General
The specification language describes protocol message formats based on types. For each type of the specification language a description of its syntax and semantics and an example is given.
A simple variant of Backus-Naur Form is used to describe the syntax.
Optional elements are enclosed in [brackets].
Elements which occur zero or more times are enclosed in {curly brackets}.
Terms are grouped using (parentheses).
Alternatives are delimited by the | character.
Reserved keywords and literals are marked in 'single quotes' or "double quotes".
Literal ranges denote a character range from a start character to an end character separated by .. (e.g. "a" .. "f").
A production can use a non-formal verbal description starting with a # sign (e.g. # All non-whitespace Unicode characters).
Non-terminals are referenced using hyperlinks (e.g. prefix_non_terminal).
To convey some semantic information regarding the usage of the non-terminal in a given context its name can be prefixed by an identifier when referencing it. Non-terminals with prefixes are equivalent to the respective non-terminals without the prefix.
See the section Variable Declaration for an example.
Basic Elements
The following basic elements are used to describe the syntax of the language. The type system is inspired by Ada, but differs in some details. In contrast to Ada, integer variables are considered type-compatible. Explicit type conversions of integer variables are neither required nor supported.
Names
Syntax
name ::=letter{ [ "_" ]alphanum{alphanum} } alphanum ::=letter|dec_digitletter ::= "A" .. "Z" | "a" .. "z" qualified_name ::=name{ "::"name}
By convention a name starts with a capital and after each underscore follows a capital as well. A qualified name starts with the package hierachy the respective entity is located in, separated by ::.
Example
Mixed_Case_With_Underscores
Numbers
Syntax
number ::=baseless_number|bin_number|oct_number|dec_number|hex_numberbaseless_number ::=dec_digit{ [ "_" ]dec_digit{dec_digit} } bin_digit ::= "0" | "1" oct_digit ::= "0" .. "7" dec_digit ::= "0" .. "9" hex_digit ::=dec_digit| "A" .. "F" bin_number ::= "2" "#"bin_digit{ [ "_" ]bin_digit{bin_digit} } "#" oct_number ::= "8" "#"oct_digit{ [ "_" ]oct_digit{oct_digit} } "#" dec_number ::= "10" "#"dec_digit{ [ "_" ]dec_digit{dec_digit} } "#" hex_number ::= "16" "#"hex_digit{ [ "_" ]hex_digit{hex_digit} } "#"
Static Semantics
A baseless number (one not containing a "#" symbol) is considered base 10. Numbers with an explicit base (2, 8, 10, or 16) are considered to have that respective base.
Example
16#DEAD_C0DE#
Strings
Syntax
string ::= '"' { character } '"'
character ::= # Any Unicode character except QUOTATION MARK
Static Semantics
Internally, strings are UTF-8 encoded. There is currently no way to represent the QUOTATION MARK character (U+0022).
Example
"Hello World"
Mathematical Expressions
Expressions of this category yield an integer result.
Syntax
math_expression ::= (math_expression( "+" | "-" )math_unop_term) |math_unop_termmath_unop_term ::= ( "-"math_term) |math_termmath_term ::= (math_term( "*" | "/" | "mod" )math_factor) |math_factormath_factor ::= (math_primary"**"math_primary) |math_suffixmath_suffix ::= (math_suffix"'"math_attribute) |math_primarymath_attribute ::= "First" | "Size" | "Last" math_primary ::=number|qualified_name|selected|paren_math_expressionparen_math_expression ::= "("math_expression")"
Example
(2 ** (V'First + 23) - 12) + 1
Sequence Expressions
Syntax
seq_expression ::=seq_paren_expression|comprehension|aggregate|selected|string|qualified_nameseq_paren_expression ::= "("seq_expression")"
Boolean Expressions
Expressions of this category yield a boolean result. Standard boolean expressions are used in then-clauses and refinements of messages.
