Chapter 4. The class File Format

Table of Contents

4.1. The ClassFile Structure
4.2. Names
4.2.1. Binary Class and Interface Names
4.2.2. Unqualified Names
4.2.3. Module and Package Names
4.3. Descriptors
4.3.1. Grammar Notation
4.3.2. Field Descriptors
4.3.3. Method Descriptors
4.4. The Constant Pool
4.4.1. The CONSTANT_Class_infoStructure
4.4.2. The CONSTANT_Fieldref_infoCONSTANT_Methodref_info, and CONSTANT_InterfaceMethodref_infoStructures
4.4.3. The CONSTANT_String_infoStructure
4.4.4. The CONSTANT_Integer_info and CONSTANT_Float_info Structures
4.4.5. The CONSTANT_Long_info and CONSTANT_Double_info Structures
4.4.6. The CONSTANT_NameAndType_infoStructure
4.4.7. The CONSTANT_Utf8_info Structure
4.4.8. The CONSTANT_MethodHandle_info Structure
4.4.9. The CONSTANT_MethodType_infoStructure
4.4.10. The CONSTANT_InvokeDynamic_infoStructure
4.4.11. The CONSTANT_Module_infoStructure
4.4.12. The CONSTANT_Package_infoStructure
4.5. Fields
4.6. Methods
4.7. Attributes
4.7.1. Defining and Naming New Attributes
4.7.2. The ConstantValue Attribute
4.7.3. The Code Attribute
4.7.4. The StackMapTable Attribute
4.7.5. The Exceptions Attribute
4.7.6. The InnerClasses Attribute
4.7.7. The EnclosingMethod Attribute
4.7.8. The Synthetic Attribute
4.7.9. The Signature Attribute
4.7.9.1. Signatures
4.7.10. The SourceFile Attribute
4.7.11. The SourceDebugExtensionAttribute
4.7.12. The LineNumberTable Attribute
4.7.13. The LocalVariableTableAttribute
4.7.14. The LocalVariableTypeTableAttribute
4.7.15. The Deprecated Attribute
4.7.16. The RuntimeVisibleAnnotations Attribute
4.7.16.1. The element_value structure
4.7.17. The RuntimeInvisibleAnnotationsAttribute
4.7.18. The RuntimeVisibleParameterAnnotationsAttribute
4.7.19. The RuntimeInvisibleParameterAnnotationsAttribute
4.7.20. The RuntimeVisibleTypeAnnotationsAttribute
4.7.20.1. The target_info union
4.7.20.2. The type_path structure
4.7.21. The RuntimeInvisibleTypeAnnotationsAttribute
4.7.22. The AnnotationDefault Attribute
4.7.23. The BootstrapMethods Attribute
4.7.24. The MethodParameters Attribute
4.7.25. The Module Attribute
4.7.26. The ModulePackages Attribute
4.7.27. The ModuleMainClass Attribute
4.8. Format Checking
4.9. Constraints on Java Virtual Machine Code
4.9.1. Static Constraints
4.9.2. Structural Constraints
4.10. Verification of class Files
4.10.1. Verification by Type Checking
4.10.1.1. Accessors for Java Virtual Machine Artifacts
4.10.1.2. Verification Type System
4.10.1.3. Instruction Representation
4.10.1.4. Stack Map Frames and Type Transitions
4.10.1.5. Type Checking Abstract and Native Methods
4.10.1.6. Type Checking Methods with Code
4.10.1.7. Type Checking Load and Store Instructions
4.10.1.8. Type Checking for protectedMembers
4.10.1.9. Type Checking Instructions
aaload
aastore
aconst_null
aloadaload_<n>
anewarray
areturn
arraylength
astoreastore_<n>
athrow
baload
bastore
bipush
caload
castore
checkcast
d2fd2id2l
dadd
daload
dastore
dcmp<op>
dconst_<d>
ddiv
dloaddload_<n>
dmul
dneg
drem
dreturn
dstoredstore_<n>
dsub
dup
dup_x1
dup_x2
dup2
dup2_x1
dup2_x2
f2df2if2l
fadd
faload
fastore
fcmp<op>
fconst_<f>
fdiv
floadfload_<n>
fmul
fneg
frem
freturn
fstorefstore_<n>
fsub
getfield
getstatic
gotogoto_w
i2bi2ci2di2fi2li2s
iadd
iaload
iand
iastore
iconst_<i>
idiv
if_acmp<cond>
if_icmp<cond>
if<cond>
ifnonnullifnull
iinc
iloadiload_<n>
imul
ineg
instanceof
invokedynamic
invokeinterface
invokespecial
invokestatic
invokevirtual
iorirem
ireturn
ishlishriushr
istoreistore_<n>
isubixor
l2dl2fl2i
ladd
laload
land
lastore
lcmp
lconst_<l>
ldcldc_wldc2_w
ldiv
lloadlload_<n>
lmul
lneg
lookupswitch
lorlrem
lreturn
lshllshrlushr
lstorelstore_<n>
lsublxor
monitorentermonitorexit
multianewarray
new
newarray
nop
poppop2
putfield
putstatic
return
saload
sastore
sipush
swap
tableswitch
wide
4.10.2. Verification by Type Inference
4.10.2.1. The Process of Verification by Type Inference
4.10.2.2. The Bytecode Verifier
4.10.2.3. Values of Types long and double
4.10.2.4. Instance Initialization Methods and Newly Created Objects
4.10.2.5. Exceptions and finally
4.11. Limitations of the Java Virtual Machine

This chapter describes the class file format of the Java Virtual Machine. Each class file contains the definition of a single class, interface, or module. Although a class, interface, or module need not have an external representation literally contained in a file (for instance, because the class is generated by a class loader), we will colloquially refer to any valid representation of a class, interface, or module as being in the class file format.

class file consists of a stream of 8-bit bytes. 16-bit and 32-bit quantities are constructed by reading in two and four consecutive 8-bit bytes, respectively. Multibyte data items are always stored in big-endian order, where the high bytes come first. This chapter defines the data types u1u2, and u4 to represent an unsigned one-, two-, or four-byte quantity, respectively.

In the Java SE Platform API, the class file format is supported by interfaces java.io.DataInput and java.io.DataOutput and classes such as java.io.DataInputStream and java.io.DataOutputStream. For example, values of the types u1u2, and u4 may be read by methods such as readUnsignedBytereadUnsignedShort, and readInt of the interface java.io.DataInput.

This chapter presents the class file format using pseudostructures written in a C-like structure notation. To avoid confusion with the fields of classes and class instances, etc., the contents of the structures describing the class file format are referred to as items. Successive items are stored in the class file sequentially, without padding or alignment.

Tables, consisting of zero or more variable-sized items, are used in several class file structures. Although we use C-like array syntax to refer to table items, the fact that tables are streams of varying-sized structures means that it is not possible to translate a table index directly to a byte offset into the table.

Where we refer to a data structure as an array, it consists of zero or more contiguous fixed-sized items and can be indexed like an array.

Reference to an ASCII character in this chapter should be interpreted to mean the Unicode code point corresponding to the ASCII character.

4.1. The ClassFile Structure

class file consists of a single ClassFile structure:

ClassFile {
u4 magic;
u2 minor_version;
u2 major_version;
u2 constant_pool_count;
cp_info constant_pool[constant_pool_count-1];
u2 access_flags;
u2 this_class;
u2 super_class;
u2 interfaces_count;
u2 interfaces[interfaces_count];
u2 fields_count;
field_info fields[fields_count];
u2 methods_count;
method_info methods[methods_count];
u2 attributes_count;
attribute_info attributes[attributes_count];
}

The items in the ClassFile structure are as follows:

magic

The magic item supplies the magic number identifying the class file format; it has the value 0xCAFEBABE.

minor_version, major_version

The values of the minor_version and major_version items are the minor and major version numbers of this class file. Together, a major and a minor version number determine the version of the class file format. If a class file has major version number M and minor version number m, we denote the version of its class file format as M.m. Thus, class file format versions may be ordered lexicographically, for example, 1.5 < 2.0 < 2.1.

A Java Virtual Machine implementation can support a class file format of version v if and only if v lies in some contiguous range Mi.0 ≤ v ≤ Mj.m. The release level of the Java SE Platform to which a Java Virtual Machine implementation conforms is responsible for determining the range.

Oracle's Java Virtual Machine implementation in JDK release 1.0.2 supports class file format versions 45.0 through 45.3 inclusive. JDK releases 1.1.* support class file format versions in the range 45.0 through 45.65535 inclusive. For k ≥ 2, JDK release 1.k supports class file format versions in the range 45.0 through 44+k.0 inclusive.

constant_pool_count

The value of the constant_pool_count item is equal to the number of entries in the constant_pool table plus one. A constant_pool index is considered valid if it is greater than zero and less than constant_pool_count, with the exception for constants of type long and double noted in §4.4.5.

constant_pool[]

The constant_pool is a table of structures (§4.4) representing various string constants, class and interface names, field names, and other constants that are referred to within the ClassFile structure and its substructures. The format of each constant_pool table entry is indicated by its first "tag" byte.

The constant_pool table is indexed from 1 to constant_pool_count - 1.

access_flags

The value of the access_flags item is a mask of flags used to denote access permissions to and properties of this class or interface. The interpretation of each flag, when set, is specified in Table 4.1-A.

Table 4.1-A. Class access and property modifiers

Flag Name Value Interpretation
ACC_PUBLIC 0x0001 Declared public; may be accessed from outside its package.
ACC_FINAL 0x0010 Declared final; no subclasses allowed.
ACC_SUPER 0x0020 Treat superclass methods specially when invoked by the invokespecial instruction.
ACC_INTERFACE 0x0200 Is an interface, not a class.
ACC_ABSTRACT 0x0400 Declared abstract; must not be instantiated.
ACC_SYNTHETIC 0x1000 Declared synthetic; not present in the source code.
ACC_ANNOTATION 0x2000 Declared as an annotation type.
ACC_ENUM 0x4000 Declared as an enum type.

ACC_MODULE

0x8000 Is a module, not a class or interface.

The ACC_MODULE flag indicates that this class file defines a module, not a class or interface. If the ACC_MODULE flag is set, then special rules apply to the class file which are given at the end of this section. If the ACC_MODULE flag is not set, then the rules immediately below the current paragraph apply to the class file.

An interface is distinguished by the ACC_INTERFACE flag being set. If the ACC_INTERFACE flag is not set, this class file defines a class, not an interface or module.

If the ACC_INTERFACE flag is set, the ACC_ABSTRACT flag must also be set, and the ACC_FINALACC_SUPERACC_ENUM, and ACC_MODULE flags set must not be set.

If the ACC_INTERFACE flag is not set, any of the other flags in Table 4.1-A may be set except ACC_ANNOTATION and ACC_MODULE. However, such a class file must not have both its ACC_FINAL and ACC_ABSTRACT flags set (JLS §8.1.1.2).

The ACC_SUPER flag indicates which of two alternative semantics is to be expressed by the invokespecial instruction (§invokespecial) if it appears in this class or interface. Compilers to the instruction set of the Java Virtual Machine should set the ACC_SUPER flag. In Java SE 8 and above, the Java Virtual Machine considers the ACC_SUPER flag to be set in every class file, regardless of the actual value of the flag in the class file and the version of the class file.

The ACC_SUPER flag exists for backward compatibility with code compiled by older compilers for the Java programming language. In JDK releases prior to 1.0.2, the compiler generated access_flags in which the flag now representing ACC_SUPER had no assigned meaning, and Oracle's Java Virtual Machine implementation ignored the flag if it was set.

The ACC_SYNTHETIC flag indicates that this class or interface was generated by a compiler and does not appear in source code.

