defines:
645 event.header := struct {
654 long seq_a[stream.event.header.seq_len];
655 char seq_b[stream.event.header.seq_len];
659 The sequence elements follow the "array" specifications.
663 Strings are an array of bytes of variable size and are terminated by a '\0'
664 "NULL" character. Their encoding is described in the TSDL meta-data. In
665 absence of encoding attribute information, the default encoding is
668 TSDL meta-data representation of a named string type:
671 encoding = UTF8 OR ASCII;
674 A nameless string type can be declared as a field type:
676 string field_name; /* Use default UTF8 encoding */
678 Strings are always aligned on byte size.
680 5. Event Packet Header
682 The event packet header consists of two parts: the "event packet header"
683 is the same for all streams of a trace. The second part, the "event
684 packet context", is described on a per-stream basis. Both are described
685 in the TSDL meta-data. The packets are aligned on architecture-page-sized
688 Event packet header (all fields are optional, specified by TSDL meta-data):
690 - Magic number (CTF magic number: 0xC1FC1FC1) specifies that this is a
691 CTF packet. This magic number is optional, but when present, it should
692 come at the very beginning of the packet.
693 - Trace UUID, used to ensure the event packet match the meta-data used.
694 (note: we cannot use a meta-data checksum in every cases instead of a
695 UUID because meta-data can be appended to while tracing is active)
696 This field is optional.
697 - Stream ID, used as reference to stream description in meta-data.
698 This field is optional if there is only one stream description in the
699 meta-data, but becomes required if there are more than one stream in
700 the TSDL meta-data description.
702 Event packet context (all fields are optional, specified by TSDL meta-data):
704 - Event packet content size (in bits).
705 - Event packet size (in bits, includes padding).
706 - Event packet content checksum. Checksum excludes the event packet
708 - Per-stream event packet sequence count (to deal with UDP packet loss). The
709 number of significant sequence counter bits should also be present, so
710 wrap-arounds are dealt with correctly.
711 - Time-stamp at the beginning and time-stamp at the end of the event packet.
712 Both timestamps are written in the packet header, but sampled respectively
713 while (or before) writing the first event and while (or after) writing the
714 last event in the packet. The inclusive range between these timestamps should
715 include all event timestamps assigned to events contained within the packet.
716 - Events discarded count
717 - Snapshot of a per-stream free-running counter, counting the number of
718 events discarded that were supposed to be written in the stream after
719 the last event in the event packet.
720 * Note: producer-consumer buffer full condition can fill the current
721 event packet with padding so we know exactly where events have been
722 discarded. However, if the buffer full condition chooses not
723 to fill the current event packet with padding, all we know
724 about the timestamp range in which the events have been
725 discarded is that it is somewhere between the beginning and
726 the end of the packet.
727 - Lossless compression scheme used for the event packet content. Applied
728 directly to raw data. New types of compression can be added in following
729 versions of the format.
730 0: no compression scheme
734 - Cypher used for the event packet content. Applied after compression.
737 - Checksum scheme used for the event packet content. Applied after encryption.
743 5.1 Event Packet Header Description
745 The event packet header layout is indicated by the trace packet.header
746 field. Here is a recommended structure type for the packet header with
747 the fields typically expected (although these fields are each optional):
749 struct event_packet_header {
757 packet.header := struct event_packet_header;
760 If the magic number is not present, tools such as "file" will have no
761 mean to discover the file type.
763 If the uuid is not present, no validation that the meta-data actually
764 corresponds to the stream is performed.
766 If the stream_id packet header field is missing, the trace can only
767 contain a single stream. Its "id" field can be left out, and its events
768 don't need to declare a "stream_id" field.
771 5.2 Event Packet Context Description
773 Event packet context example. These are declared within the stream declaration
774 in the meta-data. All these fields are optional. If the packet size field is
775 missing, the whole stream only contains a single packet. If the content
776 size field is missing, the packet is filled (no padding). The content
777 and packet sizes include all headers.