Syntax
bool_expression ::= (bool_expression( "and" | "or" )bool_unop_term) |bool_unop_termbool_unop_term ::= ( "not"bool_term) |bool_termbool_term ::= "True" | "False" |bool_suffix|qualified_name|bool_relation|math_relation|seq_relation|bool_paren_expressionbool_suffix ::=name"'"bool_attributebool_attribute ::= "Valid_Checksum" bool_relation ::=bool_expression( "=" | "/=" )bool_expressionmath_relation ::=math_expression( "=" | "/=" | "<=" | "<" | ">=" | ">" )math_expressionseq_relation ::=seq_expression( "=" | "/=" )seq_expressionseq_membership ::=expression( "in" | "not in" )seq_expressionbool_paren_expression ::= "("bool_expression")"
Static Semantics
The semantics of the Valid_Checksum attribute is explained in the section Message Types.
Example
V < 20 and U = True
Extended Boolean Expressions
Expressions of this category yield a boolean result. Extended boolean expressions are used in state machines.
Syntax
ext_bool_expression ::= (ext_bool_expression( "and" | "or" )ext_bool_unop_term) |ext_bool_unop_termext_bool_unop_term ::= ( "not"ext_bool_term) |ext_bool_termext_bool_term ::= "True" | "False" |ext_bool_suffix|qualified_name|ext_bool_relation|math_relation|seq_relation|seq_membership|quantified_expression|selected|ext_bool_paren_expressionext_bool_suffix ::=expression"'"ext_bool_attributeext_bool_attribute ::= "Valid" | "Has_Data" | "Present" ext_bool_relation ::=ext_bool_expression( "=" | "/=" )ext_bool_expressionext_bool_paren_expression ::= "("ext_bool_expression")"
Static Semantics
The Valid attribute allows to determine the validity of a message or sequence.
Whether a channel contains data can be checked with the Has_Data attribute.
Sizes
A size aspect defines the size of a type or message field in bits.
Syntax
size_aspect ::= "Size" "=>"size_expressionsize_expression ::=math_expression
Scalar Types
Integer Types
An integer type is used to represent whole numbers. In RecordFlux integer types can be specified in several ways as explained next.
Syntax
integer_type ::=range_type|unsigned_type
Range Integer Types
A range integer type is the most general form of integer type specifications. It allows one to explicitly specify the lower and upper bound for the type, as well as the size of the type in bits. At the moment only non-negative integers are supported.
Syntax
range_type ::= "type"name"is" "range"first".."last"with"size_aspectfirst ::=math_expressionlast ::=math_expression
Static Semantics
The set of values of a range integer type consists of all numbers from the lower bound to the upper bound. The lower bound must be >= 0 and <= the upper bound. The bit size has to be specified explicitly and must be a value between 1 and 63 (inclusive). It does not have to be a multiple of 8 bits. However, the size of the type must be able to accommodate the upper bound.
Example
type Type_Length is range 46 .. 2 ** 16 - 1 with Size => 16
Unsigned Integer Types
For unsigned integers which cover the whole range of their specified size the following shorthand syntax is available.
Syntax
unsigned_type ::= "type"name"is" "unsigned"size_expression
Static Semantics
The above syntax is equivalent to the definition of a range integer type where the lower limit is 0 and upper limit 2 ** size - 1.
Example
-- Value range: 0 .. 63
type Address is unsigned 6
Enumeration Types
An enumeration type represents a value out of a list of possible values.
Syntax
enumeration_type ::= "type"name"is" "("literals")" "with"enumeration_aspectsliterals ::=enumeration_literal{ ","enumeration_literal} enumeration_literal ::=name[ "=>"number] enumeration_aspects ::=enumeration_aspect{ ","enumeration_aspect} enumeration_aspect ::=size_aspect|always_valid_aspectalways_valid_aspect ::= "Always_Valid" [ "=>" ( "True" | "False" ) ]
Static Semantics
The set of values of an enumeration type consists of the list of declared enumeration literals. Each enumeration literal has a distinct value. If no explicit value is given, the first literal is zero, and the value of each subsequent literal is incremented by one. Literals with and without explicit value must not be intermixed in one definition. The bit size of the enumeration type must be specified explicitly. Optionally, an enumeration type can be flagged as always valid. A message field with such type is always considered valid, whether or not its value corresponds to one of the specified literals.
