[ctf.git] / common-trace-format-proposal.txt

RFC: Common Trace Format (CTF) Proposal (pre-v1.7)

Mathieu Desnoyers, EfficiOS Inc.

The goal of the present document is to propose a trace format that suits the
needs of the embedded, telecom, high-performance and kernel communities. It is
based on the Common Trace Format Requirements (v1.4) document. It is designed to
allow traces to be natively generated by the Linux kernel, Linux user-space
applications written in C/C++, and hardware components. One major element of
CTF is the Trace Stream Description Language (TSDL) which flexibility
enables description of various binary trace stream layouts.

The latest version of this document can be found at:

  git tree:   git://git.efficios.com/ctf.git
  gitweb:     http://git.efficios.com/?p=ctf.git

A reference implementation of a library to read and write this trace format is
being implemented within the BabelTrace project, a converter between trace
formats. The development tree is available at:

  git tree:   git://git.efficios.com/babeltrace.git
  gitweb:     http://git.efficios.com/?p=babeltrace.git


1. Preliminary definitions

  - Event Trace: An ordered sequence of events.
  - Event Stream: An ordered sequence of events, containing a subset of the
                  trace event types.
  - Event Packet: A sequence of physically contiguous events within an event
                  stream.
  - Event: This is the basic entry in a trace. (aka: a trace record).
    - An event identifier (ID) relates to the class (a type) of event within
      an event stream.
        e.g. event: irq_entry.
    - An event (or event record) relates to a specific instance of an event
      class.
        e.g. event: irq_entry, at time X, on CPU Y
  - Source Architecture: Architecture writing the trace.
  - Reader Architecture: Architecture reading the trace.


2. High-level representation of a trace

A trace is divided into multiple event streams. Each event stream contains a
subset of the trace event types.

The final output of the trace, after its generation and optional transport over
the network, is expected to be either on permanent or temporary storage in a
virtual file system. Because each event stream is appended to while a trace is
being recorded, each is associated with a separate file for output.  Therefore,
a stored trace can be represented as a directory containing one file per stream.

Meta-data description associated with the trace contains information on
trace event types expressed in the Trace Stream Description Language
(TSDL). This language describes:

- Trace version.
- Types available.
- Per-trace event header description.
- Per-stream event header description.
- Per-stream event context description.
- Per-event
  - Event type to stream mapping.
  - Event type to name mapping.
  - Event type to ID mapping.
  - Event context description.
  - Event fields description.


3. Event stream

An event stream can be divided into contiguous event packets of variable
size. These subdivisions have a variable size. An event packet can
contain a certain amount of padding at the end. The stream header is
repeated at the beginning of each event packet. The rationale for the
event stream design choices is explained in Appendix B. Stream Header
Rationale.

The event stream header will therefore be referred to as the "event packet
header" throughout the rest of this document.


4. Types

Types are organized as type classes. Each type class belong to either of two
kind of types: basic types or compound types.

4.1 Basic types

A basic type is a scalar type, as described in this section. It includes
integers, GNU/C bitfields, enumerations, and floating point values.

4.1.1 Type inheritance

Type specifications can be inherited to allow deriving types from a
type class. For example, see the uint32_t named type derived from the "integer"
type class below ("Integers" section). Types have a precise binary
representation in the trace. A type class has methods to read and write these
types, but must be derived into a type to be usable in an event field.

4.1.2 Alignment

We define "byte-packed" types as aligned on the byte size, namely 8-bit.
We define "bit-packed" types as following on the next bit, as defined by the
"Integers" section.

Each basic type must specify its alignment, in bits. Examples of
possible alignments are: bit-packed (align = 1), byte-packed (align =
8), or word-aligned (e.g. align = 32 or align = 64). The choice depends
on the architecture preference and compactness vs performance trade-offs
of the implementation.  Architectures providing fast unaligned write
byte-packed basic types to save space, aligning each type on byte
boundaries (8-bit). Architectures with slow unaligned writes align types
on specific alignment values. If no specific alignment is declared for a
type, it is assumed to be bit-packed for integers with size not multiple
of 8 bits and for gcc bitfields. All other basic types are byte-packed
by default. It is however recommended to always specify the alignment
explicitly. Alignment values must be power of two. Compound types are
aligned as specified in their individual specification.

TSDL meta-data attribute representation of a specific alignment:

  align = value;                                /* value in bits */

4.1.3 Byte order

By default, the native endianness of the source architecture the trace is used.
Byte order can be overridden for a basic type by specifying a "byte_order"
attribute. Typical use-case is to specify the network byte order (big endian:
"be") to save data captured from the network into the trace without conversion.
If not specified, the byte order is native.

TSDL meta-data representation:

  byte_order = native OR network OR be OR le;   /* network and be are aliases */

4.1.4 Size

Type size, in bits, for integers and floats is that returned by "sizeof()" in C
multiplied by CHAR_BIT.
We require the size of "char" and "unsigned char" types (CHAR_BIT) to be fixed
to 8 bits for cross-endianness compatibility.