779 An example event packet context type:
781 struct event_packet_context {
782 uint64_t timestamp_begin;
783 uint64_t timestamp_end;
785 uint32_t stream_packet_count;
786 uint32_t events_discarded;
788 uint64_t/uint32_t/uint16_t content_size;
789 uint64_t/uint32_t/uint16_t packet_size;
790 uint8_t compression_scheme;
791 uint8_t encryption_scheme;
792 uint8_t checksum_scheme;
798 The overall structure of an event is:
800 1 - Stream Packet Context (as specified by the stream meta-data)
801 2 - Event Header (as specified by the stream meta-data)
802 3 - Stream Event Context (as specified by the stream meta-data)
803 4 - Event Context (as specified by the event meta-data)
804 5 - Event Payload (as specified by the event meta-data)
806 This structure defines an implicit dynamic scoping, where variants
807 located in inner structures (those with a higher number in the listing
808 above) can refer to the fields of outer structures (with lower number in
809 the listing above). See Section 7.3 TSDL Scopes for more detail.
813 Event headers can be described within the meta-data. We hereby propose, as an
814 example, two types of events headers. Type 1 accommodates streams with less than
815 31 event IDs. Type 2 accommodates streams with 31 or more event IDs.
817 One major factor can vary between streams: the number of event IDs assigned to
818 a stream. Luckily, this information tends to stay relatively constant (modulo
819 event registration while trace is being recorded), so we can specify different
820 representations for streams containing few event IDs and streams containing
821 many event IDs, so we end up representing the event ID and time-stamp as
822 densely as possible in each case.
824 The header is extended in the rare occasions where the information cannot be
825 represented in the ranges available in the standard event header. They are also
826 used in the rare occasions where the data required for a field could not be
827 collected: the flag corresponding to the missing field within the missing_fields
828 array is then set to 1.
830 Types uintX_t represent an X-bit unsigned integer, as declared with
833 typealias integer { size = X; align = X; signed = false } := uintX_t;
837 typealias integer { size = X; align = 1; signed = false } := uintX_t;
839 6.1.1 Type 1 - Few event IDs
841 - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
843 - Native architecture byte ordering.
844 - For "compact" selection
845 - Fixed size: 32 bits.
846 - For "extended" selection
847 - Size depends on the architecture and variant alignment.
849 struct event_header_1 {
852 * id 31 is reserved to indicate an extended header.
854 enum : uint5_t { compact = 0 ... 30, extended = 31 } id;
860 uint32_t id; /* 32-bit event IDs */
861 uint64_t timestamp; /* 64-bit timestamps */
864 } align(32); /* or align(8) */
867 6.1.2 Type 2 - Many event IDs
869 - Aligned on 16-bit (or 8-bit if byte-packed, depending on the architecture
871 - Native architecture byte ordering.
872 - For "compact" selection
873 - Size depends on the architecture and variant alignment.
874 - For "extended" selection
875 - Size depends on the architecture and variant alignment.
877 struct event_header_2 {
879 * id: range: 0 - 65534.
880 * id 65535 is reserved to indicate an extended header.
882 enum : uint16_t { compact = 0 ... 65534, extended = 65535 } id;
888 uint32_t id; /* 32-bit event IDs */
889 uint64_t timestamp; /* 64-bit timestamps */
892 } align(16); /* or align(8) */
897 The event context contains information relative to the current event.
898 The choice and meaning of this information is specified by the TSDL
899 stream and event meta-data descriptions. The stream context is applied
900 to all events within the stream. The stream context structure follows
901 the event header. The event context is applied to specific events. Its
902 structure follows the stream context structure.
904 An example of stream-level event context is to save the event payload size with
905 each event, or to save the current PID with each event. These are declared
906 within the stream declaration within the meta-data:
910 event.context := struct {
912 uint16_t payload_size;
916 An example of event-specific event context is to declare a bitmap of missing
917 fields, only appended after the stream event context if the extended event
918 header is selected. NR_FIELDS is the number of fields within the event (a
926 uint1_t missing_fields[NR_FIELDS]; /* missing event fields bitmap */
935 An event payload contains fields specific to a given event type. The fields
936 belonging to an event type are described in the event-specific meta-data
937 within a structure type.
941 No padding at the end of the event payload. This differs from the ISO/C standard
942 for structures, but follows the CTF standard for structures. In a trace, even
943 though it makes sense to align the beginning of a structure, it really makes no
944 sense to add padding at the end of the structure, because structures are usually
945 not followed by a structure of the same type.