An annotation type (JLS §9.6) must have its ACC_ANNOTATION flag set. If the ACC_ANNOTATION flag is set, the ACC_INTERFACE flag must also be set.

The ACC_ENUM flag indicates that this class or its superclass is declared as an enumerated type (JLS §8.9).

All bits of the access_flags item not assigned in Table 4.1-A are reserved for future use. They should be set to zero in generated class files and should be ignored by Java Virtual Machine implementations.

this_class

The value of the this_class item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Class_info structure (§4.4.1) representing the class or interface defined by this class file.

super_class

For a class, the value of the super_class item either must be zero or must be a valid index into the constant_pool table. If the value of the super_class item is nonzero, theconstant_pool entry at that index must be a CONSTANT_Class_info structure representing the direct superclass of the class defined by this class file. Neither the direct superclass nor any of its superclasses may have the ACC_FINAL flag set in the access_flags item of its ClassFile structure.

If the value of the super_class item is zero, then this class file must represent the class Object, the only class or interface without a direct superclass.

For an interface, the value of the super_class item must always be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Class_info structure representing the class Object.

interfaces_count

The value of the interfaces_count item gives the number of direct superinterfaces of this class or interface type.

interfaces[]

Each value in the interfaces array must be a valid index into the constant_pool table. The constant_pool entry at each value of interfaces[i], where 0 ≤ i < interfaces_count, must be a CONSTANT_Class_info structure representing an interface that is a direct superinterface of this class or interface type, in the left-to-right order given in the source for the type.

fields_count

The value of the fields_count item gives the number of field_info structures in the fields table. The field_info structures represent all fields, both class variables and instance variables, declared by this class or interface type.

fields[]

Each value in the fields table must be a field_info structure (§4.5) giving a complete description of a field in this class or interface. The fields table includes only those fields that are declared by this class or interface. It does not include items representing fields that are inherited from superclasses or superinterfaces.

methods_count

The value of the methods_count item gives the number of method_info structures in the methods table.

methods[]

Each value in the methods table must be a method_info structure (§4.6) giving a complete description of a method in this class or interface. If neither of the ACC_NATIVE and ACC_ABSTRACT flags are set in the access_flags item of a method_info structure, the Java Virtual Machine instructions implementing the method are also supplied.

The method_info structures represent all methods declared by this class or interface type, including instance methods, class methods, instance initialization methods (§2.9.1), and any class or interface initialization method (§2.9.2). The methods table does not include items representing methods that are inherited from superclasses or superinterfaces.

attributes_count

The value of the attributes_count item gives the number of attributes in the attributes table of this class.

attributes[]

Each value of the attributes table must be an attribute_info structure (§4.7).

The attributes defined by this specification as appearing in the attributes table of a ClassFile structure are listed in Table 4.7-C.

The rules concerning attributes defined to appear in the attributes table of a ClassFile structure are given in §4.7.

The rules concerning non-predefined attributes in the attributes table of a ClassFile structure are given in §4.7.1.

If the ACC_MODULE flag is set in the access_flags item, then no other flag in the access_flags item may be set, and the following rules apply to the rest of the ClassFile structure:

  • major_versionminor_version: ≥ 53.0 (i.e. Java SE 9 and above)

  • this_classmodule-info

  • super_classinterfaces_countfields_countmethods_count: zero

  • attributes: One Module attribute must be present. Except for ModuleModulePackagesModuleMainClassInnerClassesSourceFileSourceDebugExtensionRuntimeVisibleAnnotations, and RuntimeInvisibleAnnotations, none of the pre-defined attributes (§4.7) may appear.

4.2. Names

4.2.1. Binary Class and Interface Names

Class and interface names that appear in class file structures are always represented in a fully qualified form known as binary names (JLS §13.1). Such names are always represented as CONSTANT_Utf8_info structures (§4.4.7) and thus may be drawn, where not further constrained, from the entire Unicode codespace. Class and interface names are referenced from those CONSTANT_NameAndType_info structures (§4.4.6) which have such names as part of their descriptor (§4.3), and from all CONSTANT_Class_info structures (§4.4.1).

For historical reasons, the syntax of binary names that appear in class file structures differs from the syntax of binary names documented in JLS §13.1. In this internal form, the ASCII periods (.) that normally separate the identifiers which make up the binary name are replaced by ASCII forward slashes (/). The identifiers themselves must be unqualified names (§4.2.2).

For example, the normal binary name of class Thread is java.lang.Thread. In the internal form used in descriptors in the class file format, a reference to the name of class Thread is implemented using a CONSTANT_Utf8_info structure representing the string java/lang/Thread.

4.2.2. Unqualified Names

Names of methods, fields, local variables, and formal parameters are stored as unqualified names. An unqualified name must contain at least one Unicode code point and must not contain any of the ASCII characters . ; [ / (that is, period or semicolon or left square bracket or forward slash).

Method names are further constrained so that, with the exception of the special method names <init> and <clinit> (§2.9), they must not contain the ASCII characters < or > (that is, left angle bracket or right angle bracket).

Note that a field name or interface method name may be <init> or <clinit>, but no method invocation instruction may reference <clinit> and only the invokespecial instruction (§invokespecial) may reference <init>.

4.2.3. Module and Package Names

Module names referenced from the Module attribute are stored in CONSTANT_Module_info structures in the constant pool (§4.4.11). A CONSTANT_Module_info structure wraps a CONSTANT_Utf8_info structure that denotes the module name. Module names are not encoded in "internal form" like class and interface names, that is, the ASCII periods (.) that separate the identifiers in a module name are not replaced by ASCII forward slashes (/).

Module names may be drawn from the entire Unicode codespace, subject to the following constraints:

  • A module name must not contain any code point in the range '\u0000' to '\u001F' inclusive.

  • The ASCII backslash (\) is reserved for use as an escape character in module names. It must not appear in a module name unless it is followed by an ASCII backslash, an ASCII colon (:), or an ASCII at-sign (@). The ASCII character sequence \\ may be used to encode a backslash in a module name.

  • The ASCII colon (:) and at-sign (@) are reserved for future use in module names. They must not appear in module names unless they are escaped. The ASCII character sequences \: and \@may be used to encode a colon and an at-sign in a module name.

Package names referenced from the Module attribute are stored in CONSTANT_Package_info structures in the constant pool (§4.4.12). A CONSTANT_Package_info structure wraps a CONSTANT_Utf8_info structure that represents a package name encoded in internal form.

4.3. Descriptors

descriptor is a string representing the type of a field or method. Descriptors are represented in the class file format using modified UTF-8 strings (§4.4.7) and thus may be drawn, where not further constrained, from the entire Unicode codespace.

4.3.1. Grammar Notation

Descriptors are specified using a grammar. The grammar is a set of productions that describe how sequences of characters can form syntactically correct descriptors of various kinds. Terminal symbols of the grammar are shown in fixed width font. Nonterminal symbols are shown in italic type. The definition of a nonterminal is introduced by the name of the nonterminal being defined, followed by a colon. One or more alternative definitions for the nonterminal then follow on succeeding lines.

The syntax {x} on the right-hand side of a production denotes zero or more occurrences of x.

The phrase (one of) on the right-hand side of a production signifies that each of the terminal symbols on the following line or lines is an alternative definition.

4.3.2. Field Descriptors

field descriptor represents the type of a class, instance, or local variable.

FieldDescriptor:
BaseType:
(one of) 
B C D F I J S Z
ObjectType:
L ClassName ;
ArrayType:
ComponentType:

The characters of BaseType, the L and ; of ObjectType, and the [ of ArrayType are all ASCII characters.

ClassName represents a binary class or interface name encoded in internal form (§4.2.1).

The interpretation of field descriptors as types is shown in Table 4.3-A.

A field descriptor representing an array type is valid only if it represents a type with 255 or fewer dimensions.

Table 4.3-A. Interpretation of field descriptors

FieldType term Type Interpretation
B byte signed byte
C char Unicode character code point in the Basic Multilingual Plane, encoded with UTF-16
D double double-precision floating-point value
F float single-precision floating-point value
I int integer
J long long integer
L ClassName ; reference an instance of class ClassName
S short signed short
Z boolean true or false
[ reference one array dimension

The field descriptor of an instance variable of type int is simply I.

The field descriptor of an instance variable of type Object is Ljava/lang/Object;. Note that the internal form of the binary name for class Object is used.

The field descriptor of an instance variable of the multidimensional array type double[][][] is [[[D.

4.3.3. Method Descriptors

method descriptor contains zero or more parameter descriptors, representing the types of parameters that the method takes, and a return descriptor, representing the type of the value (if any) that the method returns.

MethodDescriptor:
ParameterDescriptor:
ReturnDescriptor:
VoidDescriptor:
V

The character V indicates that the method returns no value (its result is void).

The method descriptor for the method:

Object m(int i, double d, Thread t) {...}

is:

(IDLjava/lang/Thread;)Ljava/lang/Object;

Note that the internal forms of the binary names of Thread and Object are used.

A method descriptor is valid only if it represents method parameters with a total length of 255 or less, where that length includes the contribution for this in the case of instance or interface method invocations. The total length is calculated by summing the contributions of the individual parameters, where a parameter of type long or double contributes two units to the length and a parameter of any other type contributes one unit.

A method descriptor is the same whether the method it describes is a class method or an instance method. Although an instance method is passed this, a reference to the object on which the method is being invoked, in addition to its intended arguments, that fact is not reflected in the method descriptor. The reference to this is passed implicitly by the Java Virtual Machine instructions which invoke instance methods (§2.6.1§4.11).

4.4. The Constant Pool

Java Virtual Machine instructions do not rely on the run-time layout of classes, interfaces, class instances, or arrays. Instead, instructions refer to symbolic information in the constant_pooltable.

All constant_pool table entries have the following general format:

cp_info {
u1 tag;
u1 info[];
}

Each entry in the constant_pool table must begin with a 1-byte tag indicating the type of constant denoted by the entry. The constant types and their corresponding tag values are listed inTable 4.4-A. Each constant type is accompanied by the first version of the class file format in which it was defined, and the corresponding version of the Java SE Platform. (For old class file versions, the JDK release is used instead of the Java SE Platform version.)

Each tag byte must be followed by two or more bytes giving information about the specific constant. The format of the additional information depends on the tag byte, that is, the content of the info array varies with the value of tag.

Table 4.4-A. Constant pool tags

Constant Type Value class file Java SE
CONSTANT_Class 7 45.3 1.0.2
CONSTANT_Fieldref 9 45.3 1.0.2
CONSTANT_Methodref 10 45.3 1.0.2
CONSTANT_InterfaceMethodref 11 45.3 1.0.2
CONSTANT_String 8 45.3 1.0.2
CONSTANT_Integer 3 45.3 1.0.2
CONSTANT_Float 4 45.3 1.0.2
CONSTANT_Long 5 45.3 1.0.2
CONSTANT_Double 6 45.3 1.0.2
CONSTANT_NameAndType 12 45.3 1.0.2
CONSTANT_Utf8 1 45.3 1.0.2
CONSTANT_MethodHandle 15 51.0 7
CONSTANT_MethodType 16 51.0 7
CONSTANT_InvokeDynamic 18 51.0 7

CONSTANT_Module

19 53.0 9

CONSTANT_Package

20 53.0 9

4.4.1. The CONSTANT_Class_info Structure

The CONSTANT_Class_info structure is used to represent a class or an interface:

CONSTANT_Class_info {
u1 tag;
u2 name_index;
}

The items of the CONSTANT_Class_info structure are as follows:

tag

The tag item has the value CONSTANT_Class (7).

name_index

The value of the name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure (§4.4.7) representing a valid binary class or interface name encoded in internal form (§4.2.1).