Example
type Tag is (Msg_Error, Msg_Data) with Size => 1
type Ether_Type is
(ET_IPv4 => 16#0800#,
ET_ARP => 16#0806#,
ET_VLAN_Tag => 16#8100#,
ET_IPv6 => 16#86DD#,
ET_VLAN_Tag_Double => 16#9100#)
with Size => 16, Always_Valid
Boolean
Boolean is a built-in enumeration type with the literals False => 0 and True => 1 with a size of 1 bit.
Parameters and Arguments
Parameters define the objects directly visible within functions or parameterized messages and their associated types.
parameter ::=name":"qualified_nameparameter_list ::= "("parameter{ ";"parameter} ")"
Named arguments associate a parameter with an expression.
named_argument ::= parameter_name"=>"expressionnamed_argument_list ::=named_argument{ ","named_argument}
Message Types
A message type is a collection of fields. Additional then clauses enable the definition of conditions and dependencies between fields.
Syntax
message_type ::= "type"name[parameter_list] "is" ( "message" [null_field]field{field} "end" "message" [ "with"message_aspects] | "null" "message" ) field ::=name":"qualified_name[ "("named_argument_list")" ] [ "with"aspects] {then_clause} ";" null_field ::= "null"then_clause{then_clause} ";" target_field ::= field_name| "null" then_clause ::= "then"target_field[ "with"aspects] [ "if"bool_expression] aspects ::=aspect{ ","aspect} aspect ::=first_aspect|size_aspectfirst_aspect ::= "First" "=>"math_expressionmessage_aspects ::=message_aspect{ ","message_aspect} message_aspect ::=checksum_aspect|byteorder_aspectchecksum_aspect ::= "Checksum" "=>" "("checksum_definition{ ","checksum_definition} ")" checksum_definition ::=name"=>" "("checksum_element{ ","checksum_element} ")" checksum_element ::=name|name"'" "Size" |field_rangefield_range ::=field_range_first".."field_range_lastfield_range_first ::=name"'" "First" |name"'" "Last" "+" "1" field_range_last ::=name"'" "Last" |name"'" "First" "-" "1" byteorder_aspect ::= "Byte_Order" "=>"byteorder_definitionbyteorder_definition ::= "High_Order_First" | "Low_Order_First"
Static Semantics
A message type specifies the message format of a protocol. A message is represented by a graph-based model. Each node in the graph corresponds to one field in a message. The links in the graph define the order of the fields. A link is represented by a then clause in the specification. If no then clause is given, it is assumed that always the next field of the message follows. If no further field follows, it is assumed that the message ends with this field. The end of a message can also be denoted explicitly by adding a then clause to null. Optionally, a then clause can contain a condition under which the corresponding field follows and aspects which enable the definition of the size of the next field and the location of its first bit. These aspects can also be specified for the field directly. Each aspect can be specified either for the field or in all incoming then clauses, but not in both. The condition can refer to previous fields (including the field containing the then clause). If required, a null field can be used to specify the size of the first field in the message. An empty message can be represented by a null message.
A message can be parameterized. Message parameters can be used in conditions and aspects and enable the definition of message formats that depend on prior negotiation. Only scalar types are allowed for parameters.
The field type Opaque represents an unconstrained sequence of bytes.
The size of opaque fields and sequence fields must be defined by a size aspect, if another field can follow.
If no size aspect is given, the field size is implicitly defined by the available space (defined by the outer message when parsing or by the written data when serializing).
Opaque fields and sequence fields must be byte aligned.
The size of a message must be a multiple of 8 bit.
A checksum aspect specifies which parts of a message are covered by a checksum.
The definition of the checksum calculation is not part of the specification.
Code based on the message specification must provide a function which is able to verify a checksum using the specified checksum elements.
A checksum element can be a field value, a field size or a range of fields.
The point where a checksum should be checked during parsing or generated during serialization must be defined for each checksum.
For this purpose the Valid_Checksum attribute is added to a condition.