947 This trick can be done by adding a zero-length "end" field at the end of the C
948 structures, and by using the offset of this field rather than using sizeof()
949 when calculating the size of a structure (see Appendix "A. Helper macros").
953 The event payload is aligned on the largest alignment required by types
954 contained within the payload. (This follows the ISO/C standard for structures)
957 7. Trace Stream Description Language (TSDL)
959 The Trace Stream Description Language (TSDL) allows expression of the
960 binary trace streams layout in a C99-like Domain Specific Language
966 The trace stream layout description is located in the trace meta-data.
967 The meta-data is itself located in a stream identified by its name:
970 The meta-data description can be expressed in two different formats:
971 text-only and packet-based. The text-only description facilitates
972 generation of meta-data and provides a convenient way to enter the
973 meta-data information by hand. The packet-based meta-data provides the
974 CTF stream packet facilities (checksumming, compression, encryption,
975 network-readiness) for meta-data stream generated and transported by a
978 The text-only meta-data file is a plain-text TSDL description. This file
979 must begin with the following characters to identify the file as a CTF
980 TSDL text-based metadata file (without the double-quotes) :
984 It must be followed by a space, and the version of the specification
985 followed by the CTF trace, e.g.:
989 These characters allow automated discovery of file type and CTF
990 specification version. They are interpreted as a the beginning of a
991 comment by the TSDL metadata parser. The comment can be continued to
992 contain extra commented characters before it is closed.
994 The packet-based meta-data is made of "meta-data packets", which each
995 start with a meta-data packet header. The packet-based meta-data
996 description is detected by reading the magic number "0x75D11D57" at the
997 beginning of the file. This magic number is also used to detect the
998 endianness of the architecture by trying to read the CTF magic number
999 and its counterpart in reversed endianness. The events within the
1000 meta-data stream have no event header nor event context. Each event only
1001 contains a special "sequence" payload, which is a sequence of bits which
1002 length is implicitly calculated by using the
1003 "trace.packet.header.content_size" field, minus the packet header size.
1004 The formatting of this sequence of bits is a plain-text representation
1005 of the TSDL description. Each meta-data packet start with a special
1006 packet header, specific to the meta-data stream, which contains,
1009 struct metadata_packet_header {
1010 uint32_t magic; /* 0x75D11D57 */
1011 uint8_t uuid[16]; /* Unique Universal Identifier */
1012 uint32_t checksum; /* 0 if unused */
1013 uint32_t content_size; /* in bits */
1014 uint32_t packet_size; /* in bits */
1015 uint8_t compression_scheme; /* 0 if unused */
1016 uint8_t encryption_scheme; /* 0 if unused */
1017 uint8_t checksum_scheme; /* 0 if unused */
1018 uint8_t major; /* CTF spec version major number */
1019 uint8_t minor; /* CTF spec version minor number */
1022 The packet-based meta-data can be converted to a text-only meta-data by
1023 concatenating all the strings it contains.
1025 In the textual representation of the meta-data, the text contained
1026 within "/*" and "*/", as well as within "//" and end of line, are
1027 treated as comments. Boolean values can be represented as true, TRUE,
1028 or 1 for true, and false, FALSE, or 0 for false. Within the string-based
1029 meta-data description, the trace UUID is represented as a string of
1030 hexadecimal digits and dashes "-". In the event packet header, the trace
1031 UUID is represented as an array of bytes.
1034 7.2 Declaration vs Definition
1036 A declaration associates a layout to a type, without specifying where
1037 this type is located in the event structure hierarchy (see Section 6).
1038 This therefore includes typedef, typealias, as well as all type
1039 specifiers. In certain circumstances (typedef, structure field and
1040 variant field), a declaration is followed by a declarator, which specify
1041 the newly defined type name (for typedef), or the field name (for
1042 declarations located within structure and variants). Array and sequence,
1043 declared with square brackets ("[" "]"), are part of the declarator,
1044 similarly to C99. The enumeration base type is specified by
1045 ": enum_base", which is part of the type specifier. The variant tag
1046 name, specified between "<" ">", is also part of the type specifier.
