2 RFC: Common Trace Format (CTF) Proposal (pre-v1.7)
4 Mathieu Desnoyers, EfficiOS Inc.
6 The goal of the present document is to propose a trace format that suits the
7 needs of the embedded, telecom, high-performance and kernel communities. It is
8 based on the Common Trace Format Requirements (v1.4) document. It is designed to
9 allow traces to be natively generated by the Linux kernel, Linux user-space
10 applications written in C/C++, and hardware components.
12 The latest version of this document can be found at:
14 git tree: git://git.efficios.com/ctf.git
15 gitweb: http://git.efficios.com/?p=ctf.git
17 A reference implementation of a library to read and write this trace format is
18 being implemented within the BabelTrace project, a converter between trace
19 formats. The development tree is available at:
21 git tree: git://git.efficios.com/babeltrace.git
22 gitweb: http://git.efficios.com/?p=babeltrace.git
25 1. Preliminary definitions
27 - Event Trace: An ordered sequence of events.
28 - Event Stream: An ordered sequence of events, containing a subset of the
30 - Event Packet: A sequence of physically contiguous events within an event
32 - Event: This is the basic entry in a trace. (aka: a trace record).
33 - An event identifier (ID) relates to the class (a type) of event within
35 e.g. event: irq_entry.
36 - An event (or event record) relates to a specific instance of an event
38 e.g. event: irq_entry, at time X, on CPU Y
39 - Source Architecture: Architecture writing the trace.
40 - Reader Architecture: Architecture reading the trace.
43 2. High-level representation of a trace
45 A trace is divided into multiple event streams. Each event stream contains a
46 subset of the trace event types.
48 The final output of the trace, after its generation and optional transport over
49 the network, is expected to be either on permanent or temporary storage in a
50 virtual file system. Because each event stream is appended to while a trace is
51 being recorded, each is associated with a separate file for output. Therefore,
52 a stored trace can be represented as a directory containing one file per stream.
54 A metadata event stream contains information on trace event types. It describes:
58 - Per-stream event header description.
59 - Per-stream event header selection.
60 - Per-stream event context fields.
62 - Event type to stream mapping.
63 - Event type to name mapping.
64 - Event type to ID mapping.
65 - Event fields description.
70 An event stream is divided in contiguous event packets of variable size. These
71 subdivisions have a variable size. An event packet can contain a certain
72 amount of padding at the end. The stream header is repeated at the
73 beginning of each event packet. The rationale for the event stream
74 design choices is explained in Appendix B. Stream Header Rationale.
76 The event stream header will therefore be referred to as the "event packet
77 header" throughout the rest of this document.
82 Types are organized as type classes. Each type class belong to either of two
83 kind of types: basic types or compound types.
87 A basic type is a scalar type, as described in this section. It includes
88 integers, GNU/C bitfields, enumerations, and floating point values.
90 4.1.1 Type inheritance
92 Type specifications can be inherited to allow deriving types from a
93 type class. For example, see the uint32_t named type derived from the "integer"
94 type class below ("Integers" section). Types have a precise binary
95 representation in the trace. A type class has methods to read and write these
96 types, but must be derived into a type to be usable in an event field.
100 We define "byte-packed" types as aligned on the byte size, namely 8-bit.
101 We define "bit-packed" types as following on the next bit, as defined by the
104 All basic types, except bitfields, are either aligned on an architecture-defined
105 specific alignment or byte-packed, depending on the architecture preference.
106 Architectures providing fast unaligned write byte-packed basic types to save
107 space, aligning each type on byte boundaries (8-bit). Architectures with slow
108 unaligned writes align types on specific alignment values. If no specific
109 alignment is declared for a type, it is assumed to be bit-packed for
110 integers with size not multiple of 8 bits and for gcc bitfields. All
111 other types are byte-packed.
113 Metadata attribute representation of a specific alignment:
115 align = value; /* value in bits */
119 By default, the native endianness of the source architecture the trace is used.
120 Byte order can be overridden for a basic type by specifying a "byte_order"
121 attribute. Typical use-case is to specify the network byte order (big endian:
122 "be") to save data captured from the network into the trace without conversion.