TSDL meta-data representation:

  size = value;    (value is in bits)

4.1.5 Integers

Signed integers are represented in two-complement. Integer alignment,
size, signedness and byte ordering are defined in the TSDL meta-data.
Integers aligned on byte size (8-bit) and with length multiple of byte
size (8-bit) correspond to the C99 standard integers. In addition,
integers with alignment and/or size that are _not_ a multiple of the
byte size are permitted; these correspond to the C99 standard bitfields,
with the added specification that the CTF integer bitfields have a fixed
binary representation. A MIT-licensed reference implementation of the
CTF portable bitfields is available at:

  http://git.efficios.com/?p=babeltrace.git;a=blob;f=include/babeltrace/bitfield.h

Binary representation of integers:

- On little and big endian:
  - Within a byte, high bits correspond to an integer high bits, and low bits
    correspond to low bits.
- On little endian:
  - Integer across multiple bytes are placed from the less significant to the
    most significant.
  - Consecutive integers are placed from lower bits to higher bits (even within
    a byte).
- On big endian:
  - Integer across multiple bytes are placed from the most significant to the
    less significant.
  - Consecutive integers are placed from higher bits to lower bits (even within
    a byte).

This binary representation is derived from the bitfield implementation in GCC
for little and big endian. However, contrary to what GCC does, integers can
cross units boundaries (no padding is required). Padding can be explicitly
added (see 4.1.6 GNU/C bitfields) to follow the GCC layout if needed.

TSDL meta-data representation:

  integer {
    signed = true OR false;                     /* default false */
    byte_order = native OR network OR be OR le; /* default native */
    size = value;                               /* value in bits, no default */
    align = value;                              /* value in bits */
  }

Example of type inheritance (creation of a uint32_t named type):

typealias integer {
  size = 32;
  signed = false;
  align = 32;
} := uint32_t;

Definition of a named 5-bit signed bitfield:

typealias integer {
  size = 5;
  signed = true;
  align = 1;
} := int5_t;

4.1.6 GNU/C bitfields

The GNU/C bitfields follow closely the integer representation, with a
particularity on alignment: if a bitfield cannot fit in the current unit, the
unit is padded and the bitfield starts at the following unit. The unit size is
defined by the size of the type "unit_type".

TSDL meta-data representation:

  unit_type name:size:

As an example, the following structure declared in C compiled by GCC:

struct example {
  short a:12;
  short b:5;
};

The example structure is aligned on the largest element (short). The second
bitfield would be aligned on the next unit boundary, because it would not fit in
the current unit.

4.1.7 Floating point

The floating point values byte ordering is defined in the TSDL meta-data.

Floating point values follow the IEEE 754-2008 standard interchange formats.
Description of the floating point values include the exponent and mantissa size
in bits. Some requirements are imposed on the floating point values:

- FLT_RADIX must be 2.
- mant_dig is the number of digits represented in the mantissa. It is specified
  by the ISO C99 standard, section 5.2.4, as FLT_MANT_DIG, DBL_MANT_DIG and
  LDBL_MANT_DIG as defined by <float.h>.
- exp_dig is the number of digits represented in the exponent. Given that
  mant_dig is one bit more than its actual size in bits (leading 1 is not
  needed) and also given that the sign bit always takes one bit, exp_dig can be
  specified as:

  - sizeof(float) * CHAR_BIT - FLT_MANT_DIG
  - sizeof(double) * CHAR_BIT - DBL_MANT_DIG
  - sizeof(long double) * CHAR_BIT - LDBL_MANT_DIG

TSDL meta-data representation:

floating_point {
  exp_dig = value;
  mant_dig = value;
  byte_order = native OR network OR be OR le;
  align = value;
}

Example of type inheritance:

typealias floating_point {
  exp_dig = 8;         /* sizeof(float) * CHAR_BIT - FLT_MANT_DIG */
  mant_dig = 24;       /* FLT_MANT_DIG */
  byte_order = native;
  align = 32;
} := float;

TODO: define NaN, +inf, -inf behavior.

Bit-packed, byte-packed or larger alignments can be used for floating
point values, similarly to integers.

4.1.8 Enumerations

Enumerations are a mapping between an integer type and a table of strings. The
numerical representation of the enumeration follows the integer type specified
by the meta-data. The enumeration mapping table is detailed in the enumeration
description within the meta-data. The mapping table maps inclusive value
ranges (or single values) to strings. Instead of being limited to simple
"value -> string" mappings, these enumerations map
"[ start_value ... end_value ] -> string", which map inclusive ranges of
values to strings.  An enumeration from the C language can be represented in
this format by having the same start_value and end_value for each element, which
is in fact a range of size 1. This single-value range is supported without
repeating the start and end values with the value = string declaration.

enum name : integer_type {
  somestring          = start_value1 ... end_value1,
  "other string"      = start_value2 ... end_value2,
  yet_another_string,   /* will be assigned to end_value2 + 1 */
  "some other string" = value,
  ...
};

If the values are omitted, the enumeration starts at 0 and increment of 1 for
each entry:

enum name : unsigned int {
  ZERO,
  ONE,
  TWO,
  TEN = 10,
  ELEVEN,
};

Overlapping ranges within a single enumeration are implementation defined.