1048 A definition associates a type to a location in the event structure
1049 hierarchy (see Section 6). This association is denoted by ":=", as shown
1055 TSDL uses three different types of scoping: a lexical scope is used for
1056 declarations and type definitions, and static and dynamic scopes are
1057 used for variants references to tag fields (with relative and absolute
1058 path lookups) and for sequence references to length fields.
1062 Each of "trace", "env", "stream", "event", "struct" and "variant" have
1063 their own nestable declaration scope, within which types can be declared
1064 using "typedef" and "typealias". A root declaration scope also contains
1065 all declarations located outside of any of the aforementioned
1066 declarations. An inner declaration scope can refer to type declared
1067 within its container lexical scope prior to the inner declaration scope.
1068 Redefinition of a typedef or typealias is not valid, although hiding an
1069 upper scope typedef or typealias is allowed within a sub-scope.
1071 7.3.2 Static and Dynamic Scopes
1073 A local static scope consists in the scope generated by the declaration
1074 of fields within a compound type. A static scope is a local static scope
1075 augmented with the nested sub-static-scopes it contains.
1077 A dynamic scope consists in the static scope augmented with the
1078 implicit event structure definition hierarchy presented at Section 6.
1080 Multiple declarations of the same field name within a local static scope
1081 is not valid. It is however valid to re-use the same field name in
1082 different local scopes.
1084 Nested static and dynamic scopes form lookup paths. These are used for
1085 variant tag and sequence length references. They are used at the variant
1086 and sequence definition site to look up the location of the tag field
1087 associated with a variant, and to lookup up the location of the length
1088 field associated with a sequence.
1090 Variants and sequences can refer to a tag field either using a relative
1091 path or an absolute path. The relative path is relative to the scope in
1092 which the variant or sequence performing the lookup is located.
1093 Relative paths are only allowed to lookup within the same static scope,
1094 which includes its nested static scopes. Lookups targeting parent static
1095 scopes need to be performed with an absolute path.
1097 Absolute path lookups use the full path including the dynamic scope
1098 followed by a "." and then the static scope. Therefore, variants (or
1099 sequences) in lower levels in the dynamic scope (e.g. event context) can
1100 refer to a tag (or length) field located in upper levels (e.g. in the
1101 event header) by specifying, in this case, the associated tag with
1102 <stream.event.header.field_name>. This allows, for instance, the event
1103 context to define a variant referring to the "id" field of the event
1106 The dynamic scope prefixes are thus:
1108 - Trace Environment: <env. >,
1109 - Trace Packet Header: <trace.packet.header. >,
1110 - Stream Packet Context: <stream.packet.context. >,
1111 - Event Header: <stream.event.header. >,
1112 - Stream Event Context: <stream.event.context. >,
1113 - Event Context: <event.context. >,
1114 - Event Payload: <event.fields. >.
1117 The target dynamic scope must be specified explicitly when referring to
1118 a field outside of the static scope (absolute scope reference). No
1119 conflict can occur between relative and dynamic paths, because the
1120 keywords "trace", "stream", and "event" are reserved, and thus
1121 not permitted as field names. It is recommended that field names
1122 clashing with CTF and C99 reserved keywords use an underscore prefix to
1123 eliminate the risk of generating a description containing an invalid
1124 field name. Consequently, fields starting with an underscore should have
1125 their leading underscore removed by the CTF trace readers.
1128 The information available in the dynamic scopes can be thought of as the
1129 current tracing context. At trace production, information about the
1130 current context is saved into the specified scope field levels. At trace
1131 consumption, for each event, the current trace context is therefore
1132 readable by accessing the upper dynamic scopes.
1137 The grammar representing the TSDL meta-data is presented in Appendix C.
1138 TSDL Grammar. This section presents a rather lighter reading that
1139 consists in examples of TSDL meta-data, with template values.
1141 The stream "id" can be left out if there is only one stream in the
1142 trace. The event "id" field can be left out if there is only one event
1146 major = value; /* CTF spec version major number */
1147 minor = value; /* CTF spec version minor number */
1148 uuid = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"; /* Trace UUID */
1149 byte_order = be OR le; /* Endianness (required) */
1150 packet.header := struct {
1158 * The "env" (environment) scope contains assignment expressions. The
1159 * field names and content are implementation-defined.