123 If not specified, the byte order is native.
125 Metadata representation:
127 byte_order = native OR network OR be OR le; /* network and be are aliases */
131 Type size, in bits, for integers and floats is that returned by "sizeof()" in C
132 multiplied by CHAR_BIT.
133 We require the size of "char" and "unsigned char" types (CHAR_BIT) to be fixed
134 to 8 bits for cross-endianness compatibility.
136 Metadata representation:
138 size = value; (value is in bits)
142 Signed integers are represented in two-complement. Integer alignment, size,
143 signedness and byte ordering are defined in the metadata. Integers aligned on
144 byte size (8-bit) and with length multiple of byte size (8-bit) correspond to
145 the C99 standard integers. In addition, integers with alignment and/or size that
146 are _not_ a multiple of the byte size are permitted; these correspond to the C99
147 standard bitfields, with the added specification that the CTF integer bitfields
148 have a fixed binary representation. A MIT-licensed reference implementation of
149 the CTF portable bitfields is available at:
151 http://git.efficios.com/?p=babeltrace.git;a=blob;f=include/babeltrace/bitfield.h
153 Binary representation of integers:
155 - On little and big endian:
156 - Within a byte, high bits correspond to an integer high bits, and low bits
157 correspond to low bits.
159 - Integer across multiple bytes are placed from the less significant to the
161 - Consecutive integers are placed from lower bits to higher bits (even within
164 - Integer across multiple bytes are placed from the most significant to the
166 - Consecutive integers are placed from higher bits to lower bits (even within
169 This binary representation is derived from the bitfield implementation in GCC
170 for little and big endian. However, contrary to what GCC does, integers can
171 cross units boundaries (no padding is required). Padding can be explicitely
172 added (see 4.1.6 GNU/C bitfields) to follow the GCC layout if needed.
174 Metadata representation:
177 signed = true OR false; /* default false */
178 byte_order = native OR network OR be OR le; /* default native */
179 size = value; /* value in bits, no default */
180 align = value; /* value in bits */
183 Example of type inheritance (creation of a uint32_t named type):
191 Definition of a named 5-bit signed bitfield:
199 4.1.6 GNU/C bitfields
201 The GNU/C bitfields follow closely the integer representation, with a
202 particularity on alignment: if a bitfield cannot fit in the current unit, the
203 unit is padded and the bitfield starts at the following unit. The unit size is
204 defined by the size of the type "unit_type".
206 Metadata representation:
210 As an example, the following structure declared in C compiled by GCC:
217 The example structure is aligned on the largest element (short). The second
218 bitfield would be aligned on the next unit boundary, because it would not fit in
223 The floating point values byte ordering is defined in the metadata.
225 Floating point values follow the IEEE 754-2008 standard interchange formats.
226 Description of the floating point values include the exponent and mantissa size
227 in bits. Some requirements are imposed on the floating point values:
229 - FLT_RADIX must be 2.
230 - mant_dig is the number of digits represented in the mantissa. It is specified
231 by the ISO C99 standard, section 5.2.4, as FLT_MANT_DIG, DBL_MANT_DIG and
232 LDBL_MANT_DIG as defined by <float.h>.
233 - exp_dig is the number of digits represented in the exponent. Given that
234 mant_dig is one bit more than its actual size in bits (leading 1 is not
235 needed) and also given that the sign bit always takes one bit, exp_dig can be
238 - sizeof(float) * CHAR_BIT - FLT_MANT_DIG
239 - sizeof(double) * CHAR_BIT - DBL_MANT_DIG
240 - sizeof(long double) * CHAR_BIT - LDBL_MANT_DIG
242 Metadata representation:
247 byte_order = native OR network OR be OR le;
250 Example of type inheritance:
252 typealias floating_point {
253 exp_dig = 8; /* sizeof(float) * CHAR_BIT - FLT_MANT_DIG */
254 mant_dig = 24; /* FLT_MANT_DIG */
258 TODO: define NaN, +inf, -inf behavior.
262 Enumerations are a mapping between an integer type and a table of strings. The
263 numerical representation of the enumeration follows the integer type specified
264 by the metadata. The enumeration mapping table is detailed in the enumeration
265 description within the metadata. The mapping table maps inclusive value ranges
266 (or single values) to strings. Instead of being limited to simple
267 "value -> string" mappings, these enumerations map
268 "[ start_value ... end_value ] -> string", which map inclusive ranges of
269 values to strings. An enumeration from the C language can be represented in
270 this format by having the same start_value and end_value for each element, which
271 is in fact a range of size 1. This single-value range is supported without
272 repeating the start and end values with the value = string declaration.