Because arrays are objects, the opcodes anewarray and multianewarray - but not the opcode new - can reference array "classes" via CONSTANT_Class_info structures in the constant_pooltable. For such array classes, the name of the class is the descriptor of the array type (§4.3.2).

For example, the class name representing the two-dimensional array type int[][] is [[I, while the class name representing the type Thread[] is [Ljava/lang/Thread;.

An array type descriptor is valid only if it represents 255 or fewer dimensions.

4.4.2. The CONSTANT_Fieldref_infoCONSTANT_Methodref_info, and CONSTANT_InterfaceMethodref_info Structures

Fields, methods, and interface methods are represented by similar structures:

CONSTANT_Fieldref_info {
u1 tag;
u2 class_index;
u2 name_and_type_index;
} CONSTANT_Methodref_info {
u1 tag;
u2 class_index;
u2 name_and_type_index;
} CONSTANT_InterfaceMethodref_info {
u1 tag;
u2 class_index;
u2 name_and_type_index;
}

The items of these structures are as follows:

tag

The tag item of a CONSTANT_Fieldref_info structure has the value CONSTANT_Fieldref (9).

The tag item of a CONSTANT_Methodref_info structure has the value CONSTANT_Methodref (10).

The tag item of a CONSTANT_InterfaceMethodref_info structure has the value CONSTANT_InterfaceMethodref (11).

class_index

The value of the class_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Class_info structure (§4.4.1) representing a class or interface type that has the field or method as a member.

The class_index item of a CONSTANT_Methodref_info structure must be a class type, not an interface type.

The class_index item of a CONSTANT_InterfaceMethodref_info structure must be an interface type.

The class_index item of a CONSTANT_Fieldref_info structure may be either a class type or an interface type.

name_and_type_index

The value of the name_and_type_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_NameAndType_infostructure (§4.4.6). This constant_pool entry indicates the name and descriptor of the field or method.

In a CONSTANT_Fieldref_info, the indicated descriptor must be a field descriptor (§4.3.2). Otherwise, the indicated descriptor must be a method descriptor (§4.3.3).

If the name of the method of a CONSTANT_Methodref_info structure begins with a '<' ('\u003c'), then the name must be the special name <init>, representing an instance initialization method (§2.9.1). The return type of such a method must be void.

4.4.3. The CONSTANT_String_info Structure

The CONSTANT_String_info structure is used to represent constant objects of the type String:

CONSTANT_String_info {
u1 tag;
u2 string_index;
}

The items of the CONSTANT_String_info structure are as follows:

tag

The tag item has the value CONSTANT_String (8).

string_index

The value of the string_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure (§4.4.7) representing the sequence of Unicode code points to which the String object is to be initialized.

4.4.4. The CONSTANT_Integer_info and CONSTANT_Float_info Structures

The CONSTANT_Integer_info and CONSTANT_Float_info structures represent 4-byte numeric (int and float) constants:

CONSTANT_Integer_info {
u1 tag;
u4 bytes;
} CONSTANT_Float_info {
u1 tag;
u4 bytes;
}

The items of these structures are as follows:

tag

The tag item of the CONSTANT_Integer_info structure has the value CONSTANT_Integer (3).

The tag item of the CONSTANT_Float_info structure has the value CONSTANT_Float (4).

bytes

The bytes item of the CONSTANT_Integer_info structure represents the value of the int constant. The bytes of the value are stored in big-endian (high byte first) order.

The bytes item of the CONSTANT_Float_info structure represents the value of the float constant in IEEE 754 floating-point single format (§2.3.2). The bytes of the single format representation are stored in big-endian (high byte first) order.

The value represented by the CONSTANT_Float_info structure is determined as follows. The bytes of the value are first converted into an int constant bits. Then:

  • If bits is 0x7f800000, the float value will be positive infinity.

  • If bits is 0xff800000, the float value will be negative infinity.

  • If bits is in the range 0x7f800001 through 0x7fffffff or in the range 0xff800001 through 0xffffffff, the float value will be NaN.

  • In all other cases, let se, and m be three values that might be computed from bits:

    int s = ((bits >> 31) == 0) ? 1 : -1;
    int e = ((bits >> 23) & 0xff);
    int m = (e == 0) ?
    (bits & 0x7fffff) << 1 :
    (bits & 0x7fffff) | 0x800000;

Then the float value equals the result of the mathematical expression s · m · 2e-150.

4.4.5. The CONSTANT_Long_info and CONSTANT_Double_info Structures

The CONSTANT_Long_info and CONSTANT_Double_info represent 8-byte numeric (long and double) constants:

CONSTANT_Long_info {
u1 tag;
u4 high_bytes;
u4 low_bytes;
} CONSTANT_Double_info {
u1 tag;
u4 high_bytes;
u4 low_bytes;
}

All 8-byte constants take up two entries in the constant_pool table of the class file. If a CONSTANT_Long_info or CONSTANT_Double_info structure is the item in the constant_pooltable at index n, then the next usable item in the pool is located at index n+2. The constant_pool index n+1 must be valid but is considered unusable.

In retrospect, making 8-byte constants take two constant pool entries was a poor choice.

The items of these structures are as follows:

tag

The tag item of the CONSTANT_Long_info structure has the value CONSTANT_Long (5).

The tag item of the CONSTANT_Double_info structure has the value CONSTANT_Double (6).

high_bytes, low_bytes

The unsigned high_bytes and low_bytes items of the CONSTANT_Long_info structure together represent the value of the long constant

((long) high_bytes << 32) + low_bytes
      

where the bytes of each of high_bytes and low_bytes are stored in big-endian (high byte first) order.

The high_bytes and low_bytes items of the CONSTANT_Double_info structure together represent the double value in IEEE 754 floating-point double format (§2.3.2). The bytes of each item are stored in big-endian (high byte first) order.

The value represented by the CONSTANT_Double_info structure is determined as follows. The high_bytes and low_bytes items are converted into the long constant bits, which is equal to

((long) high_bytes << 32) + low_bytes
      

Then:

  • If bits is 0x7ff0000000000000L, the double value will be positive infinity.

  • If bits is 0xfff0000000000000L, the double value will be negative infinity.

  • If bits is in the range 0x7ff0000000000001L through 0x7fffffffffffffffL or in the range 0xfff0000000000001L through 0xffffffffffffffffL, the double value will be NaN.

  • In all other cases, let se, and m be three values that might be computed from bits:

    int s = ((bits >> 63) == 0) ? 1 : -1;
    int e = (int)((bits >> 52) & 0x7ffL);
    long m = (e == 0) ?
    (bits & 0xfffffffffffffL) << 1 :
    (bits & 0xfffffffffffffL) | 0x10000000000000L;

Then the floating-point value equals the double value of the mathematical expression s · m · 2e-1075.

4.4.6. The CONSTANT_NameAndType_info Structure

The CONSTANT_NameAndType_info structure is used to represent a field or method, without indicating which class or interface type it belongs to:

CONSTANT_NameAndType_info {
u1 tag;
u2 name_index;
u2 descriptor_index;
}

The items of the CONSTANT_NameAndType_info structure are as follows:

tag

The tag item has the value CONSTANT_NameAndType (12).

name_index

The value of the name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure (§4.4.7) representing either a valid unqualified name denoting a field or method (§4.2.2), or the special method name <init> (§2.9.1).

descriptor_index

The value of the descriptor_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure (§4.4.7) representing a valid field descriptor or method descriptor (§4.3.2§4.3.3).

4.4.7. The CONSTANT_Utf8_info Structure

The CONSTANT_Utf8_info structure is used to represent constant string values:

CONSTANT_Utf8_info {
u1 tag;
u2 length;
u1 bytes[length];
}

The items of the CONSTANT_Utf8_info structure are as follows:

tag

The tag item has the value CONSTANT_Utf8 (1).

length

The value of the length item gives the number of bytes in the bytes array (not the length of the resulting string).

bytes[]

The bytes array contains the bytes of the string.

No byte may have the value (byte)0.

No byte may lie in the range (byte)0xf0 to (byte)0xff.

String content is encoded in modified UTF-8. Modified UTF-8 strings are encoded so that code point sequences that contain only non-null ASCII characters can be represented using only 1 byte per code point, but all code points in the Unicode codespace can be represented. Modified UTF-8 strings are not null-terminated. The encoding is as follows:

  • Code points in the range '\u0001' to '\u007F' are represented by a single byte:

    Table 4.4. 

    0 bits 6-0

    The 7 bits of data in the byte give the value of the code point represented.

  • The null code point ('\u0000') and code points in the range '\u0080' to '\u07FF' are represented by a pair of bytes x and y :

    Table 4.5. 

    x:

    Table 4.6. 

    1 1 0 bits 10-6
    y:

    Table 4.7. 

    1 0 bits 5-0

    The two bytes represent the code point with the value:

    ((x & 0x1f) << 6) + (y & 0x3f)
        
  • Code points in the range '\u0800' to '\uFFFF' are represented by 3 bytes xy, and z :

    Table 4.8. 

    x:

    Table 4.9. 

    1 1 1 0 bits 15-12
    y:

    Table 4.10. 

    1 0 bits 11-6
    z:

    Table 4.11. 

    1 0 bits 5-0

    The three bytes represent the code point with the value:

    ((x & 0xf) << 12) + ((y & 0x3f) << 6) + (z & 0x3f)
        
  • Characters with code points above U+FFFF (so-called supplementary characters) are represented by separately encoding the two surrogate code units of their UTF-16 representation. Each of the surrogate code units is represented by three bytes. This means supplementary characters are represented by six bytes, uvwxy, and z :

    Table 4.12. 

    u:

    Table 4.13. 

    1 1 1 0 1 1 0 1
    v:

    Table 4.14. 

    1 0 1 0 (bits 20-16)-1
    w:

    Table 4.15. 

    1 0 bits 15-10
    x:

    Table 4.16. 

    1 1 1 0 1 1 0 1
    y:

    Table 4.17. 

    1 0 1 1 bits 9-6
    z:

    Table 4.18. 

    1 0 bits 5-0

    The six bytes represent the code point with the value:

    0x10000 + ((v & 0x0f) << 16) + ((w & 0x3f) << 10) +
    ((y & 0x0f) << 6) + (z & 0x3f)

The bytes of multibyte characters are stored in the class file in big-endian (high byte first) order.

There are two differences between this format and the "standard" UTF-8 format. First, the null character (char)0 is encoded using the 2-byte format rather than the 1-byte format, so that modified UTF-8 strings never have embedded nulls. Second, only the 1-byte, 2-byte, and 3-byte formats of standard UTF-8 are used. The Java Virtual Machine does not recognize the four-byte format of standard UTF-8; it uses its own two-times-three-byte format instead.

For more information regarding the standard UTF-8 format, see Section 3.9 Unicode Encoding Forms of The Unicode Standard, Version 8.0.0.

4.4.8. The CONSTANT_MethodHandle_info Structure

The CONSTANT_MethodHandle_info structure is used to represent a method handle:

CONSTANT_MethodHandle_info {
u1 tag;
u1 reference_kind;
u2 reference_index;
}

The items of the CONSTANT_MethodHandle_info structure are the following:

tag

The tag item has the value CONSTANT_MethodHandle (15).

reference_kind

The value of the reference_kind item must be in the range 1 to 9. The value denotes the kind of this method handle, which characterizes its bytecode behavior (§5.4.3.5).

reference_index

The value of the reference_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be as follows:

  • If the value of the reference_kind item is 1 (REF_getField), 2 (REF_getStatic), 3 (REF_putField), or 4 (REF_putStatic), then the constant_pool entry at that index must be a CONSTANT_Fieldref_info (§4.4.2) structure representing a field for which a method handle is to be created.