1162 pid = value; /* example */
1163 proc_name = "name"; /* example */
1169 /* Type 1 - Few event IDs; Type 2 - Many event IDs. See section 6.1. */
1170 event.header := event_header_1 OR event_header_2;
1171 event.context := struct {
1174 packet.context := struct {
1180 name = "event_name";
1181 id = value; /* Numeric identifier within the stream */
1182 stream_id = stream_id;
1184 model.emf.uri = "string";
1194 name = "event_name";
1201 /* More detail on types in section 4. Types */
1206 * Type declarations behave similarly to the C standard.
1209 typedef aliased_type_specifiers new_type_declarators;
1211 /* e.g.: typedef struct example new_type_name[10]; */
1216 * The "typealias" declaration can be used to give a name (including
1217 * pointer declarator specifier) to a type. It should also be used to
1218 * map basic C types (float, int, unsigned long, ...) to a CTF type.
1219 * Typealias is a superset of "typedef": it also allows assignment of a
1220 * simple variable identifier to a type.
1223 typealias type_class {
1225 } := type_specifiers type_declarator;
1229 * typealias integer {
1233 * } := struct page *;
1235 * typealias integer {
1250 enum name : integer_type {
1256 * Unnamed types, contained within compound type fields, typedef or typealias.
1271 enum : integer_type {
1275 typedef type new_type[length];
1278 type field_name[length];
1281 typedef type new_type[length_type];
1284 type field_name[length_type];
1296 integer_type field_name:size; /* GNU/C bitfield */
1306 Clock metadata allows to describe the clock topology of the system, as
1307 well as to detail each clock parameter. In absence of clock description,
1308 it is assumed that all fields named "timestamp" use the same clock
1309 source, which increments once per nanosecond.
1311 Describing a clock and how it is used by streams is threefold: first,
1312 the clock and clock topology should be described in a "clock"
1313 description block, e.g.:
1316 name = cycle_counter_sync;
1317 uuid = "62189bee-96dc-11e0-91a8-cfa3d89f3923";
1318 description = "Cycle counter synchronized across CPUs";
1319 freq = 1000000000; /* frequency, in Hz */
1320 /* precision in seconds is: 1000 * (1/freq) */
1323 * clock value offset from Epoch is:
1324 * offset_s + (offset * (1/freq))
1326 offset_s = 1326476837;
1331 The mandatory "name" field specifies the name of the clock identifier,
1332 which can later be used as a reference. The optional field "uuid" is the
1333 unique identifier of the clock. It can be used to correlate different
1334 traces that use the same clock. An optional textual description string
1335 can be added with the "description" field. The "freq" field is the
1336 initial frequency of the clock, in Hz. If the "freq" field is not
1337 present, the frequency is assumed to be 1000000000 (providing clock
1338 increment of 1 ns). The optional "precision" field details the
1339 uncertainty on the clock measurements, in (1/freq) units. The "offset_s"
1340 and "offset" fields indicate the offset from POSIX.1 Epoch, 1970-01-01
1341 00:00:00 +0000 (UTC), to the zero of value of the clock. The "offset_s"
1342 field is in seconds. The "offset" field is in (1/freq) units. If any of
1343 the "offset_s" or "offset" field is not present, it is assigned the 0
1344 value. The field "absolute" is TRUE if the clock is a global reference
1345 across different clock uuid (e.g. NTP time). Otherwise, "absolute" is
1346 FALSE, and the clock can be considered as synchronized only with other
1347 clocks that have the same uuid.
1350 Secondly, a reference to this clock should be added within an integer
1354 size = 64; align = 1; signed = false;
1355 map = clock.cycle_counter_sync.value;
1358 Thirdly, stream declarations can reference the clock they use as a
1361 struct packet_context {
1362 uint64_ccnt_t ccnt_begin;
1363 uint64_ccnt_t ccnt_end;
1369 event.header := struct {
1370 uint64_ccnt_t timestamp;
1373 packet.context := struct packet_context;
1376 For a N-bit integer type referring to a clock, if the integer overflows
1377 compared to the N low order bits of the clock prior value, then it is
1378 assumed that one, and only one, overflow occurred. It is therefore
1379 important that events encoding time on a small number of bits happen
1380 frequently enough to detect when more than one N-bit overflow occurs.