274 enum name : integer_type {
275 somestring = start_value1 ... end_value1,
276 "other string" = start_value2 ... end_value2,
277 yet_another_string, /* will be assigned to end_value2 + 1 */
278 "some other string" = value,
282 If the values are omitted, the enumeration starts at 0 and increment of 1 for
285 enum name : unsigned int {
293 Overlapping ranges within a single enumeration are implementation defined.
295 A nameless enumeration can be declared as a field type or as part of a typedef:
297 enum : integer_type {
301 Enumerations omitting the container type ": integer_type" use the "int"
302 type (for compatibility with C99). The "int" type must be previously
305 typealias integer { size = 32; align = 32; signed = true } := int;
314 Compound are aggregation of type declarations. Compound types include
315 structures, variant, arrays, sequences, and strings.
319 Structures are aligned on the largest alignment required by basic types
320 contained within the structure. (This follows the ISO/C standard for structures)
322 Metadata representation of a named structure:
325 field_type field_name;
326 field_type field_name;
333 integer { /* Nameless type */
338 uint64_t second_field_name; /* Named type declared in the metadata */
341 The fields are placed in a sequence next to each other. They each possess a
342 field name, which is a unique identifier within the structure.
344 A nameless structure can be declared as a field type or as part of a typedef:
350 4.2.2 Variants (Discriminated/Tagged Unions)
352 A CTF variant is a selection between different types. A CTF variant must
353 always be defined within the scope of a structure or within fields
354 contained within a structure (defined recursively). A "tag" enumeration
355 field must appear in either the same lexical scope, prior to the variant
356 field (in field declaration order), in an uppermost lexical scope (see
357 Section 7.2.1), or in an uppermost dynamic scope (see Section 7.2.2).
358 The type selection is indicated by the mapping from the enumeration
359 value to the string used as variant type selector. The field to use as
360 tag is specified by the "tag_field", specified between "< >" after the
361 "variant" keyword for unnamed variants, and after "variant name" for
364 The alignment of the variant is the alignment of the type as selected by the tag
365 value for the specific instance of the variant. The alignment of the type
366 containing the variant is independent of the variant alignment. The size of the
367 variant is the size as selected by the tag value for the specific instance of
370 A named variant declaration followed by its definition within a structure
381 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
383 variant name <tag_field> v;
386 An unnamed variant definition within a structure is expressed by the following
390 enum : integer_type { sel1, sel2, sel3, ... } tag_field;
392 variant <tag_field> {
400 Example of a named variant within a sequence that refers to a single tag field:
409 enum : uint2_t { a, b, c } choice;
410 variant example <choice> v[unsigned int];
413 Example of an unnamed variant:
416 enum : uint2_t { a, b, c, d } choice;
417 /* Unrelated fields can be added between the variant and its tag */
430 Example of an unnamed variant within an array:
433 enum : uint2_t { a, b, c } choice;
441 Example of a variant type definition within a structure, where the defined type
442 is then declared within an array of structures. This variant refers to a tag
443 located in an upper lexical scope. This example clearly shows that a variant
444 type definition referring to the tag "x" uses the closest preceding field from
445 the lexical scope of the type definition.
448 enum : uint2_t { a, b, c, d } x;
450 typedef variant <x> { /*
451 * "x" refers to the preceding "x" enumeration in the
452 * lexical scope of the type definition.
460 enum : int { x, y, z } x; /* This enumeration is not used by "v". */
461 example_variant v; /*
462 * "v" uses the "enum : uint2_t { a, b, c, d }"
470 Arrays are fixed-length. Their length is declared in the type declaration within
471 the metadata. They contain an array of "inner type" elements, which can refer to
472 any type not containing the type of the array being declared (no circular
473 dependency). The length is the number of elements in an array.
475 Metadata representation of a named array:
477 typedef elem_type name[length];
479 A nameless array can be declared as a field type within a structure, e.g.:
481 uint8_t field_name[10];
486 Sequences are dynamically-sized arrays. They start with an integer that specify
487 the length of the sequence, followed by an array of "inner type" elements.