All message parts on which the checksum depends have to be known at this point.
The Byte_Order aspect allows the user to specify the endianness of the message, with the two possible choices High_Order_First (big endian, or network byte order) and Low_Order_First (little endian).
If the Byte_Order aspect is not specified, the byte order of the message is set to High_Order_First.
Message’First, Message’Last and Message’Size can be used in expressions to refer to the position of the first or last bit of the message or the size of the message.
All bytes which were received when parsing or were written when serializing are considered as part of the message.
Example
type Frame is
message
Destination : Address;
Source : Address;
Type_Length_TPID : Type_Length
then TPID
with First => Type_Length_TPID'First
if Type_Length_TPID = 16#8100#
then Payload
with Size => Type_Length_TPID * 8
if Type_Length_TPID <= 1500
then Ether_Type
with First => Type_Length_TPID'First
if Type_Length_TPID >= 1536 and Type_Length_TPID /= 16#8100#;
TPID : TPID;
TCI : TCI;
Ether_Type : Ether_Type;
Payload : Opaque
then null
if Payload'Size / 8 >= 46 and Payload'Size / 8 <= 1500;
end message
type Empty_Message is null message
Type Refinements
A type refinement describes the relation of an opaque field in a message type to another message type.
Syntax
type_refinement ::= "for" refined_qualified_name"use" "(" refined_field_name"=>" message_qualified_name")" [ "if"bool_expression]
Static Semantics
A type refinement describes under which condition a specific message can be expected inside of a payload field.
Only fields of type Opaque can be refined.
Types defined in other packages are referenced by a qualified name in the form Package_Name::Message_Type_Name.
The condition can refer to fields of the refined type.
To indicate that a refined field is empty (i.e. does not exist) under a certain condition, a null message can be used as message type.
Example
for Ethernet::Frame use (Payload => IPv4::Packet)
if Ether_Type = Ethernet::IPV4
Type Derivations
A type derivation enables the creation of a new message type based on an existing message type.
Syntax
type_derivation ::= "type"name"is" "new" base_type_qualified_name
Static Semantics
A derived message type derives its specification from a base type. Type refinements of a base message type are not inherited by the derived message type.
Example
type Specific_Extension is new Extension
Sequence Types
A sequence type represents a list of similar elements.
Syntax
sequence_type ::= "type"name"is" "sequence" "of" element_qualified_name
Static Semantics
A sequence consists of a number of elements with the same type. Scalar types as well as message types can be used as element type.
Example
type Options is sequence of Option
State Machines
A state machine defines the dynamic behavior of a protocol using a finite state machine. The first defined state is considered the initial state. The external interface of a state machine is defined by parameters. The declaration part enables the declaration of state machine global variables. The main part of a state machine definition is the state definitions.
Syntax
state_machine ::= "generic"
{ state_machine_parameter }
"machine" name "is"
{ state_machine_declaration }
"begin"
state
{ state }
"end" name
Example
generic
X : Channel with Readable, Writable;
with function F return T;
with function G (P : T) return Boolean;
machine S is
Y : Boolean := False;
begin
state A
with Desc => "rfc1149.txt+51:4-52:9"
is
Z : Boolean := Y;
M : TLV::Message;
begin
X'Read (M);
transition
goto null
with Desc => "rfc1149.txt+45:4-47:8"
if Z = True
and G (F) = True
goto A
end A;
end S
State Machine Parameters
Functions and channels can be defined as state machine parameters.
Syntax
state_machine_parameter ::= (function_declaration|channel_declaration) ";"
Functions
Functions enable the execution of externally defined code.
Syntax
function_declaration ::= "with" "function"name[parameter_list] "return" type_qualified_name
Static Semantics
Allowed parameter types:
Scalars
Definite messages
Opaque fields of messages
Allowed return types:
Scalars
Definite messages
Definite messages are messages with no optional fields and an explicit size (i.e. all size aspects contain no reference to Message).
SPARK
For each function declaration in the state machine specification a procedure declaration is added to the corresponding state machine package. The return type and parameters of a function are represented by the first and subsequent parameters of the generated procedure declaration.