A nameless enumeration can be declared as a field type or as part of a typedef:

enum : integer_type {
  ...
}

Enumerations omitting the container type ": integer_type" use the "int"
type (for compatibility with C99). The "int" type must be previously
declared. E.g.:

typealias integer { size = 32; align = 32; signed = true } := int;

enum {
  ...
}


4.2 Compound types

Compound are aggregation of type declarations. Compound types include
structures, variant, arrays, sequences, and strings.

4.2.1 Structures

Structures are aligned on the largest alignment required by basic types
contained within the structure. (This follows the ISO/C standard for structures)

TSDL meta-data representation of a named structure:

struct name {
  field_type field_name;
  field_type field_name;
  ...
}; 

Example:

struct example {
  integer {                       /* Nameless type */
    size = 16;
    signed = true;
    align = 16;
  } first_field_name;
  uint64_t second_field_name;  /* Named type declared in the meta-data */
};

The fields are placed in a sequence next to each other. They each possess a
field name, which is a unique identifier within the structure.

A nameless structure can be declared as a field type or as part of a typedef:

struct {
  ...
}

Alignment for a structure compound type can be forced to a minimum value
by adding an "align" specifier after the declaration of a structure
body. This attribute is read as: align(value). The value is specified in
bits. The structure will be aligned on the maximum value between this
attribute and the alignment required by the basic types contained within
the structure. e.g.

struct {
  ...
} align(32)

4.2.2 Variants (Discriminated/Tagged Unions)

A CTF variant is a selection between different types. A CTF variant must
always be defined within the scope of a structure or within fields
contained within a structure (defined recursively). A "tag" enumeration
field must appear in either the same lexical scope, prior to the variant
field (in field declaration order), in an uppermost lexical scope (see
Section 7.3.1), or in an uppermost dynamic scope (see Section 7.3.2).
The type selection is indicated by the mapping from the enumeration
value to the string used as variant type selector. The field to use as
tag is specified by the "tag_field", specified between "< >" after the
"variant" keyword for unnamed variants, and after "variant name" for
named variants.

The alignment of the variant is the alignment of the type as selected by the tag
value for the specific instance of the variant. The alignment of the type
containing the variant is independent of the variant alignment.  The size of the
variant is the size as selected by the tag value for the specific instance of
the variant.

A named variant declaration followed by its definition within a structure
declaration:

variant name {
  field_type sel1;
  field_type sel2;
  field_type sel3;
  ...
};

struct {
  enum : integer_type { sel1, sel2, sel3, ... } tag_field;
  ...
  variant name <tag_field> v;
}

An unnamed variant definition within a structure is expressed by the following
TSDL meta-data:

struct {
  enum : integer_type { sel1, sel2, sel3, ... } tag_field;
  ...
  variant <tag_field> {
    field_type sel1;
    field_type sel2;
    field_type sel3;
    ...
  } v;
}

Example of a named variant within a sequence that refers to a single tag field:

variant example {
  uint32_t a;
  uint64_t b;
  short c;
};

struct {
  enum : uint2_t { a, b, c } choice;
  variant example <choice> v[unsigned int];
}

Example of an unnamed variant:

struct {
  enum : uint2_t { a, b, c, d } choice;
  /* Unrelated fields can be added between the variant and its tag */
  int32_t somevalue;
  variant <choice> {
    uint32_t a;
    uint64_t b;
    short c;
    struct {
      unsigned int field1;
      uint64_t field2;
    } d;
  } s;
}

Example of an unnamed variant within an array:

struct {
  enum : uint2_t { a, b, c } choice;
  variant <choice> {
    uint32_t a;
    uint64_t b;
    short c;
  } v[10];
}

Example of a variant type definition within a structure, where the defined type
is then declared within an array of structures. This variant refers to a tag
located in an upper lexical scope. This example clearly shows that a variant
type definition referring to the tag "x" uses the closest preceding field from
the lexical scope of the type definition.

struct {
  enum : uint2_t { a, b, c, d } x;

  typedef variant <x> { /*
                         * "x" refers to the preceding "x" enumeration in the
                         * lexical scope of the type definition.
                         */
    uint32_t a;
    uint64_t b;
    short c;
  } example_variant;

  struct {
    enum : int { x, y, z } x;   /* This enumeration is not used by "v". */
    example_variant v;          /*
                                 * "v" uses the "enum : uint2_t { a, b, c, d }"
                                 * tag.
                                 */
  } a[10];
}

4.2.3 Arrays

Arrays are fixed-length. Their length is declared in the type
declaration within the meta-data. They contain an array of "inner type"
elements, which can refer to any type not containing the type of the
array being declared (no circular dependency). The length is the number
of elements in an array.

TSDL meta-data representation of a named array:

typedef elem_type name[length];

A nameless array can be declared as a field type within a structure, e.g.:

  uint8_t field_name[10];

Arrays are always aligned on their element alignment requirement.

4.2.4 Sequences

Sequences are dynamically-sized arrays. They start with an integer that specify
the length of the sequence, followed by an array of "inner type" elements.
The length is the number of elements in the sequence.

TSDL meta-data representation for a named sequence:

typedef elem_type name[length_type];

A nameless sequence can be declared as a field type, e.g.:

long field_name[int];

The length type follows the integer types specifications, and the sequence
elements follow the "array" specifications.