488 The length is the number of elements in the sequence.
490 Metadata representation for a named sequence:
492 typedef elem_type name[length_type];
494 A nameless sequence can be declared as a field type, e.g.:
496 long field_name[int];
498 The length type follows the integer types specifications, and the sequence
499 elements follow the "array" specifications.
503 Strings are an array of bytes of variable size and are terminated by a '\0'
504 "NULL" character. Their encoding is described in the metadata. In absence of
505 encoding attribute information, the default encoding is UTF-8.
507 Metadata representation of a named string type:
510 encoding = UTF8 OR ASCII;
513 A nameless string type can be declared as a field type:
515 string field_name; /* Use default UTF8 encoding */
517 5. Event Packet Header
519 The event packet header consists of two part: one is mandatory and have a fixed
520 layout. The second part, the "event packet context", has its layout described in
523 - Aligned on page size. Fixed size. Fields either aligned or packed (depending
524 on the architecture preference).
525 No padding at the end of the event packet header. Native architecture byte
528 Fixed layout (event packet header):
530 - Magic number (CTF magic numbers: 0xC1FC1FC1 and its reverse endianness
531 representation: 0xC11FFCC1) It needs to have a non-symmetric bytewise
532 representation. Used to distinguish between big and little endian traces (this
533 information is determined by knowing the endianness of the architecture
534 reading the trace and comparing the magic number against its value and the
535 reverse, 0xC11FFCC1). This magic number specifies that we use the CTF metadata
536 description language described in this document. Different magic numbers
537 should be used for other metadata description languages.
538 - Trace UUID, used to ensure the event packet match the metadata used.
539 (note: we cannot use a metadata checksum because metadata can be appended to
540 while tracing is active)
541 - Stream ID, used as reference to stream description in metadata.
543 Metadata-defined layout (event packet context):
545 - Event packet content size (in bytes).
546 - Event packet size (in bytes, includes padding).
547 - Event packet content checksum (optional). Checksum excludes the event packet
549 - Per-stream event packet sequence count (to deal with UDP packet loss). The
550 number of significant sequence counter bits should also be present, so
551 wrap-arounds are deal with correctly.
552 - Timestamp at the beginning and timestamp at the end of the event packet.
553 Both timestamps are written in the packet header, but sampled respectively
554 while (or before) writing the first event and while (or after) writing the
555 last event in the packet. The inclusive range between these timestamps should
556 include all event timestamps assigned to events contained within the packet.
557 - Events discarded count
558 - Snapshot of a per-stream free-running counter, counting the number of
559 events discarded that were supposed to be written in the stream prior to
560 the first event in the event packet.
561 * Note: producer-consumer buffer full condition should fill the current
562 event packet with padding so we know exactly where events have been
564 - Lossless compression scheme used for the event packet content. Applied
565 directly to raw data. New types of compression can be added in following
566 versions of the format.
567 0: no compression scheme
571 - Cypher used for the event packet content. Applied after compression.
574 - Checksum scheme used for the event packet content. Applied after encryption.
580 5.1 Event Packet Header Fixed Layout Description
582 struct event_packet_header {
584 uint8_t trace_uuid[16];
588 5.2 Event Packet Context Description
590 Event packet context example. These are declared within the stream declaration
591 in the metadata. All these fields are optional except for "content_size" and
592 "packet_size", which must be present in the context.
594 An example event packet context type:
596 struct event_packet_context {
597 uint64_t timestamp_begin;
598 uint64_t timestamp_end;
600 uint32_t stream_packet_count;
601 uint32_t events_discarded;
603 uint32_t/uint16_t content_size;
604 uint32_t/uint16_t packet_size;
605 uint8_t stream_packet_count_bits; /* Significant counter bits */
606 uint8_t compression_scheme;
607 uint8_t encryption_scheme;
608 uint8_t checksum_scheme;
614 The overall structure of an event is:
616 1 - Stream Packet Context (as specified by the stream metadata)
617 2 - Event Header (as specified by the stream metadata)
618 3 - Stream Event Context (as specified by the stream metadata)
619 4 - Event Context (as specified by the event metadata)
620 5 - Event Payload (as specified by the event metadata)
622 This structure defines an implicit dynamic scoping, where variants
623 located in inner structures (those with a higher number in the listing
624 above) can refer to the fields of outer structures (with lower number in
625 the listing above). See Section 7.2 Metadata Scopes for more detail.