1382 In a packet context, clock field names ending with "_begin" and "_end"
1383 have a special meaning: this refers to the time-stamps at, respectively,
1384 the beginning and the end of each packet.
1389 The two following macros keep track of the size of a GNU/C structure without
1390 padding at the end by placing HEADER_END as the last field. A one byte end field
1391 is used for C90 compatibility (C99 flexible arrays could be used here). Note
1392 that this does not affect the effective structure size, which should always be
1393 calculated with the header_sizeof() helper.
1395 #define HEADER_END char end_field
1396 #define header_sizeof(type) offsetof(typeof(type), end_field)
1399 B. Stream Header Rationale
1401 An event stream is divided in contiguous event packets of variable size. These
1402 subdivisions allow the trace analyzer to perform a fast binary search by time
1403 within the stream (typically requiring to index only the event packet headers)
1404 without reading the whole stream. These subdivisions have a variable size to
1405 eliminate the need to transfer the event packet padding when partially filled
1406 event packets must be sent when streaming a trace for live viewing/analysis.
1407 An event packet can contain a certain amount of padding at the end. Dividing
1408 streams into event packets is also useful for network streaming over UDP and
1409 flight recorder mode tracing (a whole event packet can be swapped out of the
1410 buffer atomically for reading).
1412 The stream header is repeated at the beginning of each event packet to allow
1413 flexibility in terms of:
1415 - streaming support,
1416 - allowing arbitrary buffers to be discarded without making the trace
1418 - allow UDP packet loss handling by either dealing with missing event packet
1419 or asking for re-transmission.
1420 - transparently support flight recorder mode,
1421 - transparently support crash dump.
1427 * Common Trace Format (CTF) Trace Stream Description Language (TSDL) Grammar.
1429 * Inspired from the C99 grammar:
1430 * http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf (Annex A)
1431 * and c++1x grammar (draft)
1432 * http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3291.pdf (Annex A)
1434 * Specialized for CTF needs by including only constant and declarations from
1435 * C99 (excluding function declarations), and by adding support for variants,
1436 * sequences and CTF-specific specifiers. Enumeration container types
1437 * semantic is inspired from c++1x enum-base.
1442 1.1) Lexical elements
1489 identifier identifier-nondigit
1492 identifier-nondigit:
1494 universal-character-name
1495 any other implementation-defined characters
1499 [a-zA-Z] /* regular expression */
1502 [0-9] /* regular expression */
1504 1.4) Universal character names
1506 universal-character-name:
1508 \U hex-quad hex-quad
1511 hexadecimal-digit hexadecimal-digit hexadecimal-digit hexadecimal-digit
1517 enumeration-constant
1521 decimal-constant integer-suffix-opt
1522 octal-constant integer-suffix-opt
1523 hexadecimal-constant integer-suffix-opt
1527 decimal-constant digit
1531 octal-constant octal-digit
1533 hexadecimal-constant:
1534 hexadecimal-prefix hexadecimal-digit
1535 hexadecimal-constant hexadecimal-digit
1545 unsigned-suffix long-suffix-opt
1546 unsigned-suffix long-long-suffix
1547 long-suffix unsigned-suffix-opt
1548 long-long-suffix unsigned-suffix-opt
1562 enumeration-constant:
1568 L' c-char-sequence '
1572 c-char-sequence c-char
1575 any member of source charset except single-quote ('), backslash
1576 (\), or new-line character.
1580 simple-escape-sequence
1581 octal-escape-sequence
1582 hexadecimal-escape-sequence
1583 universal-character-name
1585 simple-escape-sequence: one of
1586 \' \" \? \\ \a \b \f \n \r \t \v
1588 octal-escape-sequence:
1590 \ octal-digit octal-digit
1591 \ octal-digit octal-digit octal-digit
1593 hexadecimal-escape-sequence:
1594 \x hexadecimal-digit
1595 hexadecimal-escape-sequence hexadecimal-digit
1597 1.6) String literals
1600 " s-char-sequence-opt "
1601 L" s-char-sequence-opt "
1605 s-char-sequence s-char
1608 any member of source charset except double-quote ("), backslash
1609 (\), or new-line character.