4.2.5 Strings

Strings are an array of bytes of variable size and are terminated by a '\0'
"NULL" character.  Their encoding is described in the TSDL meta-data. In
absence of encoding attribute information, the default encoding is
UTF-8.

TSDL meta-data representation of a named string type:

typealias string {
  encoding = UTF8 OR ASCII;
} := name;

A nameless string type can be declared as a field type:

string field_name;      /* Use default UTF8 encoding */

Strings are always aligned on byte size.

5. Event Packet Header

The event packet header consists of two parts: the "event packet header"
is the same for all streams of a trace. The second part, the "event
packet context", is described on a per-stream basis. Both are described
in the TSDL meta-data. The packets are aligned on architecture-page-sized
addresses.

Event packet header (all fields are optional, specified by TSDL meta-data):

- Magic number (CTF magic number: 0xC1FC1FC1) specifies that this is a
  CTF packet. This magic number is optional, but when present, it should
  come at the very beginning of the packet.
- Trace UUID, used to ensure the event packet match the meta-data used.
  (note: we cannot use a meta-data checksum in every cases instead of a
   UUID because meta-data can be appended to while tracing is active)
  This field is optional.
- Stream ID, used as reference to stream description in meta-data.
  This field is optional if there is only one stream description in the
  meta-data, but becomes required if there are more than one stream in
  the TSDL meta-data description.

Event packet context (all fields are optional, specified by TSDL meta-data):

- Event packet content size (in bytes).
- Event packet size (in bytes, includes padding).
- Event packet content checksum (optional). Checksum excludes the event packet
  header.
- Per-stream event packet sequence count (to deal with UDP packet loss). The
  number of significant sequence counter bits should also be present, so
  wrap-arounds are dealt with correctly.
- Time-stamp at the beginning and time-stamp at the end of the event packet.
  Both timestamps are written in the packet header, but sampled respectively
  while (or before) writing the first event and while (or after) writing the
  last event in the packet. The inclusive range between these timestamps should
  include all event timestamps assigned to events contained within the packet.
- Events discarded count
  - Snapshot of a per-stream free-running counter, counting the number of
    events discarded that were supposed to be written in the stream prior to
    the first event in the event packet.
    * Note: producer-consumer buffer full condition should fill the current
            event packet with padding so we know exactly where events have been
            discarded.
- Lossless compression scheme used for the event packet content. Applied
  directly to raw data. New types of compression can be added in following
  versions of the format.
  0: no compression scheme
  1: bzip2
  2: gzip
  3: xz
- Cypher used for the event packet content. Applied after compression.
  0: no encryption
  1: AES
- Checksum scheme used for the event packet content. Applied after encryption.
  0: no checksum
  1: md5
  2: sha1
  3: crc32

5.1 Event Packet Header Description

The event packet header layout is indicated by the trace packet.header
field. Here is a recommended structure type for the packet header with
the fields typically expected (although these fields are each optional):

struct event_packet_header {
  uint32_t magic;
  uint8_t  trace_uuid[16];
  uint32_t stream_id;
};

trace {
  ...
  packet.header := struct event_packet_header;
};

If the magic number is not present, tools such as "file" will have no
mean to discover the file type.

If the trace_uuid is not present, no validation that the meta-data
actually corresponds to the stream is performed.

If the stream_id packet header field is missing, the trace can only
contain a single stream. Its "id" field can be left out, and its events
don't need to declare a "stream_id" field.


5.2 Event Packet Context Description

Event packet context example. These are declared within the stream declaration
in the meta-data. All these fields are optional. If the packet size field is
missing, the whole stream only contains a single packet. If the content
size field is missing, the packet is filled (no padding). The content
and packet sizes include all headers.

An example event packet context type:

struct event_packet_context {
  uint64_t timestamp_begin;
  uint64_t timestamp_end;
  uint32_t checksum;
  uint32_t stream_packet_count;
  uint32_t events_discarded;
  uint32_t cpu_id;
  uint32_t/uint16_t content_size;
  uint32_t/uint16_t packet_size;
  uint8_t  stream_packet_count_bits;    /* Significant counter bits */
  uint8_t  compression_scheme;
  uint8_t  encryption_scheme;
  uint8_t  checksum_scheme;
};


6. Event Structure

The overall structure of an event is:

1 - Stream Packet Context (as specified by the stream meta-data)
 2 - Event Header (as specified by the stream meta-data)
  3 - Stream Event Context (as specified by the stream meta-data)
   4 - Event Context (as specified by the event meta-data)
    5 - Event Payload (as specified by the event meta-data)

This structure defines an implicit dynamic scoping, where variants
located in inner structures (those with a higher number in the listing
above) can refer to the fields of outer structures (with lower number in
the listing above). See Section 7.3 TSDL Scopes for more detail.

6.1 Event Header

Event headers can be described within the meta-data. We hereby propose, as an
example, two types of events headers. Type 1 accommodates streams with less than
31 event IDs. Type 2 accommodates streams with 31 or more event IDs.

One major factor can vary between streams: the number of event IDs assigned to
a stream. Luckily, this information tends to stay relatively constant (modulo
event registration while trace is being recorded), so we can specify different
representations for streams containing few event IDs and streams containing
many event IDs, so we end up representing the event ID and time-stamp as
densely as possible in each case.