Example
with function Decrypt
(Key_Update_Message : Key_Update_Message;
Sequence_Number : Sequence_Number;
Encrypted_Record : Opaque)
return
TLS_Inner_Plaintext
Channels
Channels provide a way for communicating with other systems using messages.
Syntax
channel_declaration ::=name":" "Channel" "with"channel_aspect{ ","channel_aspect} channel_aspect ::= "Readable" | "Writable"
Static Semantics
Channels can be readable or writable (non-exclusive).
Example
Data_Channel : Channel with Readable, Writable
Declarations
Variables and renamings can be globally declared (i.e. for the scope of the complete state machine).
Syntax
state_machine_declaration ::= (variable_declaration|renaming_declaration) ";"
Variable Declaration
A declared variable must have a type and can be optionally initialized using an expression.
Syntax
variable_declaration ::= variable_name":" type_qualified_name[ ":=" initialization_expression]
Example
Error_Sent : Boolean := False
Renaming Declaration
Syntax
renaming_declaration ::=name":" message_qualified_name"renames" message_variable_name"." field_name
Example
Client_Hello_Message : TLS_Handshake::Client_Hello renames Client_Hello_Handshake_Message.Payload
States
A state defines the to be executed actions and the transitions to subsequent states.
Variable declarations and renaming declarations in a state have a state-local scope, i.e., local declarations cannot be accessed from other states.
Syntax
state ::= "state"name[ "with"description_aspect] "is" { local_state_machine_declaration} "begin" {state_action} "transition" {conditional_transition}transition[ "exception"transition] "end"name";" description_aspect ::= "Desc" "=>"string
Static Semantics
An exception transition must be defined just in case any action might lead to a critical (potentially non-recoverable) error:
Insufficient memory for setting a field of a message
Insufficient memory for appending an element to a sequence or extending a sequence by another sequence
Exception transitions are currently also used for other cases.
A local declaration must not hide a global declaration.
The states where the actions include either Read or Write operations are referred to as IO states.
See the sections Read Attribute Statements and Write Attribute Statements for more information about those operations.
IO states cannot contain any declarations and they must not contain any other operations than channel IO.
Each channel and each message can be read or written at most once in a given IO state.
Dynamic Semantics
After entering a state the declarations and actions of the state are executed. If a non-recoverable error occurs, the execution is aborted and the state is changed based on the exception transition. When all action were executed successfully, the conditions of the transitions are checked in the given order. If a condition is fulfilled, the corresponding transition is taken to change the state. If no condition could be fulfilled or no conditional transitions were defined, the default transition is used.
Example
state A
with Desc => "rfc1149.txt+51:4-52:9"
is
Z : Boolean := Y;
M : TLV::Message;
begin
X'Read (M);
transition
goto B
with Desc => "rfc1149.txt+45:4-47:8"
if Z = True and G (F) = True
goto A
end A
State Transitions
State transitions define the conditions for the change to subsequent states. An arbitrary number of conditional transitions can be defined. The last transition in a state definition is the default transition, which does not contain any condition. The transition target must be either a state name or null, which represents the final state.
Syntax
conditional_transition ::=transition"if" conditional_ext_bool_expressiontransition ::= "goto" state_name[ "with"description_aspect]
Example
goto B
with Desc => "rfc1149.txt+45:4-47:8"
if Z = True and G (F) = True
State Actions
The state actions are executed after entering a state.
Syntax
state_action ::= (message_field_assignment|assignment|append|extend|reset|read|write) ";"
Assignment Statements
An assignment sets the value of variable.
Syntax
assignment ::= variable_name":="expression
Dynamic Semantics
An assignment always creates a copy of the original object.
Example
Error_Sent := True
Message Field Assignment Statements
A message field assignment sets the value of a message field.
Syntax
message_field_assignment ::= variable_name"." field_name":="expression
Dynamic Semantics
Message fields must be set in order. Trying to set a message field which is not a valid next field leads to an exception transition. All subsequent fields of the set message field are invalidated.