  • If the value of the reference_kind item is 5 (REF_invokeVirtual) or 8 (REF_newInvokeSpecial), then the constant_pool entry at that index must be aCONSTANT_Methodref_info structure (§4.4.2) representing a class's method or constructor (§2.9.1) for which a method handle is to be created.

  • If the value of the reference_kind item is 6 (REF_invokeStatic) or 7 (REF_invokeSpecial), then if the class file version number is less than 52.0, the constant_pool entry at that index must be a CONSTANT_Methodref_info structure representing a class's method for which a method handle is to be created; if the class file version number is 52.0 or above, the constant_pool entry at that index must be either a CONSTANT_Methodref_info structure or a CONSTANT_InterfaceMethodref_info structure (§4.4.2) representing a class's or interface's method for which a method handle is to be created.

  • If the value of the reference_kind item is 9 (REF_invokeInterface), then the constant_pool entry at that index must be a CONSTANT_InterfaceMethodref_info structure representing an interface's method for which a method handle is to be created.

If the value of the reference_kind item is 5 (REF_invokeVirtual), 6 (REF_invokeStatic), 7 (REF_invokeSpecial), or 9 (REF_invokeInterface), the name of the method represented by a CONSTANT_Methodref_info structure or a CONSTANT_InterfaceMethodref_info structure must not be <init> or <clinit>.

If the value is 8 (REF_newInvokeSpecial), the name of the method represented by a CONSTANT_Methodref_info structure must be <init>.

4.4.9. The CONSTANT_MethodType_info Structure

The CONSTANT_MethodType_info structure is used to represent a method type:

CONSTANT_MethodType_info {
u1 tag;
u2 descriptor_index;
}

The items of the CONSTANT_MethodType_info structure are as follows:

tag

The tag item has the value CONSTANT_MethodType (16).

descriptor_index

The value of the descriptor_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure (§4.4.7) representing a method descriptor (§4.3.3).

4.4.10. The CONSTANT_InvokeDynamic_info Structure

The CONSTANT_InvokeDynamic_info structure is used by an invokedynamic instruction (§invokedynamic) to specify a bootstrap method, the dynamic invocation name, the argument and return types of the call, and optionally, a sequence of additional constants called static arguments to the bootstrap method.

CONSTANT_InvokeDynamic_info {
u1 tag;
u2 bootstrap_method_attr_index;
u2 name_and_type_index;
}

The items of the CONSTANT_InvokeDynamic_info structure are as follows:

tag

The tag item has the value CONSTANT_InvokeDynamic (18).

bootstrap_method_attr_index

The value of the bootstrap_method_attr_index item must be a valid index into the bootstrap_methods array of the bootstrap method table (§4.7.23) of this class file.

name_and_type_index

The value of the name_and_type_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_NameAndType_infostructure (§4.4.6) representing a method name and method descriptor (§4.3.3).

4.4.11. The CONSTANT_Module_info Structure

The CONSTANT_Module_info structure is used to represent a module:

CONSTANT_Module_info {
u1 tag;
u2 name_index;
}

The items of the CONSTANT_Module_info structure are as follows:

tag

The tag item has the value CONSTANT_Module (19).

name_index

The value of the name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure representing a valid module name (§4.2.3).

CONSTANT_Module_info structure is permitted only in the constant pool of a class file that declares a module, that is, a ClassFile structure where the access_flags item has the ACC_MODULE flag set. In all other class files, a CONSTANT_Module_info structure is illegal.

4.4.12. The CONSTANT_Package_info Structure

The CONSTANT_Package_info structure is used to represent a package exported or opened by a module:

CONSTANT_Package_info {
u1 tag;
u2 name_index;
}

The items of the CONSTANT_Package_info structure are as follows:

tag

The tag item has the value CONSTANT_Package (20).

name_index

The value of the name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure representing a valid package name encoded in internal form (§4.2.3).

CONSTANT_Package_info structure is permitted only in the constant pool of a class file that declares a module, that is, a ClassFile structure where the access_flags item has the ACC_MODULE flag set. In all other class files, a CONSTANT_Package_info structure is illegal.

4.5. Fields

Each field is described by a field_info structure.

No two fields in one class file may have the same name and descriptor (§4.3.2).

The structure has the following format:

field_info {
u2 access_flags;
u2 name_index;
u2 descriptor_index;
u2 attributes_count;
attribute_info attributes[attributes_count];
}

The items of the field_info structure are as follows:

access_flags

The value of the access_flags item is a mask of flags used to denote access permission to and properties of this field. The interpretation of each flag, when set, is specified in Table 4.5-A.

Table 4.5-A. Field access and property flags

Flag Name Value Interpretation
ACC_PUBLIC 0x0001 Declared public; may be accessed from outside its package.
ACC_PRIVATE 0x0002 Declared private; usable only within the defining class.
ACC_PROTECTED 0x0004 Declared protected; may be accessed within subclasses.
ACC_STATIC 0x0008 Declared static.
ACC_FINAL 0x0010 Declared final; never directly assigned to after object construction (JLS §17.5).
ACC_VOLATILE 0x0040 Declared volatile; cannot be cached.
ACC_TRANSIENT 0x0080 Declared transient; not written or read by a persistent object manager.
ACC_SYNTHETIC 0x1000 Declared synthetic; not present in the source code.
ACC_ENUM 0x4000 Declared as an element of an enum.

Fields of classes may set any of the flags in Table 4.5-A. However, each field of a class may have at most one of its ACC_PUBLICACC_PRIVATE, and ACC_PROTECTED flags set (JLS §8.3.1), and must not have both its ACC_FINAL and ACC_VOLATILE flags set (JLS §8.3.1.4).

Fields of interfaces must have their ACC_PUBLICACC_STATIC, and ACC_FINAL flags set; they may have their ACC_SYNTHETIC flag set and must not have any of the other flags in Table 4.5-A set (JLS §9.3).

The ACC_SYNTHETIC flag indicates that this field was generated by a compiler and does not appear in source code.

The ACC_ENUM flag indicates that this field is used to hold an element of an enumerated type.

All bits of the access_flags item not assigned in Table 4.5-A are reserved for future use. They should be set to zero in generated class files and should be ignored by Java Virtual Machine implementations.

name_index

The value of the name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure (§4.4.7) which represents a valid unqualified name denoting a field (§4.2.2).

descriptor_index

The value of the descriptor_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure (§4.4.7) which represents a valid field descriptor (§4.3.2).

attributes_count

The value of the attributes_count item indicates the number of additional attributes of this field.

attributes[]

Each value of the attributes table must be an attribute_info structure (§4.7).

A field can have any number of optional attributes associated with it.

The attributes defined by this specification as appearing in the attributes table of a field_info structure are listed in Table 4.7-C.

The rules concerning attributes defined to appear in the attributes table of a field_info structure are given in §4.7.

The rules concerning non-predefined attributes in the attributes table of a field_info structure are given in §4.7.1.

4.6. Methods

Each method, including each instance initialization method (§2.9.1) and the class or interface initialization method (§2.9.2), is described by a method_info structure.

No two methods in one class file may have the same name and descriptor (§4.3.3).

The structure has the following format:

method_info {
u2 access_flags;
u2 name_index;
u2 descriptor_index;
u2 attributes_count;
attribute_info attributes[attributes_count];
}

The items of the method_info structure are as follows:

access_flags

The value of the access_flags item is a mask of flags used to denote access permission to and properties of this method. The interpretation of each flag, when set, is specified in Table 4.6-A.

Table 4.6-A. Method access and property flags

Flag Name Value Interpretation
ACC_PUBLIC 0x0001 Declared public; may be accessed from outside its package.
ACC_PRIVATE 0x0002 Declared private; accessible only within the defining class.
ACC_PROTECTED 0x0004 Declared protected; may be accessed within subclasses.
ACC_STATIC 0x0008 Declared static.
ACC_FINAL 0x0010 Declared final; must not be overridden (§5.4.5).
ACC_SYNCHRONIZED 0x0020 Declared synchronized; invocation is wrapped by a monitor use.
ACC_BRIDGE 0x0040 A bridge method, generated by the compiler.
ACC_VARARGS 0x0080 Declared with variable number of arguments.
ACC_NATIVE 0x0100 Declared native; implemented in a language other than the Java programming language.
ACC_ABSTRACT 0x0400 Declared abstract; no implementation is provided.
ACC_STRICT 0x0800 Declared strictfp; floating-point mode is FP-strict.
ACC_SYNTHETIC 0x1000 Declared synthetic; not present in the source code.

Methods of classes may have any of the flags in Table 4.6-A set. However, each method of a class may have at most one of its ACC_PUBLICACC_PRIVATE, and ACC_PROTECTED flags set (JLS §8.4.3).

Methods of interfaces may have any of the flags in Table 4.6-A set except ACC_PROTECTEDACC_FINALACC_SYNCHRONIZED, and ACC_NATIVE (JLS §9.4). In a class file whose version number is less than 52.0, each method of an interface must have its ACC_PUBLIC and ACC_ABSTRACT flags set; in a class file whose version number is 52.0 or above, each method of an interface must have exactly one of its ACC_PUBLIC and ACC_PRIVATE flags set.

If a method of a class or interface has its ACC_ABSTRACT flag set, it must not have any of its ACC_PRIVATEACC_STATICACC_FINALACC_SYNCHRONIZEDACC_NATIVE, orACC_STRICT flags set.

An instance initialization method (§2.9.1) may have at most one of its ACC_PUBLICACC_PRIVATE, and ACC_PROTECTED flags set, and may also have its ACC_VARARGSACC_STRICT, and ACC_SYNTHETIC flags set, but must not have any of the other flags in Table 4.6-A set.

A class or interface initialization method (§2.9.2) is called implicitly by the Java Virtual Machine, whereupon the value of its access_flags item is ignored except for the setting of the ACC_STRICT flag. Because the value of its access_flags item is irrelevant (except for the ACC_STRICT flag), a class or interface initialization method is exempt from the preceding rules about legal combinations of flags.

The ACC_BRIDGE flag is used to indicate a bridge method generated by a compiler for the Java programming language.

The ACC_VARARGS flag indicates that this method takes a variable number of arguments at the source code level. A method declared to take a variable number of arguments must be compiled with the ACC_VARARGS flag set to 1. All other methods must be compiled with the ACC_VARARGS flag set to 0.

The ACC_SYNTHETIC flag indicates that this method was generated by a compiler and does not appear in source code, unless it is one of the methods named in §4.7.8.

All bits of the access_flags item not assigned in Table 4.6-A are reserved for future use. They should be set to zero in generated class files and should be ignored by Java Virtual Machine implementations.

name_index

The value of the name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure (§4.4.7) representing either a valid unqualified name denoting a method (§4.2.2), or (if this method is in a class rather than an interface) the special method name <init>, or the special method name <clinit>.

descriptor_index

The value of the descriptor_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure representing a valid method descriptor (§4.3.3). Furthermore:

  • If this method is in a class rather than an interface, and the name of the method is <init>, then the descriptor must denote a void method.

  • If the name of the method is <clinit>, then the descriptor must denote a void method that takes no arguments.

A future edition of this specification may require that the last parameter descriptor of the method descriptor is an array type if the ACC_VARARGS flag is set in the access_flags item.

attributes_count

The value of the attributes_count item indicates the number of additional attributes of this method.

attributes[]

Each value of the attributes table must be an attribute_info structure (§4.7).

A method can have any number of optional attributes associated with it.

The attributes defined by this specification as appearing in the attributes table of a method_info structure are listed in Table 4.7-C.

The rules concerning attributes defined to appear in the attributes table of a method_info structure are given in §4.7.

The rules concerning non-predefined attributes in the attributes table of a method_info structure are given in §4.7.1.