The header is extended in the rare occasions where the information cannot be
represented in the ranges available in the standard event header. They are also
used in the rare occasions where the data required for a field could not be
collected: the flag corresponding to the missing field within the missing_fields
array is then set to 1.

Types uintX_t represent an X-bit unsigned integer, as declared with
either:

  typealias integer { size = X; align = X; signed = false } := uintX_t;

    or

  typealias integer { size = X; align = 1; signed = false } := uintX_t;

6.1.1 Type 1 - Few event IDs

  - Aligned on 32-bit (or 8-bit if byte-packed, depending on the architecture
    preference).
  - Native architecture byte ordering.
  - For "compact" selection
    - Fixed size: 32 bits.
  - For "extended" selection
    - Size depends on the architecture and variant alignment.

struct event_header_1 {
  /*
   * id: range: 0 - 30.
   * id 31 is reserved to indicate an extended header.
   */
  enum : uint5_t { compact = 0 ... 30, extended = 31 } id;
  variant <id> {
    struct {
      uint27_t timestamp;
    } compact;
    struct {
      uint32_t id;                       /* 32-bit event IDs */
      uint64_t timestamp;                /* 64-bit timestamps */
    } extended;
  } v;
};


6.1.2 Type 2 - Many event IDs

  - Aligned on 16-bit (or 8-bit if byte-packed, depending on the architecture
    preference).
  - Native architecture byte ordering.
  - For "compact" selection
    - Size depends on the architecture and variant alignment.
  - For "extended" selection
    - Size depends on the architecture and variant alignment.

struct event_header_2 {
  /*
   * id: range: 0 - 65534.
   * id 65535 is reserved to indicate an extended header.
   */
  enum : uint16_t { compact = 0 ... 65534, extended = 65535 } id;
  variant <id> {
    struct {
      uint32_t timestamp;
    } compact;
    struct {
      uint32_t id;                       /* 32-bit event IDs */
      uint64_t timestamp;                /* 64-bit timestamps */ 
    } extended;
  } v;
};


6.2 Event Context

The event context contains information relative to the current event.
The choice and meaning of this information is specified by the TSDL
stream and event meta-data descriptions. The stream context is applied
to all events within the stream. The stream context structure follows
the event header. The event context is applied to specific events. Its
structure follows the stream context structure.

An example of stream-level event context is to save the event payload size with
each event, or to save the current PID with each event.  These are declared
within the stream declaration within the meta-data:

  stream {
    ...
    event.context := struct {
        uint pid;
        uint16_t payload_size;
    };
  };

An example of event-specific event context is to declare a bitmap of missing
fields, only appended after the stream event context if the extended event
header is selected. NR_FIELDS is the number of fields within the event (a
numeric value).

  event {
    context = struct {
      variant <id> {
        struct { } compact;
        struct {
          uint1_t missing_fields[NR_FIELDS]; /* missing event fields bitmap */
        } extended;
      } v;
    };
    ...
  }

6.3 Event Payload

An event payload contains fields specific to a given event type. The fields
belonging to an event type are described in the event-specific meta-data
within a structure type.

6.3.1 Padding

No padding at the end of the event payload. This differs from the ISO/C standard
for structures, but follows the CTF standard for structures. In a trace, even
though it makes sense to align the beginning of a structure, it really makes no
sense to add padding at the end of the structure, because structures are usually
not followed by a structure of the same type.

This trick can be done by adding a zero-length "end" field at the end of the C
structures, and by using the offset of this field rather than using sizeof()
when calculating the size of a structure (see Appendix "A. Helper macros").

6.3.2 Alignment

The event payload is aligned on the largest alignment required by types
contained within the payload. (This follows the ISO/C standard for structures)


7. Trace Stream Description Language (TSDL)

The Trace Stream Description Language (TSDL) allows expression of the
binary trace streams layout in a C99-like Domain Specific Language
(DSL).


7.1 Meta-data

The trace stream layout description is located in the trace meta-data.
The meta-data is itself located in a stream identified by its name:
"metadata".

The meta-data description can be expressed in two different formats:
text-only and packet-based. The text-only description facilitates
generation of meta-data and provides a convenient way to enter the
meta-data information by hand. The packet-based meta-data provides the
CTF stream packet facilities (checksumming, compression, encryption,
network-readiness) for meta-data stream generated and transported by a
tracer.

The text-only meta-data file is a plain text TSDL description.

The packet-based meta-data is made of "meta-data packets", which each
start with a meta-data packet header. The packet-based meta-data
description is detected by reading the magic number "0x75D11D57" at the
beginning of the file. This magic number is also used to detect the
endianness of the architecture by trying to read the CTF magic number
and its counterpart in reversed endianness. The events within the
meta-data stream have no event header nor event context. Each event only
contains a "string" payload. Each meta-data packet start with a special
packet header, specific to the meta-data stream, which contains,
exactly:

struct metadata_packet_header {
  uint32_t magic;                       /* 0x75D11D57 */
  uint8_t  trace_uuid[16];              /* Unique Universal Identifier */
  uint32_t checksum;                    /* 0 if unused */
  uint32_t content_size;                /* in bits */
  uint32_t packet_size;                 /* in bits */
  uint8_t  compression_scheme;          /* 0 if unused */
  uint8_t  encryption_scheme;           /* 0 if unused */
  uint8_t  checksum_scheme;             /* 0 if unused */
};

The packet-based meta-data can be converted to a text-only meta-data by
concatenating all the strings in contains.