629 Event headers can be described within the metadata. We hereby propose, as an
630 example, two types of events headers. Type 1 accommodates streams with less than
631 31 event IDs. Type 2 accommodates streams with 31 or more event IDs.
633 One major factor can vary between streams: the number of event IDs assigned to
634 a stream. Luckily, this information tends to stay relatively constant (modulo
635 event registration while trace is being recorded), so we can specify different
636 representations for streams containing few event IDs and streams containing
637 many event IDs, so we end up representing the event ID and timestamp as densely
638 as possible in each case.
640 The header is extended in the rare occasions where the information cannot be
641 represented in the ranges available in the standard event header. They are also
642 used in the rare occasions where the data required for a field could not be
643 collected: the flag corresponding to the missing field within the missing_fields
644 array is then set to 1.
646 Types uintX_t represent an X-bit unsigned integer.
649 6.1.1 Type 1 - Few event IDs
651 - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
653 - Native architecture byte ordering.
654 - For "compact" selection
655 - Fixed size: 32 bits.
656 - For "extended" selection
657 - Size depends on the architecture and variant alignment.
659 struct event_header_1 {
662 * id 31 is reserved to indicate an extended header.
664 enum : uint5_t { compact = 0 ... 30, extended = 31 } id;
670 uint32_t id; /* 32-bit event IDs */
671 uint64_t timestamp; /* 64-bit timestamps */
677 6.1.2 Type 2 - Many event IDs
679 - Aligned on 16-bit (or 8-bit if byte-packed, depending on the architecture
681 - Native architecture byte ordering.
682 - For "compact" selection
683 - Size depends on the architecture and variant alignment.
684 - For "extended" selection
685 - Size depends on the architecture and variant alignment.
687 struct event_header_2 {
689 * id: range: 0 - 65534.
690 * id 65535 is reserved to indicate an extended header.
692 enum : uint16_t { compact = 0 ... 65534, extended = 65535 } id;
698 uint32_t id; /* 32-bit event IDs */
699 uint64_t timestamp; /* 64-bit timestamps */
707 The event context contains information relative to the current event. The choice
708 and meaning of this information is specified by the metadata "stream" and
709 "event" information. The "stream" context is applied to all events within the
710 stream. The "stream" context structure follows the event header. The "event"
711 context is applied to specific events. Its structure follows the "stream"
714 An example of stream-level event context is to save the event payload size with
715 each event, or to save the current PID with each event. These are declared
716 within the stream declaration within the metadata:
724 uint16_t payload_size;
729 An example of event-specific event context is to declare a bitmap of missing
730 fields, only appended after the stream event context if the extended event
731 header is selected. NR_FIELDS is the number of fields within the event (a
739 uint1_t missing_fields[NR_FIELDS]; /* missing event fields bitmap */
748 An event payload contains fields specific to a given event type. The fields
749 belonging to an event type are described in the event-specific metadata
750 within a structure type.
754 No padding at the end of the event payload. This differs from the ISO/C standard
755 for structures, but follows the CTF standard for structures. In a trace, even
756 though it makes sense to align the beginning of a structure, it really makes no
757 sense to add padding at the end of the structure, because structures are usually
758 not followed by a structure of the same type.
760 This trick can be done by adding a zero-length "end" field at the end of the C
761 structures, and by using the offset of this field rather than using sizeof()
762 when calculating the size of a structure (see Appendix "A. Helper macros").
766 The event payload is aligned on the largest alignment required by types
767 contained within the payload. (This follows the ISO/C standard for structures)
772 The meta-data is located in a stream identified by its name: "metadata".
773 It is made of "event packets", which each start with an event packet
774 header. The event type within the metadata stream have no event header
775 nor event context. Each event only contains a null-terminated "string"
776 payload, which is a metadata description entry. The events are packed
777 one next to another. Each event packet start with an event packet
778 header, which contains, amongst other fields, the magic number and trace
779 UUID. In the event packet header, the trace UUID is represented as an
780 array of bytes. Within the string-based metadata description, the trace
781 UUID is represented as a string of hexadecimal digits and dashes "-".
783 The metadata can be parsed by reading through the metadata strings,
784 skipping null-characters. Type names are made of a single identifier,
785 and can be surrounded by prefix/postfix. Text contained within "/*" and
786 "*/", as well as within "//" and end of line, are treated as comments.