4.7. Attributes

Attributes are used in the ClassFilefield_infomethod_info, and Code_attribute structures of the class file format (§4.1§4.5§4.6§4.7.3).

All attributes have the following general format:

attribute_info {
u2 attribute_name_index;
u4 attribute_length;
u1 info[attribute_length];
}

For all attributes, the attribute_name_index item must be a valid unsigned 16-bit index into the constant pool of the class. The constant_pool entry at attribute_name_index must be aCONSTANT_Utf8_info structure (§4.4.7) representing the name of the attribute. The value of the attribute_length item indicates the length of the subsequent information in bytes. The length does not include the initial six bytes that contain the attribute_name_index and attribute_length items.

26 attributes are predefined by this specification. They are listed three times, for ease of navigation:

  • Table 4.7-A is ordered by the attributes' section numbers in this chapter. Each attribute is accompanied by the first version of the class file format in which it was defined, and the corresponding version of the Java SE Platform. (For old class file versions, the JDK release is used instead of the Java SE Platform version.)

  • Table 4.7-B is orderd by the first version of the class file format in which each attribute was defined.

  • Table 4.7-C is ordered by the location in a class file where each attribute is defined to appear.

Within the context of their use in this specification, that is, in the attributes tables of the class file structures in which they appear, the names of these predefined attributes are reserved.

Any conditions on the presence of a predefined attribute in an attributes table are specified explicitly in the section which describes the attribute. If no conditions are specified, then the attribute may appear any number of times in an attributes table.

The predefined attributes are categorized into three groups according to their purpose:

  1. Five attributes are critical to correct interpretation of the class file by the Java Virtual Machine:

    • ConstantValue

    • Code

    • StackMapTable

    • Exceptions

    • BootstrapMethods

    In a class file of version V, each of these attributes must be recognized and correctly read by an implementation of the Java Virtual Machine if the implementation recognizes class files of version V, and V is at least the version where the attribute was first defined, and the attribute appears in a location where it is defined to appear.

  2. Eight attributes are not critical to correct interpretation of the class file by the Java Virtual Machine, but are either critical to correct interpretation of the class file by the class libraries of the Java SE Platform, or are useful for tools (in which case the section that specifies an attribute describes it as "optional"):

    • InnerClasses

    • EnclosingMethod

    • Synthetic

    • Signature

    • SourceFile

    • LineNumberTable

    • LocalVariableTable

    • LocalVariableTypeTable

    In a class file of version V, each of these attributes must be recognized and correctly read by an implementation of the Java Virtual Machine if the implementation recognizes class files of version V, and V is at least the version where the attribute was first defined, and the attribute appears in a location where it is defined to appear.

  3. Thirteen attributes are not critical to correct interpretation of the class file by the Java Virtual Machine, but contain metadata about the class file that is either exposed by the class libraries of the Java SE Platform, or made available by tools (in which case the section that specifies an attribute describes it as "optional"):

    • SourceDebugExtension

    • Deprecated

    • RuntimeVisibleAnnotations

    • RuntimeInvisibleAnnotations

    • RuntimeVisibleParameterAnnotations

    • RuntimeInvisibleParameterAnnotations

    • RuntimeVisibleTypeAnnotations

    • RuntimeInvisibleTypeAnnotations

    • AnnotationDefault

    • MethodParameters

    • Module

    • ModulePackages

    • ModuleMainClass

    An implementation of the Java Virtual Machine may use the information that these attributes contain, or otherwise must silently ignore these attributes.

Table 4.7-A. Predefined class file attributes (by section)

Attribute Section class file Java SE
ConstantValue §4.7.2 45.3 1.0.2
Code §4.7.3 45.3 1.0.2
StackMapTable §4.7.4 50.0 6
Exceptions §4.7.5 45.3 1.0.2
InnerClasses §4.7.6 45.3 1.1
EnclosingMethod §4.7.7 49.0 5.0
Synthetic §4.7.8 45.3 1.1
Signature §4.7.9 49.0 5.0
SourceFile §4.7.10 45.3 1.0.2
SourceDebugExtension §4.7.11 49.0 5.0
LineNumberTable §4.7.12 45.3 1.0.2
LocalVariableTable §4.7.13 45.3 1.0.2
LocalVariableTypeTable §4.7.14 49.0 5.0
Deprecated §4.7.15 45.3 1.1
RuntimeVisibleAnnotations §4.7.16 49.0 5.0
RuntimeInvisibleAnnotations §4.7.17 49.0 5.0
RuntimeVisibleParameterAnnotations §4.7.18 49.0 5.0
RuntimeInvisibleParameterAnnotations §4.7.19 49.0 5.0
RuntimeVisibleTypeAnnotations §4.7.20 52.0 8
RuntimeInvisibleTypeAnnotations §4.7.21 52.0 8
AnnotationDefault §4.7.22 49.0 5.0
BootstrapMethods §4.7.23 51.0 7
MethodParameters §4.7.24 52.0 8

Module

§4.7.25 53.0 9

ModulePackages

§4.7.26 53.0 9

ModuleMainClass

§4.7.27 53.0 9

Table 4.7-B. Predefined class file attributes (by class file version)

Attribute class file Java SE Section
ConstantValue 45.3 1.0.2 §4.7.2
Code 45.3 1.0.2 §4.7.3
Exceptions 45.3 1.0.2 §4.7.5
SourceFile 45.3 1.0.2 §4.7.10
LineNumberTable 45.3 1.0.2 §4.7.12
LocalVariableTable 45.3 1.0.2 §4.7.13
InnerClasses 45.3 1.1 §4.7.6
Synthetic 45.3 1.1 §4.7.8
Deprecated 45.3 1.1 §4.7.15
EnclosingMethod 49.0 5.0 §4.7.7
Signature 49.0 5.0 §4.7.9
SourceDebugExtension 49.0 5.0 §4.7.11
LocalVariableTypeTable 49.0 5.0 §4.7.14
RuntimeVisibleAnnotations 49.0 5.0 §4.7.16
RuntimeInvisibleAnnotations 49.0 5.0 §4.7.17
RuntimeVisibleParameterAnnotations 49.0 5.0 §4.7.18
RuntimeInvisibleParameterAnnotations 49.0 5.0 §4.7.19
AnnotationDefault 49.0 5.0 §4.7.22
StackMapTable 50.0 6 §4.7.4
BootstrapMethods 51.0 7 §4.7.23
RuntimeVisibleTypeAnnotations 52.0 8 §4.7.20
RuntimeInvisibleTypeAnnotations 52.0 8 §4.7.21
MethodParameters 52.0 8 §4.7.24

Module

53.0 9 §4.7.25

ModulePackages

53.0 9 §4.7.26

ModuleMainClass

53.0 9 §4.7.27

Table 4.7-C. Predefined class file attributes (by location)

Attribute Location class file
SourceFile ClassFile 45.3
InnerClasses ClassFile 45.3
EnclosingMethod ClassFile 49.0
SourceDebugExtension ClassFile 49.0
BootstrapMethods ClassFile 51.0

ModuleModulePackagesModuleMainClass

ClassFile 53.0
ConstantValue field_info 45.3
Code method_info 45.3
Exceptions method_info 45.3
RuntimeVisibleParameterAnnotationsRuntimeInvisibleParameterAnnotations method_info 49.0
AnnotationDefault method_info 49.0
MethodParameters method_info 52.0
Synthetic ClassFilefield_infomethod_info 45.3
Deprecated ClassFilefield_infomethod_info 45.3
Signature ClassFilefield_infomethod_info 49.0
RuntimeVisibleAnnotationsRuntimeInvisibleAnnotations ClassFilefield_infomethod_info 49.0
LineNumberTable Code 45.3
LocalVariableTable Code 45.3
LocalVariableTypeTable Code 49.0
StackMapTable Code 50.0
RuntimeVisibleTypeAnnotationsRuntimeInvisibleTypeAnnotations ClassFilefield_infomethod_infoCode 52.0

4.7.1. Defining and Naming New Attributes

Compilers are permitted to define and emit class files containing new attributes in the attributes tables of class file structures, field_info structures, method_info structures, and Codeattributes (§4.7.3). Java Virtual Machine implementations are permitted to recognize and use new attributes found in these attributes tables. However, any attribute not defined as part of this specification must not affect the semantics of the class file. Java Virtual Machine implementations are required to silently ignore attributes they do not recognize.

For instance, defining a new attribute to support vendor-specific debugging is permitted. Because Java Virtual Machine implementations are required to ignore attributes they do not recognize, class files intended for that particular Java Virtual Machine implementation will be usable by other implementations even if those implementations cannot make use of the additional debugging information that the class files contain.

Java Virtual Machine implementations are specifically prohibited from throwing an exception or otherwise refusing to use class files simply because of the presence of some new attribute. Of course, tools operating on class files may not run correctly if given class files that do not contain all the attributes they require.

Two attributes that are intended to be distinct, but that happen to use the same attribute name and are of the same length, will conflict on implementations that recognize either attribute. Attributes defined other than in this specification should have names chosen according to the package naming convention described in The Java Language Specification, Java SE 9 Edition (JLS §6.1).

Future versions of this specification may define additional attributes.

4.7.2. The ConstantValue Attribute

The ConstantValue attribute is a fixed-length attribute in the attributes table of a field_info structure (§4.5). A ConstantValue attribute represents the value of a constant expression (JLS §15.28), and is used as follows:

  • If the ACC_STATIC flag in the access_flags item of the field_info structure is set, then the field represented by the field_info structure is assigned the value represented by itsConstantValue attribute as part of the initialization of the class or interface declaring the field (§5.5). This occurs prior to the invocation of the class or interface initialization method of that class or interface (§2.9.2).

  • Otherwise, the Java Virtual Machine must silently ignore the attribute.

There may be at most one ConstantValue attribute in the attributes table of a field_info structure.

The ConstantValue attribute has the following format:

ConstantValue_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 constantvalue_index;
}

The items of the ConstantValue_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "ConstantValue".

attribute_length

The value of the attribute_length item of a ConstantValue_attribute structure must be two.

constantvalue_index

The value of the constantvalue_index item must be a valid index into the constant_pool table. The constant_pool entry at that index gives the constant value represented by this attribute. The constant_pool entry must be of a type appropriate to the field, as specified in Table 4.7.2-A.

Table 4.7.2-A. Constant value attribute types

Field Type Entry Type
long CONSTANT_Long
float CONSTANT_Float
double CONSTANT_Double
intshortcharbyteboolean CONSTANT_Integer
String CONSTANT_String

4.7.3. The Code Attribute

The Code attribute is a variable-length attribute in the attributes table of a method_info structure (§4.6). A Code attribute contains the Java Virtual Machine instructions and auxiliary information for a method, including an instance initialization method and a class or interface initialization method (§2.9.1§2.9.2).

If the method is either native or abstract, and is not a class or interface initialization method, then its method_info structure must not have a Code attribute in its attributes table. Otherwise, its method_info structure must have exactly one Code attribute in its attributes table.

The Code attribute has the following format:

Code_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 max_stack;
u2 max_locals;
u4 code_length;
u1 code[code_length];
u2 exception_table_length;
{ u2 start_pc;
u2 end_pc;
u2 handler_pc;
u2 catch_type;
} exception_table[exception_table_length];
u2 attributes_count;
attribute_info attributes[attributes_count];
}

The items of the Code_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "Code".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

max_stack

The value of the max_stack item gives the maximum depth of the operand stack of this method (§2.6.2) at any point during execution of the method.

max_locals

The value of the max_locals item gives the number of local variables in the local variable array allocated upon invocation of this method (§2.6.1), including the local variables used to pass parameters to the method on its invocation.

The greatest local variable index for a value of type long or double is max_locals - 2. The greatest local variable index for a value of any other type is max_locals - 1.

code_length

The value of the code_length item gives the number of bytes in the code array for this method.