In the textual representation of the meta-data, the text contained
within "/*" and "*/", as well as within "//" and end of line, are
treated as comments.  Boolean values can be represented as true, TRUE,
or 1 for true, and false, FALSE, or 0 for false. Within the string-based
meta-data description, the trace UUID is represented as a string of
hexadecimal digits and dashes "-". In the event packet header, the trace
UUID is represented as an array of bytes.


7.2 Declaration vs Definition

A declaration associates a layout to a type, without specifying where
this type is located in the event structure hierarchy (see Section 6).
This therefore includes typedef, typealias, as well as all type
specifiers. In certain circumstances (typedef, structure field and
variant field), a declaration is followed by a declarator, which specify
the newly defined type name (for typedef), or the field name (for
declarations located within structure and variants). Array and sequence,
declared with square brackets ("[" "]"), are part of the declarator,
similarly to C99. The enumeration base type is specified by
": enum_base", which is part of the type specifier. The variant tag
name, specified between "<" ">", is also part of the type specifier.

A definition associates a type to a location in the event structure
hierarchy (see Section 6). This association is denoted by ":=", as shown
in Section 7.3.


7.3 TSDL Scopes

TSDL uses two different types of scoping: a lexical scope is used for
declarations and type definitions, and a dynamic scope is used for
variants references to tag fields.

7.3.1 Lexical Scope

Each of "trace", "stream", "event", "struct" and "variant" have their own
nestable declaration scope, within which types can be declared using "typedef"
and "typealias". A root declaration scope also contains all declarations
located outside of any of the aforementioned declarations. An inner
declaration scope can refer to type declared within its container
lexical scope prior to the inner declaration scope. Redefinition of a
typedef or typealias is not valid, although hiding an upper scope
typedef or typealias is allowed within a sub-scope.

7.3.2 Dynamic Scope

A dynamic scope consists in the lexical scope augmented with the
implicit event structure definition hierarchy presented at Section 6.
The dynamic scope is only used for variant tag definitions. It is used
at definition time to look up the location of the tag field associated
with a variant.

Therefore, variants in lower levels in the dynamic scope (e.g. event
context) can refer to a tag field located in upper levels (e.g. in the
event header) by specifying, in this case, the associated tag with
<header.field_name>. This allows, for instance, the event context to
define a variant referring to the "id" field of the event header as
selector.

The target dynamic scope must be specified explicitly when referring to
a field outside of the local static scope. The dynamic scope prefixes
are thus:

 - Trace Packet Header: <trace.packet.header. >,
 - Stream Packet Context: <stream.packet.context. >,
 - Event Header: <stream.event.header. >,
 - Stream Event Context: <stream.event.context. >,
 - Event Context: <event.context. >,
 - Event Payload: <event.fields. >.

Multiple declarations of the same field name within a single scope is
not valid. It is however valid to re-use the same field name in
different scopes. There is no possible conflict, because the dynamic
scope must be specified when a variant refers to a tag field located in
a different dynamic scope.

The information available in the dynamic scopes can be thought of as the
current tracing context. At trace production, information about the
current context is saved into the specified scope field levels. At trace
consumption, for each event, the current trace context is therefore
readable by accessing the upper dynamic scopes.


7.4 TSDL Examples

The grammar representing the TSDL meta-data is presented in Appendix C.
TSDL Grammar. This section presents a rather lighter reading that
consists in examples of TSDL meta-data, with template values.

The stream "id" can be left out if there is only one stream in the
trace. The event "id" field can be left out if there is only one event
in a stream.

trace {
  major = value;                                /* Trace format version */
  minor = value;
  uuid = "aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa";        /* Trace UUID */
  byte_order = be OR le;                        /* Endianness (required) */
  packet.header := struct {
    uint32_t magic;
    uint8_t  trace_uuid[16];
    uint32_t stream_id;
  };
};

stream {
  id = stream_id;
  /* Type 1 - Few event IDs; Type 2 - Many event IDs. See section 6.1. */
  event.header := event_header_1 OR event_header_2;
  event.context := struct {
    ...
  };
  packet.context := struct {
    ...
  };
};

event {
  name = event_name;
  id = value;                   /* Numeric identifier within the stream */
  stream_id = stream_id;
  context := struct {
    ...
  };
  fields := struct {
    ...
  };
};

/* More detail on types in section 4. Types */

/*
 * Named types:
 *
 * Type declarations behave similarly to the C standard.
 */

typedef aliased_type_specifiers new_type_declarators;

/* e.g.: typedef struct example new_type_name[10]; */

/*
 * typealias
 *
 * The "typealias" declaration can be used to give a name (including
 * pointer declarator specifier) to a type. It should also be used to
 * map basic C types (float, int, unsigned long, ...) to a CTF type.
 * Typealias is a superset of "typedef": it also allows assignment of a
 * simple variable identifier to a type.
 */

typealias type_class {
  ...
} := type_specifiers type_declarator;