Example
Packet.Length := 42
Append Attribute Statements
An element is added to the end of a sequence using the Append attribute.
Syntax
append ::= sequence_name"'" "Append" "("expression")"
Dynamic Semantics
Appending an element to a sequence might lead to an exception transition.
Example
Parameter_Request_List'Append (DHCP::Domain_Name_Option)
Extend Attribute Statements
The Extend attributes adds a sequence of elements to the end of a sequence.
Syntax
extend ::= sequence_name"'" "Extend" "("expression")"
Dynamic Semantics
Extending a sequence might lead to an exception transition.
Example
Parameter_Request_List'Extend (Parameters)
Reset Attribute Statements
The state of a message or sequence can be cleared using the Reset attribute.
Syntax
reset ::=name"'" "Reset" [ "("named_argument_list")" ]
Static Semantics
When resetting a parameterized message, the intended values for the parameters of the message must be defined.
Dynamic Semantics
The existing state of a message or sequence is removed (and the corresponding buffer is cleared).
Example
Message'Reset
Read Attribute Statements
The read attribute statement is used to retrieve a message from a channel.
Syntax
read ::= channel_name"'" "Read" "("expression")"
Example
Data_Channel'Read (Message)
Write Attribute Statements
A message can be sent through a channel using a write attribute statement.
Syntax
write ::= channel_name"'" "Write" "("expression")"
Dynamic Semantics
Writing an invalid message leads to an exception transition.
Example
Data_Channel'Write (Message)
Expressions
Syntax
expression ::=math_expression|ext_bool_expression|message_aggregate|attribute_reference|selected|seq_expression|comprehension|conversion|call|case_expression|name
Message Aggregates
Syntax
message_aggregate ::= message_qualified_name"'" "("message_aggregate_association_list")" message_aggregate_association_list ::=named_argument_list| "null" "message"
Dynamic Semantics
An invalid condition during message creation leads to an exception transition. Insufficient memory during the message creation leads to an exception transition.
Example
TLS_Record::TLS_Record'(Tag => TLS_Record::Alert,
Legacy_Record_Version => TLS_Record::TLS_1_2,
Length => Alert_Message'Size / 8,
Fragment => Alert_Message'Opaque)
Null_Message'(null message)
Aggregates
An aggregate is a collection of elements.
Syntax
aggregate ::= "[" [number{ ","number} ] "]"
Example
[0, 1, 2]
[]
Attribute Expressions
Syntax
attribute_reference ::=expression"'"attribute_designatorattribute_designator ::= "Opaque" | "Head"
Static Semantics
The byte representation of a message can be retrieved using the Opaque attribute.
The Head attribute returns the first element of a sequence.
Dynamic Semantics
The use of the Opaque attribute on an invalid message or the use of the Head attribute on an empty sequence leads to an exception transition.
Example
Message'Opaque
Selected Expressions
The Selected expression is used to get a value of a message field.
Syntax
selected ::= message_expression"." field_name
Dynamic Semantics
Accesses to message fields that were detected as invalid during parsing lead to an exception transition.
Example
Ethernet_Frame.Payload
List Comprehensions
A list comprehension provides a way to create a new sequence based on an existing sequence.
Syntax
comprehension ::= "["
"for" name "in" iterable_expression
[ "if" condition_ext_bool_expression ]
"=>" selector_expression
"]"
Dynamic Semantics
An access to an invalid element in iterable expression leads to an exception transition.
Example
[for O in Offer.Options if O.Code = DHCP::DHCP_Message_Type_Option => O.DHCP_Message_Type]
Quantified Expressions
Quantified expressions enable reasoning about properties of sequences.
Syntax
quantified_expression ::= "for"quantifiername"in" iterable_expression"=>" predicate_ext_bool_expressionquantifier ::= "all" | "some"
Example
for all E in Server_Hello_Message.Extensions => E.Tag /= TLS_Handshake::ET_Supported_Versions
Calls
All functions which are declared in the state machine parameters can be called.