The value of code_length must be greater than zero (as the code array must not be empty) and less than 65536.

code[]

The code array gives the actual bytes of Java Virtual Machine code that implement the method.

When the code array is read into memory on a byte-addressable machine, if the first byte of the array is aligned on a 4-byte boundary, the tableswitch and lookupswitch 32-bit offsets will be 4-byte aligned. (Refer to the descriptions of those instructions for more information on the consequences of code array alignment.)

The detailed constraints on the contents of the code array are extensive and are given in a separate section (§4.9).

exception_table_length

The value of the exception_table_length item gives the number of entries in the exception_table table.

exception_table[]

Each entry in the exception_table array describes one exception handler in the code array. The order of the handlers in the exception_table array is significant (§2.10).

Each exception_table entry contains the following four items:

start_pc, end_pc

The values of the two items start_pc and end_pc indicate the ranges in the code array at which the exception handler is active. The value of start_pc must be a valid index into the code array of the opcode of an instruction. The value of end_pc either must be a valid index into the code array of the opcode of an instruction or must be equal to code_length, the length of the code array. The value of start_pc must be less than the value of end_pc.

The start_pc is inclusive and end_pc is exclusive; that is, the exception handler must be active while the program counter is within the interval [start_pcend_pc).

The fact that end_pc is exclusive is a historical mistake in the design of the Java Virtual Machine: if the Java Virtual Machine code for a method is exactly 65535 bytes long and ends with an instruction that is 1 byte long, then that instruction cannot be protected by an exception handler. A compiler writer can work around this bug by limiting the maximum size of the generated Java Virtual Machine code for any method, instance initialization method, or static initializer (the size of any code array) to 65534 bytes.

handler_pc

The value of the handler_pc item indicates the start of the exception handler. The value of the item must be a valid index into the code array and must be the index of the opcode of an instruction.

catch_type

If the value of the catch_type item is nonzero, it must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Class_info structure (§4.4.1) representing a class of exceptions that this exception handler is designated to catch. The exception handler will be called only if the thrown exception is an instance of the given class or one of its subclasses.

The verifier checks that the class is Throwable or a subclass of Throwable (§4.9.2).

If the value of the catch_type item is zero, this exception handler is called for all exceptions.

This is used to implement finally (§3.13).

attributes_count

The value of the attributes_count item indicates the number of attributes of the Code attribute.

attributes[]

Each value of the attributes table must be an attribute_info structure (§4.7).

Code attribute can have any number of optional attributes associated with it.

The attributes defined by this specification as appearing in the attributes table of a Code attribute are listed in Table 4.7-C.

The rules concerning attributes defined to appear in the attributes table of a Code attribute are given in §4.7.

The rules concerning non-predefined attributes in the attributes table of a Code attribute are given in §4.7.1.

4.7.4. The StackMapTable Attribute

The StackMapTable attribute is a variable-length attribute in the attributes table of a Code attribute (§4.7.3). A StackMapTable attribute is used during the process of verification by type checking (§4.10.1).

There may be at most one StackMapTable attribute in the attributes table of a Code attribute.

In a class file whose version number is 50.0 or above, if a method's Code attribute does not have a StackMapTable attribute, it has an implicit stack map attribute (§4.10.1). This implicit stack map attribute is equivalent to a StackMapTable attribute with number_of_entries equal to zero.

The StackMapTable attribute has the following format:

StackMapTable_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 number_of_entries;
stack_map_frame entries[number_of_entries];
}

The items of the StackMapTable_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "StackMapTable".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

number_of_entries

The value of the number_of_entries item gives the number of stack_map_frame entries in the entries table.

entries[]

Each entry in the entries table describes one stack map frame of the method. The order of the stack map frames in the entries table is significant.

stack map frame specifies (either explicitly or implicitly) the bytecode offset at which it applies, and the verification types of local variables and operand stack entries for that offset.

Each stack map frame described in the entries table relies on the previous frame for some of its semantics. The first stack map frame of a method is implicit, and computed from the method descriptor by the type checker (§4.10.1.6). The stack_map_frame structure at entries[0] therefore describes the second stack map frame of the method.

The bytecode offset at which a stack map frame applies is calculated by taking the value offset_delta specified in the frame (either explicitly or implicitly), and adding offset_delta + 1 to the bytecode offset of the previous frame, unless the previous frame is the initial frame of the method. In that case, the bytecode offset at which the stack map frame applies is the value offset_delta specified in the frame.

By using an offset delta rather than storing the actual bytecode offset, we ensure, by definition, that stack map frames are in the correctly sorted order. Furthermore, by consistently using the formula offset_delta + 1 for all explicit frames (as opposed to the implicit first frame), we guarantee the absence of duplicates.

We say that an instruction in the bytecode has a corresponding stack map frame if the instruction starts at offset i in the code array of a Code attribute, and the Code attribute has aStackMapTable attribute whose entries array contains a stack map frame that applies at bytecode offset i.

verification type specifies the type of either one or two locations, where a location is either a single local variable or a single operand stack entry. A verification type is represented by a discriminated union, verification_type_info, that consists of a one-byte tag, indicating which item of the union is in use, followed by zero or more bytes, giving more information about the tag.

union verification_type_info {
Top_variable_info;
Integer_variable_info;
Float_variable_info;
Long_variable_info;
Double_variable_info;
Null_variable_info;
UninitializedThis_variable_info;
Object_variable_info;
Uninitialized_variable_info;
}

A verification type that specifies one location in the local variable array or in the operand stack is represented by the following items of the verification_type_info union:

  • The Top_variable_info item indicates that the local variable has the verification type top.

    Top_variable_info {
    u1 tag = ITEM_Top; /* 0 */
    }
  • The Integer_variable_info item indicates that the location has the verification type int.

    Integer_variable_info {
    u1 tag = ITEM_Integer; /* 1 */
    }
  • The Float_variable_info item indicates that the location has the verification type float.

    Float_variable_info {
    u1 tag = ITEM_Float; /* 2 */
    }
  • The Null_variable_info type indicates that the location has the verification type null.

    Null_variable_info {
    u1 tag = ITEM_Null; /* 5 */
    }
  • The UninitializedThis_variable_info item indicates that the location has the verification type uninitializedThis.

    UninitializedThis_variable_info {
    u1 tag = ITEM_UninitializedThis; /* 6 */
    }
  • The Object_variable_info item indicates that the location has the verification type which is the class represented by the CONSTANT_Class_info structure (§4.4.1) found in the constant_pool table at the index given by cpool_index.

    Object_variable_info {
    u1 tag = ITEM_Object; /* 7 */
    u2 cpool_index;
    }
  • The Uninitialized_variable_info item indicates that the location has the verification type uninitialized(Offset). The Offset item indicates the offset, in the code array of the Code attribute that contains this StackMapTable attribute, of the new instruction (§new) that created the object being stored in the location.

    Uninitialized_variable_info {
    u1 tag = ITEM_Uninitialized; /* 8 */
    u2 offset;
    }

A verification type that specifies two locations in the local variable array or in the operand stack is represented by the following items of the verification_type_info union:

  • The Long_variable_info item indicates that the first of two locations has the verification type long.

    Long_variable_info {
    u1 tag = ITEM_Long; /* 4 */
    }
  • The Double_variable_info item indicates that the first of two locations has the verification type double.

    Double_variable_info {
    u1 tag = ITEM_Double; /* 3 */
    }
  • The Long_variable_info and Double_variable_info items indicate the verification type of the second of two locations as follows:

    • If the first of the two locations is a local variable, then:

      • It must not be the local variable with the highest index.

      • The next higher numbered local variable has the verification type top.

    • If the first of the two locations is an operand stack entry, then:

      • It must not be the topmost location of the operand stack.

      • The next location closer to the top of the operand stack has the verification type top.

A stack map frame is represented by a discriminated union, stack_map_frame, which consists of a one-byte tag, indicating which item of the union is in use, followed by zero or more bytes, giving more information about the tag.

union stack_map_frame {
same_frame;
same_locals_1_stack_item_frame;
same_locals_1_stack_item_frame_extended;
chop_frame;
same_frame_extended;
append_frame;
full_frame;
}

The tag indicates the frame type of the stack map frame:

  • The frame type same_frame is represented by tags in the range [0-63]. This frame type indicates that the frame has exactly the same local variables as the previous frame and that the operand stack is empty. The offset_delta value for the frame is the value of the tag item, frame_type.

    same_frame {
    u1 frame_type = SAME; /* 0-63 */
    }
  • The frame type same_locals_1_stack_item_frame is represented by tags in the range [64, 127]. This frame type indicates that the frame has exactly the same local variables as the previous frame and that the operand stack has one entry. The offset_delta value for the frame is given by the formula frame_type - 64. The verification type of the one stack entry appears after the frame type.

    same_locals_1_stack_item_frame {
    u1 frame_type = SAME_LOCALS_1_STACK_ITEM; /* 64-127 */
    verification_type_info stack[1];
    }
  • Tags in the range [128-246] are reserved for future use.

  • The frame type same_locals_1_stack_item_frame_extended is represented by the tag 247. This frame type indicates that the frame has exactly the same local variables as the previous frame and that the operand stack has one entry. The offset_delta value for the frame is given explicitly, unlike in the frame type same_locals_1_stack_item_frame. The verification type of the one stack entry appears after offset_delta.

    same_locals_1_stack_item_frame_extended {
    u1 frame_type = SAME_LOCALS_1_STACK_ITEM_EXTENDED; /* 247 */
    u2 offset_delta;
    verification_type_info stack[1];
    }
  • The frame type chop_frame is represented by tags in the range [248-250]. This frame type indicates that the frame has the same local variables as the previous frame except that the last klocal variables are absent, and that the operand stack is empty. The value of k is given by the formula 251 - frame_type. The offset_delta value for the frame is given explicitly.

    chop_frame {
    u1 frame_type = CHOP; /* 248-250 */
    u2 offset_delta;
    }

    Assume the verification types of local variables in the previous frame are given by locals, an array structured as in the full_frame frame type. If locals[M-1] in the previous frame represented local variable X and locals[M] represented local variable Y, then the effect of removing one local variable is that locals[M-1] in the new frame represents local variable X andlocals[M] is undefined.

    It is an error if k is larger than the number of local variables in locals for the previous frame, that is, if the number of local variables in the new frame would be less than zero.

  • The frame type same_frame_extended is represented by the tag 251. This frame type indicates that the frame has exactly the same local variables as the previous frame and that the operand stack is empty. The offset_delta value for the frame is given explicitly, unlike in the frame type same_frame.

    same_frame_extended {
    u1 frame_type = SAME_FRAME_EXTENDED; /* 251 */
    u2 offset_delta;
    }
  • The frame type append_frame is represented by tags in the range [252-254]. This frame type indicates that the frame has the same locals as the previous frame except that k additional locals are defined, and that the operand stack is empty. The value of k is given by the formula frame_type - 251. The offset_delta value for the frame is given explicitly.

    append_frame {
    u1 frame_type = APPEND; /* 252-254 */
    u2 offset_delta;
    verification_type_info locals[frame_type - 251];
    }

    The 0th entry in locals represents the verification type of the first additional local variable. If locals[M] represents local variable N, then:

    • locals[M+1] represents local variable N+1 if locals[M] is one of Top_variable_infoInteger_variable_infoFloat_variable_infoNull_variable_info,UninitializedThis_variable_infoObject_variable_info, or Uninitialized_variable_info; and

    • locals[M+1] represents local variable N+2 if locals[M] is either Long_variable_info or Double_variable_info.

    It is an error if, for any index ilocals[i] represents a local variable whose index is greater than the maximum number of local variables for the method.