/*
 * e.g.: 
 * typealias integer {
 *   size = 32;
 *   align = 32;
 *   signed = false;
 * } := struct page *;
 *
 * typealias integer {
 *  size = 32;
 *  align = 32;
 *  signed = true;
 * } := int;
 */

struct name {
  ...
};

variant name {
  ...
};

enum name : integer_type {
  ...
};


/*
 * Unnamed types, contained within compound type fields, typedef or typealias.
 */

struct {
  ...
}

struct {
  ...
} align(value)

variant {
  ...
}

enum : integer_type {
  ...
}

typedef type new_type[length];

struct {
  type field_name[length];
}

typedef type new_type[length_type];

struct {
  type field_name[length_type];
}

integer {
  ...
}

floating_point {
  ...
}

struct {
  integer_type field_name:size;         /* GNU/C bitfield */
}

struct {
  string field_name;
}


A. Helper macros

The two following macros keep track of the size of a GNU/C structure without
padding at the end by placing HEADER_END as the last field. A one byte end field
is used for C90 compatibility (C99 flexible arrays could be used here). Note
that this does not affect the effective structure size, which should always be
calculated with the header_sizeof() helper.

#define HEADER_END              char end_field
#define header_sizeof(type)     offsetof(typeof(type), end_field)


B. Stream Header Rationale

An event stream is divided in contiguous event packets of variable size. These
subdivisions allow the trace analyzer to perform a fast binary search by time
within the stream (typically requiring to index only the event packet headers)
without reading the whole stream. These subdivisions have a variable size to
eliminate the need to transfer the event packet padding when partially filled
event packets must be sent when streaming a trace for live viewing/analysis.
An event packet can contain a certain amount of padding at the end. Dividing
streams into event packets is also useful for network streaming over UDP and
flight recorder mode tracing (a whole event packet can be swapped out of the
buffer atomically for reading).

The stream header is repeated at the beginning of each event packet to allow
flexibility in terms of:

  - streaming support,
  - allowing arbitrary buffers to be discarded without making the trace
    unreadable,
  - allow UDP packet loss handling by either dealing with missing event packet
    or asking for re-transmission.
  - transparently support flight recorder mode,
  - transparently support crash dump.


C. TSDL Grammar

/*
 * Common Trace Format (CTF) Trace Stream Description Language (TSDL) Grammar.
 *
 * Inspired from the C99 grammar:
 * http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1124.pdf (Annex A)
 * and c++1x grammar (draft)
 * http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2011/n3291.pdf (Annex A)
 *
 * Specialized for CTF needs by including only constant and declarations from
 * C99 (excluding function declarations), and by adding support for variants,
 * sequences and CTF-specific specifiers. Enumeration container types
 * semantic is inspired from c++1x enum-base.
 */

1) Lexical grammar

1.1) Lexical elements

token:
        keyword
        identifier
        constant
        string-literal
        punctuator

1.2) Keywords

keyword: is one of

align
const
char
double
enum
event
floating_point
float
integer
int
long
short
signed
stream
string
struct
trace
typealias
typedef
unsigned
variant
void
_Bool
_Complex
_Imaginary


1.3) Identifiers

identifier:
        identifier-nondigit
        identifier identifier-nondigit
        identifier digit

identifier-nondigit:
        nondigit
        universal-character-name
        any other implementation-defined characters

nondigit:
        _
        [a-zA-Z]        /* regular expression */

digit:
        [0-9]           /* regular expression */

1.4) Universal character names

universal-character-name:
        \u hex-quad
        \U hex-quad hex-quad

hex-quad:
        hexadecimal-digit hexadecimal-digit hexadecimal-digit hexadecimal-digit

1.5) Constants

constant:
        integer-constant
        enumeration-constant
        character-constant

integer-constant:
        decimal-constant integer-suffix-opt
        octal-constant integer-suffix-opt
        hexadecimal-constant integer-suffix-opt

decimal-constant:
        nonzero-digit
        decimal-constant digit

octal-constant:
        0
        octal-constant octal-digit

hexadecimal-constant:
        hexadecimal-prefix hexadecimal-digit
        hexadecimal-constant hexadecimal-digit

hexadecimal-prefix:
        0x
        0X

nonzero-digit:
        [1-9]

integer-suffix:
        unsigned-suffix long-suffix-opt
        unsigned-suffix long-long-suffix
        long-suffix unsigned-suffix-opt
        long-long-suffix unsigned-suffix-opt

unsigned-suffix:
        u
        U

long-suffix:
        l
        L

long-long-suffix:
        ll
        LL

enumeration-constant:
        identifier
        string-literal

character-constant:
        ' c-char-sequence '
        L' c-char-sequence '

c-char-sequence:
        c-char
        c-char-sequence c-char

c-char:
        any member of source charset except single-quote ('), backslash
        (\), or new-line character.
        escape-sequence

escape-sequence:
        simple-escape-sequence
        octal-escape-sequence
        hexadecimal-escape-sequence
        universal-character-name

simple-escape-sequence: one of
        \' \" \? \\ \a \b \f \n \r \t \v

octal-escape-sequence:
        \ octal-digit
        \ octal-digit octal-digit
        \ octal-digit octal-digit octal-digit

hexadecimal-escape-sequence:
        \x hexadecimal-digit
        hexadecimal-escape-sequence hexadecimal-digit