787 Boolean values can be represented as true, TRUE, or 1 for true, and
788 false, FALSE, or 0 for false.
791 7.1 Declaration vs Definition
793 A declaration associates a layout to a type, without specifying where
794 this type is located in the event structure hierarchy (see Section 6).
795 This therefore includes typedef, typealias, as well as all type
796 specifiers. In certain circumstances (typedef, structure field and
797 variant field), a declaration is followed by a declarator, which specify
798 the newly defined type name (for typedef), or the field name (for
799 declarations located within structure and variants). Array and sequence,
800 declared with square brackets ("[" "]"), are part of the declarator,
801 similarly to C99. The enumeration base type is specified by
802 ": base_type", which is part of the type specifier. The variant tag
803 name, specified between "<" ">", is also part of the type specifier.
805 A definition associates a type to a location in the event structure
806 hierarchy (see Section 6). This association is denoted by ":=", as shown
812 CTF metadata uses two different types of scoping: a lexical scope is
813 used for declarations and type definitions, and a dynamic scope is used
814 for variants references to tag fields.
818 Each of "trace", "stream", "event", "struct" and "variant" have their own
819 nestable declaration scope, within which types can be declared using "typedef"
820 and "typealias". A root declaration scope also contains all declarations
821 located outside of any of the aforementioned declarations. An inner
822 declaration scope can refer to type declared within its container
823 lexical scope prior to the inner declaration scope. Redefinition of a
824 typedef or typealias is not valid, although hiding an upper scope
825 typedef or typealias is allowed within a sub-scope.
829 A dynamic scope consists in the lexical scope augmented with the
830 implicit event structure definition hierarchy presented at Section 6.
831 The dynamic scope is only used for variant tag definitions. It is used
832 at definition time to look up the location of the tag field associated
835 Therefore, variants in lower levels in the dynamic scope (e.g. event
836 context) can refer to a tag field located in upper levels (e.g. in the
837 event header) by specifying, in this case, the associated tag with
838 <header.field_name>. This allows, for instance, the event context to
839 define a variant referring to the "id" field of the event header as
842 The target dynamic scope must be specified explicitly when referring to
843 a field outside of the local static scope. The dynamic scope prefixes
846 - Stream Packet Context: <stream.packet.context. >,
847 - Event Header: <stream.event.header. >,
848 - Stream Event Context: <stream.event.context. >,
849 - Event Context: <event.context. >,
850 - Event Payload: <event.fields. >.
852 Multiple declarations of the same field name within a single scope is
853 not valid. It is however valid to re-use the same field name in
854 different scopes. There is no possible conflict, because the dynamic
855 scope must be specified when a variant refers to a tag field located in
856 a different dynamic scope.
858 The information available in the dynamic scopes can be thought of as the
859 current tracing context. At trace production, information about the
860 current context is saved into the specified scope field levels. At trace
861 consumption, for each event, the current trace context is therefore
862 readable by accessing the upper dynamic scopes.
865 7.3 Metadata Examples
867 The grammar representing the CTF metadata is presented in
868 Appendix C. CTF Metadata Grammar. This section presents a rather ligher
869 reading that consists in examples of CTF metadata, with template values:
872 major = value; /* Trace format version */
874 uuid = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"; /* Trace UUID */
880 /* Type 1 - Few event IDs; Type 2 - Many event IDs. See section 6.1. */
881 event.header := event_header_1 OR event_header_2;
882 event.context := struct {
885 packet.context := struct {
892 id = value; /* Numeric identifier within the stream */
902 /* More detail on types in section 4. Types */
907 * Type declarations behave similarly to the C standard.
910 typedef aliased_type_specifiers new_type_declarators;
912 /* e.g.: typedef struct example new_type_name[10]; */
917 * The "typealias" declaration can be used to give a name (including
918 * pointer declarator specifier) to a type. It should also be used to
919 * map basic C types (float, int, unsigned long, ...) to a CTF type.
920 * Typealias is a superset of "typedef": it also allows assignment of a
921 * simple variable identifier to a type.