  • The frame type full_frame is represented by the tag 255. The offset_delta value for the frame is given explicitly.

    full_frame {
    u1 frame_type = FULL_FRAME; /* 255 */
    u2 offset_delta;
    u2 number_of_locals;
    verification_type_info locals[number_of_locals];
    u2 number_of_stack_items;
    verification_type_info stack[number_of_stack_items];
    }

    The 0th entry in locals represents the verification type of local variable 0. If locals[M] represents local variable N, then:

    • locals[M+1] represents local variable N+1 if locals[M] is one of Top_variable_infoInteger_variable_infoFloat_variable_infoNull_variable_info,UninitializedThis_variable_infoObject_variable_info, or Uninitialized_variable_info; and

    • locals[M+1] represents local variable N+2 if locals[M] is either Long_variable_info or Double_variable_info.

    It is an error if, for any index ilocals[i] represents a local variable whose index is greater than the maximum number of local variables for the method.

    The 0th entry in stack represents the verification type of the bottom of the operand stack, and subsequent entries in stack represent the verification types of stack entries closer to the top of the operand stack. We refer to the bottom of the operand stack as stack entry 0, and to subsequent entries of the operand stack as stack entry 1, 2, etc. If stack[M] represents stack entry N, then:

    • stack[M+1] represents stack entry N+1 if stack[M] is one of Top_variable_infoInteger_variable_infoFloat_variable_infoNull_variable_infoUninitializedThis_variable_infoObject_variable_info, or Uninitialized_variable_info; and

    • stack[M+1] represents stack entry N+2 if stack[M] is either Long_variable_info or Double_variable_info.

    It is an error if, for any index istack[i] represents a stack entry whose index is greater than the maximum operand stack size for the method.

4.7.5. The Exceptions Attribute

The Exceptions attribute is a variable-length attribute in the attributes table of a method_info structure (§4.6). The Exceptions attribute indicates which checked exceptions a method may throw.

There may be at most one Exceptions attribute in the attributes table of a method_info structure.

The Exceptions attribute has the following format:

Exceptions_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 number_of_exceptions;
u2 exception_index_table[number_of_exceptions];
}

The items of the Exceptions_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be the CONSTANT_Utf8_infostructure (§4.4.7) representing the string "Exceptions".

attribute_length

The value of the attribute_length item indicates the attribute length, excluding the initial six bytes.

number_of_exceptions

The value of the number_of_exceptions item indicates the number of entries in the exception_index_table.

exception_index_table[]

Each value in the exception_index_table array must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Class_infostructure (§4.4.1) representing a class type that this method is declared to throw.

A method should throw an exception only if at least one of the following three criteria is met:

  • The exception is an instance of RuntimeException or one of its subclasses.

  • The exception is an instance of Error or one of its subclasses.

  • The exception is an instance of one of the exception classes specified in the exception_index_table just described, or one of their subclasses.

These requirements are not enforced in the Java Virtual Machine; they are enforced only at compile time.

4.7.6. The InnerClasses Attribute

The InnerClasses attribute is a variable-length attribute in the attributes table of a ClassFile structure (§4.1).

If the constant pool of a class or interface C contains at least one CONSTANT_Class_info entry (§4.4.1) which represents a class or interface that is not a member of a package, then there must be exactly one InnerClasses attribute in the attributes table of the ClassFile structure for C.

The InnerClasses attribute has the following format:

InnerClasses_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 number_of_classes;
{ u2 inner_class_info_index;
u2 outer_class_info_index;
u2 inner_name_index;
u2 inner_class_access_flags;
} classes[number_of_classes];
}

The items of the InnerClasses_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "InnerClasses".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

number_of_classes

The value of the number_of_classes item indicates the number of entries in the classes array.

classes[]

Every CONSTANT_Class_info entry in the constant_pool table which represents a class or interface C that is not a package member must have exactly one corresponding entry in the classes array.

If a class or interface has members that are classes or interfaces, its constant_pool table (and hence its InnerClasses attribute) must refer to each such member (JLS §13.1), even if that member is not otherwise mentioned by the class.

In addition, the constant_pool table of every nested class and nested interface must refer to its enclosing class, so altogether, every nested class and nested interface will haveInnerClasses information for each enclosing class and for each of its own nested classes and interfaces.

Each entry in the classes array contains the following four items:

inner_class_info_index

The value of the inner_class_info_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be aCONSTANT_Class_info structure representing C.

outer_class_info_index

If C is not a member of a class or an interface - that is, if C is a top-level class or interface (JLS §7.6) or a local class (JLS §14.3) or an anonymous class (JLS §15.9.5) - then the value of the outer_class_info_index item must be zero.

Otherwise, the value of the outer_class_info_index item must be a valid index into the constant_pool table, and the entry at that index must be a CONSTANT_Class_infostructure representing the class or interface of which C is a member. The value of the outer_class_info_index item must not equal the the value of the inner_class_info_index item.

inner_name_index

If C is anonymous (JLS §15.9.5), the value of the inner_name_index item must be zero.

Otherwise, the value of the inner_name_index item must be a valid index into the constant_pool table, and the entry at that index must be a CONSTANT_Utf8_info structure that represents the original simple name of C, as given in the source code from which this class file was compiled.

inner_class_access_flags

The value of the inner_class_access_flags item is a mask of flags used to denote access permissions to and properties of class or interface C as declared in the source code from which this class file was compiled. It is used by a compiler to recover the original information when source code is not available. The flags are specified in Table 4.7.6-A.

Table 4.7.6-A. Nested class access and property flags

Flag Name Value Interpretation
ACC_PUBLIC 0x0001 Marked or implicitly public in source.
ACC_PRIVATE 0x0002 Marked private in source.
ACC_PROTECTED 0x0004 Marked protected in source.
ACC_STATIC 0x0008 Marked or implicitly static in source.
ACC_FINAL 0x0010 Marked or implicitly final in source.
ACC_INTERFACE 0x0200 Was an interface in source.
ACC_ABSTRACT 0x0400 Marked or implicitly abstract in source.
ACC_SYNTHETIC 0x1000 Declared synthetic; not present in the source code.
ACC_ANNOTATION 0x2000 Declared as an annotation type.
ACC_ENUM 0x4000 Declared as an enum type.

All bits of the inner_class_access_flags item not assigned in Table 4.7.6-A are reserved for future use. They should be set to zero in generated class files and should be ignored by Java Virtual Machine implementations.

If a class file has a version number that is 51.0 or above, and has an InnerClasses attribute in its attributes table, then for all entries in the classes array of the InnerClassesattribute, the value of the outer_class_info_index item must be zero if the value of the inner_name_index item is zero.

Oracle's Java Virtual Machine implementation does not check the consistency of an InnerClasses attribute against a class file representing a class or interface referenced by the attribute.

4.7.7. The EnclosingMethod Attribute

The EnclosingMethod attribute is a fixed-length attribute in the attributes table of a ClassFile structure (§4.1). A class must have an EnclosingMethod attribute if and only if it represents a local class or an anonymous class (JLS §14.3, JLS §15.9.5).

There may be at most one EnclosingMethod attribute in the attributes table of a ClassFile structure.

The EnclosingMethod attribute has the following format:

EnclosingMethod_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 class_index;
u2 method_index;
}

The items of the EnclosingMethod_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "EnclosingMethod".

attribute_length

The value of the attribute_length item must be four.

class_index

The value of the class_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Class_info structure (§4.4.1) representing the innermost class that encloses the declaration of the current class.

method_index

If the current class is not immediately enclosed by a method or constructor, then the value of the method_index item must be zero.

In particular, method_index must be zero if the current class was immediately enclosed in source code by an instance initializer, static initializer, instance variable initializer, or class variable initializer. (The first two concern both local classes and anonymous classes, while the last two concern anonymous classes declared on the right hand side of a field assignment.)

Otherwise, the value of the method_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be aCONSTANT_NameAndType_info structure (§4.4.6) representing the name and type of a method in the class referenced by the class_index attribute above.

It is the responsibility of a Java compiler to ensure that the method identified via the method_index is indeed the closest lexically enclosing method of the class that contains thisEnclosingMethod attribute.

4.7.8. The Synthetic Attribute

The Synthetic attribute is a fixed-length attribute in the attributes table of a ClassFilefield_info, or method_info structure (§4.1§4.5§4.6). A class member that does not appear in the source code must be marked using a Synthetic attribute, or else it must have its ACC_SYNTHETIC flag set. The only exceptions to this requirement are compiler-generated methods which are not considered implementation artifacts, namely the instance initialization method representing a default constructor of the Java programming language (§2.9.1), the class or interface initialization method (§2.9.2), and the Enum.values() and Enum.valueOf() methods.

The Synthetic attribute was introduced in JDK 1.1 to support nested classes and interfaces.

The Synthetic attribute has the following format:

Synthetic_attribute {
u2 attribute_name_index;
u4 attribute_length;
}

The items of the Synthetic_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "Synthetic".

attribute_length

The value of the attribute_length item must be zero.

4.7.9. The Signature Attribute

The Signature attribute is a fixed-length attribute in the attributes table of a ClassFilefield_info, or method_info structure (§4.1§4.5§4.6). A Signature attribute records a signature (§4.7.9.1) for a class, interface, constructor, method, or field whose declaration in the Java programming language uses type variables or parameterized types. See The Java Language Specification, Java SE 9 Edition for details about these constructs.

There may be at most one Signature attribute in the attributes table of a ClassFilefield_info, or method_info structure.

The Signature attribute has the following format:

Signature_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 signature_index;
}

The items of the Signature_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "Signature".

attribute_length

The value of the attribute_length item of a Signature_attribute structure must be two.

signature_index

The value of the signature_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure (§4.4.7) representing a class signature if this Signature attribute is an attribute of a ClassFile structure; a method signature if this Signature attribute is an attribute of amethod_info structure; or a field signature otherwise.

Oracle's Java Virtual Machine implementation does not check the well-formedness of Signature attributes during class loading or linking. Instead, Signature attributes are checked by methods of the Java SE Platform class libraries which expose generic signatures of classes, interfaces, constructors, methods, and fields. Examples include getGenericSuperclass in Class and toGenericString in java.lang.reflect.Executable.

4.7.9.1. Signatures

Signatures encode declarations written in the Java programming language that use types outside the type system of the Java Virtual Machine. They support reflection and debugging, as well as compilation when only class files are available.

A Java compiler must emit a signature for any class, interface, constructor, method, or field whose declaration uses type variables or parameterized types. Specifically, a Java compiler must emit:

  • A class signature for any class or interface declaration which is either generic, or has a parameterized type as a superclass or superinterface, or both.

  • A method signature for any method or constructor declaration which is either generic, or has a type variable or parameterized type as the return type or a formal parameter type, or has a type variable in a throws clause, or any combination thereof.

    If the throws clause of a method or constructor declaration does not involve type variables, then a compiler may treat the declaration as having no throws clause for the purpose of emitting a method signature.

  • A field signature for any field, formal parameter, or local variable declaration whose type uses a type variable or a parameterized type.

Signatures are specified using a grammar which follows the notation of §4.3.1. In addition to that notation:

  • The syntax [x] on the right-hand side of a production denotes zero or one occurrences of x. That is, x is an optional symbol. The alternative which contains the optional symbol actually defines two alternatives: one that omits the optional symbol and one that includes it.

  • A very long right-hand side may be continued on a second line by clearly indenting the second line.

The grammar includes the terminal symbol Identifier to denote the name of a type, field, method, formal parameter, local variable, or type variable, as generated by a Java compiler. Such a name must not contain any of the ASCII characters . ; [ / < > : (that is, the characters forbidden in method names (§4.2.2) and also colon) but may contain characters that must not appear in an identifier in the Java programming language (JLS §3.8).