1.6) String literals

string-literal:
        " s-char-sequence-opt "
        L" s-char-sequence-opt "

s-char-sequence:
        s-char
        s-char-sequence s-char

s-char:
        any member of source charset except double-quote ("), backslash
        (\), or new-line character.
        escape-sequence

1.7) Punctuators

punctuator: one of
        [ ] ( ) { } . -> * + - < > : ; ... = ,


2) Phrase structure grammar

primary-expression:
        identifier
        constant
        string-literal
        ( unary-expression )

postfix-expression:
        primary-expression
        postfix-expression [ unary-expression ]
        postfix-expression . identifier
        postfix-expressoin -> identifier

unary-expression:
        postfix-expression
        unary-operator postfix-expression

unary-operator: one of
        + -

assignment-operator:
        =

type-assignment-operator:
        :=

constant-expression:
        unary-expression

constant-expression-range:
        constant-expression ... constant-expression

2.2) Declarations:

declaration:
        declaration-specifiers declarator-list-opt ;
        ctf-specifier ;

declaration-specifiers:
        storage-class-specifier declaration-specifiers-opt
        type-specifier declaration-specifiers-opt
        type-qualifier declaration-specifiers-opt

declarator-list:
        declarator
        declarator-list , declarator

abstract-declarator-list:
        abstract-declarator
        abstract-declarator-list , abstract-declarator

storage-class-specifier:
        typedef

type-specifier:
        void
        char
        short
        int
        long
        float
        double
        signed
        unsigned
        _Bool
        _Complex
        _Imaginary
        struct-specifier
        variant-specifier
        enum-specifier
        typedef-name
        ctf-type-specifier

align-attribute:
        align ( constant-expression )

struct-specifier:
        struct identifier-opt { struct-or-variant-declaration-list-opt } align-attribute-opt
        struct identifier align-attribute-opt

struct-or-variant-declaration-list:
        struct-or-variant-declaration
        struct-or-variant-declaration-list struct-or-variant-declaration

struct-or-variant-declaration:
        specifier-qualifier-list struct-or-variant-declarator-list ;
        declaration-specifiers storage-class-specifier declaration-specifiers declarator-list ;
        typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list ;
        typealias declaration-specifiers abstract-declarator-list := declarator-list ;

specifier-qualifier-list:
        type-specifier specifier-qualifier-list-opt
        type-qualifier specifier-qualifier-list-opt

struct-or-variant-declarator-list:
        struct-or-variant-declarator
        struct-or-variant-declarator-list , struct-or-variant-declarator

struct-or-variant-declarator:
        declarator
        declarator-opt : constant-expression

variant-specifier:
        variant identifier-opt variant-tag-opt { struct-or-variant-declaration-list }
        variant identifier variant-tag

variant-tag:
        < identifier >

enum-specifier:
        enum identifier-opt { enumerator-list }
        enum identifier-opt { enumerator-list , }
        enum identifier
        enum identifier-opt : declaration-specifiers { enumerator-list }
        enum identifier-opt : declaration-specifiers { enumerator-list , }

enumerator-list:
        enumerator
        enumerator-list , enumerator

enumerator:
        enumeration-constant
        enumeration-constant = constant-expression
        enumeration-constant = constant-expression-range

type-qualifier:
        const

declarator:
        pointer-opt direct-declarator

direct-declarator:
        identifier
        ( declarator )
        direct-declarator [ type-specifier ]
        direct-declarator [ constant-expression ]

abstract-declarator:
        pointer-opt direct-abstract-declarator

direct-abstract-declarator:
        identifier-opt
        ( abstract-declarator )
        direct-abstract-declarator [ type-specifier ]
        direct-abstract-declarator [ constant-expression ]
        direct-abstract-declarator [ ]

pointer:
        * type-qualifier-list-opt
        * type-qualifier-list-opt pointer

type-qualifier-list:
        type-qualifier
        type-qualifier-list type-qualifier

typedef-name:
        identifier

2.3) CTF-specific declarations

ctf-specifier:
        event { ctf-assignment-expression-list-opt }
        stream { ctf-assignment-expression-list-opt }
        trace { ctf-assignment-expression-list-opt }
        typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list ;
        typealias declaration-specifiers abstract-declarator-list := declarator-list ;

ctf-type-specifier:
        floating_point { ctf-assignment-expression-list-opt }
        integer { ctf-assignment-expression-list-opt }
        string { ctf-assignment-expression-list-opt }

ctf-assignment-expression-list:
        ctf-assignment-expression
        ctf-assignment-expression-list ; ctf-assignment-expression

ctf-assignment-expression:
        unary-expression assignment-operator unary-expression
        unary-expression type-assignment-operator type-specifier
        declaration-specifiers storage-class-specifier declaration-specifiers declarator-list
        typealias declaration-specifiers abstract-declarator-list := declaration-specifiers abstract-declarator-list
        typealias declaration-specifiers abstract-declarator-list := declarator-list