924 typealias type_class {
926 } := type_specifiers type_declarator;
930 * typealias integer {
934 * } := struct page *;
936 * typealias integer {
951 enum name : integer_type {
957 * Unnamed types, contained within compound type fields, typedef or typealias.
968 enum : integer_type {
972 typedef type new_type[length];
975 type field_name[length];
978 typedef type new_type[length_type];
981 type field_name[length_type];
993 integer_type field_name:size; /* GNU/C bitfield */
1003 The two following macros keep track of the size of a GNU/C structure without
1004 padding at the end by placing HEADER_END as the last field. A one byte end field
1005 is used for C90 compatibility (C99 flexible arrays could be used here). Note
1006 that this does not affect the effective structure size, which should always be
1007 calculated with the header_sizeof() helper.
1009 #define HEADER_END char end_field
1010 #define header_sizeof(type) offsetof(typeof(type), end_field)
1013 B. Stream Header Rationale
1015 An event stream is divided in contiguous event packets of variable size. These
1016 subdivisions allow the trace analyzer to perform a fast binary search by time
1017 within the stream (typically requiring to index only the event packet headers)
1018 without reading the whole stream. These subdivisions have a variable size to
1019 eliminate the need to transfer the event packet padding when partially filled
1020 event packets must be sent when streaming a trace for live viewing/analysis.
1021 An event packet can contain a certain amount of padding at the end. Dividing
1022 streams into event packets is also useful for network streaming over UDP and
1023 flight recorder mode tracing (a whole event packet can be swapped out of the
1024 buffer atomically for reading).
1026 The stream header is repeated at the beginning of each event packet to allow
1027 flexibility in terms of:
1029 - streaming support,
1030 - allowing arbitrary buffers to be discarded without making the trace
1032 - allow UDP packet loss handling by either dealing with missing event packet
1033 or asking for re-transmission.
1034 - transparently support flight recorder mode,
1035 - transparently support crash dump.
1037 The event stream header will therefore be referred to as the "event packet
1038 header" throughout the rest of this document.
1040 C. CTF Metadata Grammar
1043 * Common Trace Format (CTF) Metadata Grammar.
1045 * Inspired from the C99 grammar:
1046 * http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf (Annex A)
1048 * Specialized for CTF needs by including only constant and declarations from
1049 * C99 (excluding function declarations), and by adding support for variants,
1050 * sequences and CTF-specific specifiers.
1055 1.1) Lexical elements
1098 identifier identifier-nondigit
1101 identifier-nondigit:
1103 universal-character-name
1104 any other implementation-defined characters
1108 [a-zA-Z] /* regular expression */
1111 [0-9] /* regular expression */
1113 1.4) Universal character names
1115 universal-character-name:
1117 \U hex-quad hex-quad
1120 hexadecimal-digit hexadecimal-digit hexadecimal-digit hexadecimal-digit
1126 enumeration-constant
1130 decimal-constant integer-suffix-opt
1131 octal-constant integer-suffix-opt
1132 hexadecimal-constant integer-suffix-opt
1136 decimal-constant digit
1140 octal-constant octal-digit
1142 hexadecimal-constant:
1143 hexadecimal-prefix hexadecimal-digit
1144 hexadecimal-constant hexadecimal-digit
1154 unsigned-suffix long-suffix-opt
1155 unsigned-suffix long-long-suffix
1156 long-suffix unsigned-suffix-opt
1157 long-long-suffix unsigned-suffix-opt
1173 digit-sequence digit
1175 hexadecimal-digit-sequence:
1177 hexadecimal-digit-sequence hexadecimal-digit
1179 enumeration-constant:
1185 L' c-char-sequence '
1189 c-char-sequence c-char
1192 any member of source charset except single-quote ('), backslash
1193 (\), or new-line character.
1197 simple-escape-sequence
1198 octal-escape-sequence
1199 hexadecimal-escape-sequence
1200 universal-character-name
1202 simple-escape-sequence: one of
1203 \' \" \? \\ \a \b \f \n \r \t \v
1205 octal-escape-sequence:
1207 \ octal-digit octal-digit
1208 \ octal-digit octal-digit octal-digit
1210 hexadecimal-escape-sequence:
1211 \x hexadecimal-digit
1212 hexadecimal-escape-sequence hexadecimal-digit
1214 1.6) String literals
1217 " s-char-sequence-opt "
1218 L" s-char-sequence-opt "
1222 s-char-sequence s-char
1225 any member of source charset except double-quote ("), backslash
1226 (\), or new-line character.