Syntax
call ::=qualified_name[ "(" argument_expression{ "," argument_expression} ")" ]
Example
Decrypt (Key_Update_Message, Sequence_Number, TLS_Record_Message.Encrypted_Record)
Conversions
An opaque field of a message can be converted to a message.
Syntax
conversion ::= message_qualified_name"(" message_expression"." field_name")"
Static Semantics
A conversion is only allowed if a refinement for the message field and the intended target type exists.
Dynamic Semantics
An invalid condition of a refinement leads to an exception transition.
Example
Key_Update_Message (Handshake_Control_Message.Data)
Case Expressions
A case expression selects one of several alternative dependent expressions for evaluation based on the value of a selecting expression.
Syntax
case_expression ::= "(" "case" selecting_expression
"is" case_expression_alternative { ","
case_expression_alternative } ")"
case_expression_alternative ::= "when" discrete_choice_list
"=>" dependent_expression
discrete_choice_list ::= discrete_choice
{ "|" discrete_choice }
discrete_choice ::= number | qualified_name
Static Semantics
The type of all the dependent expressions shall be compatible to the type of the case expression. Each value of the type of the selecting expression shall be covered by a discrete choice. Two distinct discrete choices of a case expression shall not cover the same value.
Example
(case Value is
when T::V1 | T::V2 => 2,
when T::V3 => 4)
Packages
A package is used to structure a specification.
Syntax
package ::= "package"name"is" {basic_declaration} "end"name";" basic_declaration ::= (integer_type|enumeration_type|message_type|type_refinement|type_derivation|sequence_type|state_machine) ";"
Static Semantics
A package is a collection of types and state machines. By convention one protocol is specified in one package.
Example
package Ethernet is
type Address is unsigned 48;
type Type_Length is range 46 .. 2 ** 16 - 1 with Size => 16;
type TPID is range 16#8100# .. 16#8100# with Size => 16;
type TCI is unsigned 16;
type Ether_Type is
(ET_IPv4 => 16#0800#,
ET_ARP => 16#0806#,
ET_VLAN_Tag => 16#8100#,
ET_IPv6 => 16#86DD#,
ET_VLAN_Tag_Double => 16#9100#)
with Size => 16, Always_Valid;
type Frame is
message
Destination : Address;
Source : Address;
Type_Length_TPID : Type_Length
then TPID
with First => Type_Length_TPID'First
if Type_Length_TPID = 16#8100#
then Payload
with Size => Type_Length_TPID * 8
if Type_Length_TPID <= 1500
then Ether_Type
with First => Type_Length_TPID'First
if Type_Length_TPID >= 1536 and Type_Length_TPID /= 16#8100#;
TPID : TPID;
TCI : TCI;
Ether_Type : Ether_Type;
Payload : Opaque
then null
if Payload'Size / 8 >= 46 and Payload'Size / 8 <= 1500;
end message;
generic
Input : Channel with Readable;
Output : Channel with Writable;
machine Validator is
Frame : Ethernet::Frame;
begin
state Validate
is
begin
Input'Read (Frame);
transition
goto Forward
if Frame'Valid
goto Validate
exception
goto null
end Validate;
state Forward
is
begin
Output'Write (Frame);
transition
goto Validate
end Forward;
end Validator;
end Ethernet;
Context Clauses
The context clause is used to specify the relation to other packages and consists of a list of with clauses.
Syntax
context ::= { "with" package_name ";" }
Static Semantics
For each package referenced in a file, a corresponding with clause has to be added to the beginning of the file.
Example
with Ethernet;
with IPv4;
Files
A RecordFlux specification file is recognized by the file extension .rflx.
Each specification file contains exactly one package.
The file name must match the package name in lower case characters.
Syntax
file ::=contextpackage
Example
File: in_ethernet.rflx.
with Ethernet;
with IPv4;
package In_Ethernet is
for Ethernet::Frame use (Payload => IPv4::Packet)
if Ether_Type = Ethernet::ET_IPv4;
end In_Ethernet;
Comments
Comments are started with two dashes (--) and extend to the end of the line. Block comments do not exist in RecordFlux.