1615 [ ] ( ) { } . -> * + - < > : ; ... = ,
1618 2) Phrase structure grammar
1624 ( unary-expression )
1628 postfix-expression [ unary-expression ]
1629 postfix-expression . identifier
1630 postfix-expressoin -> identifier
1634 unary-operator postfix-expression
1636 unary-operator: one of
1639 assignment-operator:
1642 type-assignment-operator:
1645 constant-expression-range:
1646 unary-expression ... unary-expression
1651 declaration-specifiers declarator-list-opt ;
1654 declaration-specifiers:
1655 storage-class-specifier declaration-specifiers-opt
1656 type-specifier declaration-specifiers-opt
1657 type-qualifier declaration-specifiers-opt
1661 declarator-list , declarator
1663 abstract-declarator-list:
1665 abstract-declarator-list , abstract-declarator
1667 storage-class-specifier:
1690 align ( unary-expression )
1693 struct identifier-opt { struct-or-variant-declaration-list-opt } align-attribute-opt
1694 struct identifier align-attribute-opt
1696 struct-or-variant-declaration-list:
1697 struct-or-variant-declaration
1698 struct-or-variant-declaration-list struct-or-variant-declaration
1700 struct-or-variant-declaration:
1701 specifier-qualifier-list struct-or-variant-declarator-list ;
1702 declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list ;
1703 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list ;
1704 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list ;
1706 specifier-qualifier-list:
1707 type-specifier specifier-qualifier-list-opt
1708 type-qualifier specifier-qualifier-list-opt
1710 struct-or-variant-declarator-list:
1711 struct-or-variant-declarator
1712 struct-or-variant-declarator-list , struct-or-variant-declarator
1714 struct-or-variant-declarator:
1716 declarator-opt : unary-expression
1719 variant identifier-opt variant-tag-opt { struct-or-variant-declaration-list }
1720 variant identifier variant-tag
1723 < unary-expression >
1726 enum identifier-opt { enumerator-list }
1727 enum identifier-opt { enumerator-list , }
1729 enum identifier-opt : declaration-specifiers { enumerator-list }
1730 enum identifier-opt : declaration-specifiers { enumerator-list , }
1734 enumerator-list , enumerator
1737 enumeration-constant
1738 enumeration-constant assignment-operator unary-expression
1739 enumeration-constant assignment-operator constant-expression-range
1745 pointer-opt direct-declarator
1750 direct-declarator [ unary-expression ]
1752 abstract-declarator:
1753 pointer-opt direct-abstract-declarator
1755 direct-abstract-declarator:
1757 ( abstract-declarator )
1758 direct-abstract-declarator [ unary-expression ]
1759 direct-abstract-declarator [ ]
1762 * type-qualifier-list-opt
1763 * type-qualifier-list-opt pointer
1765 type-qualifier-list:
1767 type-qualifier-list type-qualifier
1772 2.3) CTF-specific declarations
1775 clock { ctf-assignment-expression-list-opt }
1776 event { ctf-assignment-expression-list-opt }
1777 stream { ctf-assignment-expression-list-opt }
1778 env { ctf-assignment-expression-list-opt }
1779 trace { ctf-assignment-expression-list-opt }
1780 callsite { ctf-assignment-expression-list-opt }
1781 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1782 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list
1785 floating_point { ctf-assignment-expression-list-opt }
1786 integer { ctf-assignment-expression-list-opt }
1787 string { ctf-assignment-expression-list-opt }
1790 ctf-assignment-expression-list:
1791 ctf-assignment-expression ;
1792 ctf-assignment-expression-list ctf-assignment-expression ;
1794 ctf-assignment-expression:
1795 unary-expression assignment-operator unary-expression
1796 unary-expression type-assignment-operator type-specifier
1797 declaration-specifiers-opt storage-class-specifier declaration-specifiers-opt declarator-list
1798 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declaration-specifiers abstract-declarator-list
1799 typealias declaration-specifiers abstract-declarator-list type-assignment-operator declarator-list