1232 [ ] ( ) { } . -> * + - < > : ; ... = ,
1235 2) Phrase structure grammar
1241 ( unary-expression )
1245 postfix-expression [ unary-expression ]
1246 postfix-expression . identifier
1247 postfix-expressoin -> identifier
1251 unary-operator postfix-expression
1253 unary-operator: one of
1256 assignment-operator:
1259 type-assignment-operator:
1262 constant-expression:
1265 constant-expression-range:
1266 constant-expression ... constant-expression
1271 declaration-specifiers declarator-list-opt ;
1274 declaration-specifiers:
1275 storage-class-specifier declaration-specifiers-opt
1276 type-specifier declaration-specifiers-opt
1277 type-qualifier declaration-specifiers-opt
1281 declarator-list , declarator
1283 abstract-declarator-list:
1285 abstract-declarator-list , abstract-declarator
1287 storage-class-specifier:
1310 struct identifier-opt { struct-or-variant-declaration-list-opt }
1313 struct-or-variant-declaration-list:
1314 struct-or-variant-declaration
1315 struct-or-variant-declaration-list struct-or-variant-declaration
1317 struct-or-variant-declaration:
1318 specifier-qualifier-list struct-or-variant-declarator-list ;
1319 declaration-specifiers storage-class-specifier declaration-specifiers declarator-list ;
1320 typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list ;
1321 typealias declaration-specifiers abstract-declarator-list := declarator-list ;
1323 specifier-qualifier-list:
1324 type-specifier specifier-qualifier-list-opt
1325 type-qualifier specifier-qualifier-list-opt
1327 struct-or-variant-declarator-list:
1328 struct-or-variant-declarator
1329 struct-or-variant-declarator-list , struct-or-variant-declarator
1331 struct-or-variant-declarator:
1333 declarator-opt : constant-expression
1336 variant identifier-opt variant-tag-opt { struct-or-variant-declaration-list }
1337 variant identifier variant-tag
1343 enum identifier-opt { enumerator-list }
1344 enum identifier-opt { enumerator-list , }
1346 enum identifier-opt : declaration-specifiers { enumerator-list }
1347 enum identifier-opt : declaration-specifiers { enumerator-list , }
1351 enumerator-list , enumerator
1354 enumeration-constant
1355 enumeration-constant = constant-expression
1356 enumeration-constant = constant-expression-range
1362 pointer-opt direct-declarator
1367 direct-declarator [ type-specifier ]
1368 direct-declarator [ constant-expression ]
1370 abstract-declarator:
1371 pointer-opt direct-abstract-declarator
1373 direct-abstract-declarator:
1375 ( abstract-declarator )
1376 direct-abstract-declarator [ type-specifier ]
1377 direct-abstract-declarator [ constant-expression ]
1378 direct-abstract-declarator [ ]
1381 * type-qualifier-list-opt
1382 * type-qualifier-list-opt pointer
1384 type-qualifier-list:
1386 type-qualifier-list type-qualifier
1391 2.3) CTF-specific declarations
1394 event { ctf-assignment-expression-list-opt }
1395 stream { ctf-assignment-expression-list-opt }
1396 trace { ctf-assignment-expression-list-opt }
1397 typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list ;
1398 typealias declaration-specifiers abstract-declarator-list := declarator-list ;
1401 floating_point { ctf-assignment-expression-list-opt }
1402 integer { ctf-assignment-expression-list-opt }
1403 string { ctf-assignment-expression-list-opt }
1405 ctf-assignment-expression-list:
1406 ctf-assignment-expression
1407 ctf-assignment-expression-list ; ctf-assignment-expression
1409 ctf-assignment-expression:
1410 unary-expression assignment-operator unary-expression
1411 unary-expression type-assignment-operator type-specifier
1412 declaration-specifiers storage-class-specifier declaration-specifiers declarator-list
1413 typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list
1414 typealias declaration-specifiers abstract-declarator-list := declarator-list