Signatures rely on a hierarchy of nonterminals known as type signatures:

class signature encodes type information about a (possibly generic) class declaration. It describes any type parameters of the class, and lists its (possibly parameterized) direct superclass and direct superinterfaces, if any. A type parameter is described by its name, followed by any class bound and interface bounds.

TypeParameters:
TypeParameter:
Identifier ClassBound {InterfaceBound}
ClassBound:
InterfaceBound:
SuperclassSignature:
SuperinterfaceSignature:

method signature encodes type information about a (possibly generic) method declaration. It describes any type parameters of the method; the (possibly parameterized) types of any formal parameters; the (possibly parameterized) return type, if any; and the types of any exceptions declared in the method's throws clause.

The following production from §4.3.3 is repeated here for convenience:

VoidDescriptor:
V

A method signature encoded by the Signature attribute may not correspond exactly to the method descriptor in the method_info structure (§4.3.3). In particular, there is no assurance that the number of formal parameter types in the method signature is the same as the number of parameter descriptors in the method descriptor. The numbers are the same for most methods, but certain constructors in the Java programming language have an implicitly declared parameter which a compiler represents with a parameter descriptor but may omit from the method signature. See the note in §4.7.18 for a similar situation involving parameter annotations.

field signature encodes the (possibly parameterized) type of a field, formal parameter, or local variable declaration.

FieldSignature:

4.7.10. The SourceFile Attribute

The SourceFile attribute is an optional fixed-length attribute in the attributes table of a ClassFile structure (§4.1).

There may be at most one SourceFile attribute in the attributes table of a ClassFile structure.

The SourceFile attribute has the following format:

SourceFile_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 sourcefile_index;
}

The items of the SourceFile_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "SourceFile".

attribute_length

The value of the attribute_length item of a SourceFile_attribute structure must be two.

sourcefile_index

The value of the sourcefile_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure representing a string.

The string referenced by the sourcefile_index item will be interpreted as indicating the name of the source file from which this class file was compiled. It will not be interpreted as indicating the name of a directory containing the file or an absolute path name for the file; such platform-specific additional information must be supplied by the run-time interpreter or development tool at the time the file name is actually used.

4.7.11. The SourceDebugExtension Attribute

The SourceDebugExtension attribute is an optional attribute in the attributes table of a ClassFile structure (§4.1).

There may be at most one SourceDebugExtension attribute in the attributes table of a ClassFile structure.

The SourceDebugExtension attribute has the following format:

SourceDebugExtension_attribute {
u2 attribute_name_index;
u4 attribute_length;
u1 debug_extension[attribute_length];
}

The items of the SourceDebugExtension_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "SourceDebugExtension".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

debug_extension[]

The debug_extension array holds extended debugging information which has no semantic effect on the Java Virtual Machine. The information is represented using a modified UTF-8 string (§4.4.7) with no terminating zero byte.

Note that the debug_extension array may denote a string longer than that which can be represented with an instance of class String.

4.7.12. The LineNumberTable Attribute

The LineNumberTable attribute is an optional variable-length attribute in the attributes table of a Code attribute (§4.7.3). It may be used by debuggers to determine which part of the codearray corresponds to a given line number in the original source file.

If multiple LineNumberTable attributes are present in the attributes table of a Code attribute, then they may appear in any order.

There may be more than one LineNumberTable attribute per line of a source file in the attributes table of a Code attribute. That is, LineNumberTable attributes may together represent a given line of a source file, and need not be one-to-one with source lines.

The LineNumberTable attribute has the following format:

LineNumberTable_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 line_number_table_length;
{ u2 start_pc;
u2 line_number;
} line_number_table[line_number_table_length];
}

The items of the LineNumberTable_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "LineNumberTable".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

line_number_table_length

The value of the line_number_table_length item indicates the number of entries in the line_number_table array.

line_number_table[]

Each entry in the line_number_table array indicates that the line number in the original source file changes at a given point in the code array. Each line_number_table entry must contain the following two items:

start_pc

The value of the start_pc item must be a valid index into the code array of this Code attribute. The item indicates the index into the code array at which the code for a new line in the original source file begins.

line_number

The value of the line_number item gives the corresponding line number in the original source file.

4.7.13. The LocalVariableTable Attribute

The LocalVariableTable attribute is an optional variable-length attribute in the attributes table of a Code attribute (§4.7.3). It may be used by debuggers to determine the value of a given local variable during the execution of a method.

If multiple LocalVariableTable attributes are present in the attributes table of a Code attribute, then they may appear in any order.

There may be no more than one LocalVariableTable attribute per local variable in the attributes table of a Code attribute.

The LocalVariableTable attribute has the following format:

LocalVariableTable_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 local_variable_table_length;
{ u2 start_pc;
u2 length;
u2 name_index;
u2 descriptor_index;
u2 index;
} local_variable_table[local_variable_table_length];
}

The items of the LocalVariableTable_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "LocalVariableTable".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

local_variable_table_length

The value of the local_variable_table_length item indicates the number of entries in the local_variable_table array.

local_variable_table[]

Each entry in the local_variable_table array indicates a range of code array offsets within which a local variable has a value, and indicates the index into the local variable array of the current frame at which that local variable can be found. Each entry must contain the following five items:

start_pc, length

The value of the start_pc item must be a valid index into the code array of this Code attribute and must be the index of the opcode of an instruction.

The value of start_pc + length must either be a valid index into the code array of this Code attribute and be the index of the opcode of an instruction, or it must be the first index beyond the end of that code array.

The start_pc and length items indicate that the given local variable has a value at indices into the code array in the interval [start_pcstart_pc + length), that is, between start_pc inclusive and start_pc + length exclusive.

name_index

The value of the name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must contain a CONSTANT_Utf8_info structure representing a valid unqualified name denoting a local variable (§4.2.2).

descriptor_index

The value of the descriptor_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must contain a CONSTANT_Utf8_infostructure representing a field descriptor which encodes the type of a local variable in the source program (§4.3.2).

index

The value of the index item must be a valid index into the local variable array of the current frame. The given local variable is at index in the local variable array of the current frame.

If the given local variable is of type double or long, it occupies both index and index + 1.

4.7.14. The LocalVariableTypeTable Attribute

The LocalVariableTypeTable attribute is an optional variable-length attribute in the attributes table of a Code attribute (§4.7.3). It may be used by debuggers to determine the value of a given local variable during the execution of a method.

If multiple LocalVariableTypeTable attributes are present in the attributes table of a given Code attribute, then they may appear in any order.

There may be no more than one LocalVariableTypeTable attribute per local variable in the attributes table of a Code attribute.

The LocalVariableTypeTable attribute differs from the LocalVariableTable attribute (§4.7.13) in that it provides signature information rather than descriptor information. This difference is only significant for variables whose type uses a type variable or parameterized type. Such variables will appear in both tables, while variables of other types will appear only in LocalVariableTable.

The LocalVariableTypeTable attribute has the following format:

LocalVariableTypeTable_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 local_variable_type_table_length;
{ u2 start_pc;
u2 length;
u2 name_index;
u2 signature_index;
u2 index;
} local_variable_type_table[local_variable_type_table_length];
}

The items of the LocalVariableTypeTable_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "LocalVariableTypeTable".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

local_variable_type_table_length

The value of the local_variable_type_table_length item indicates the number of entries in the local_variable_type_table array.

local_variable_type_table[]

Each entry in the local_variable_type_table array indicates a range of code array offsets within which a local variable has a value, and indicates the index into the local variable array of the current frame at which that local variable can be found. Each entry must contain the following five items:

start_pc, length

The value of the start_pc item must be a valid index into the code array of this Code attribute and must be the index of the opcode of an instruction.

The value of start_pc + length must either be a valid index into the code array of this Code attribute and be the index of the opcode of an instruction, or it must be the first index beyond the end of that code array.

The start_pc and length items indicate that the given local variable has a value at indices into the code array in the interval [start_pcstart_pc + length), that is, between start_pc inclusive and start_pc + length exclusive.

name_index

The value of the name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must contain a CONSTANT_Utf8_info structure representing a valid unqualified name denoting a local variable (§4.2.2).

signature_index

The value of the signature_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must contain a CONSTANT_Utf8_infostructure representing a field signature which encodes the type of a local variable in the source program (§4.7.9.1).

index

The value of the index item must be a valid index into the local variable array of the current frame. The given local variable is at index in the local variable array of the current frame.

If the given local variable is of type double or long, it occupies both index and index + 1.

4.7.15. The Deprecated Attribute

The Deprecated attribute is an optional fixed-length attribute in the attributes table of a ClassFilefield_info, or method_info structure (§4.1§4.5§4.6). A class, interface, method, or field may be marked using a Deprecated attribute to indicate that the class, interface, method, or field has been superseded.

A run-time interpreter or tool that reads the class file format, such as a compiler, can use this marking to advise the user that a superseded class, interface, method, or field is being referred to. The presence of a Deprecated attribute does not alter the semantics of a class or interface.

The Deprecated attribute has the following format:

Deprecated_attribute {
u2 attribute_name_index;
u4 attribute_length;
}

The items of the Deprecated_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "Deprecated".

attribute_length

The value of the attribute_length item must be zero.

4.7.16. The RuntimeVisibleAnnotations Attribute

The RuntimeVisibleAnnotations attribute is a variable-length attribute in the attributes table of a ClassFilefield_info, or method_info structure (§4.1§4.5§4.6). TheRuntimeVisibleAnnotations attribute records run-time visible annotations on the declaration of the corresponding class, field, or method.

There may be at most one RuntimeVisibleAnnotations attribute in the attributes table of a ClassFilefield_info, or method_info structure.

The RuntimeVisibleAnnotations attribute has the following format:

RuntimeVisibleAnnotations_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 num_annotations;
annotation annotations[num_annotations];
}

The items of the RuntimeVisibleAnnotations_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "RuntimeVisibleAnnotations".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

num_annotations

The value of the num_annotations item gives the number of run-time visible annotations represented by the structure.

annotations[]

Each entry in the annotations table represents a single run-time visible annotation on a declaration. The annotation structure has the following format:

annotation {
u2 type_index;
u2 num_element_value_pairs;
{ u2 element_name_index;
element_value value;
} element_value_pairs[num_element_value_pairs];
}

The items of the annotation structure are as follows:

type_index

The value of the type_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure (§4.4.7) representing a field descriptor (§4.3.2). The field descriptor denotes the type of the annotation represented by this annotation structure.

num_element_value_pairs

The value of the num_element_value_pairs item gives the number of element-value pairs of the annotation represented by this annotation structure.

element_value_pairs[]

Each value of the element_value_pairs table represents a single element-value pair in the annotation represented by this annotation structure. Each element_value_pairs entry contains the following two items:

element_name_index

The value of the element_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be aCONSTANT_Utf8_info structure (§4.4.7). The constant_pool entry denotes the name of the element of the element-value pair represented by this element_value_pairs entry.

In other words, the entry denotes an element of the annotation type specified by type_index.

value

The value of the value item represents the value of the element-value pair represented by this element_value_pairs entry.

4.7.16.1. The element_value structure

The element_value structure is a discriminated union representing the value of an element-value pair. It has the following format:

element_value {
u1 tag;
union {
u2 const_value_index; { u2 type_name_index;
u2 const_name_index;
} enum_const_value; u2 class_info_index; annotation annotation_value; { u2 num_values;
element_value values[num_values];
} array_value;
} value;
}

The tag item uses a single ASCII character to indicate the type of the value of the element-value pair. This determines which item of the value union is in use. Table 4.7.16.1-A shows the valid characters for the tag item, the type indicated by each character, and the item used in the value union for each character. The table's fourth column is used in the description below of one item of the value union.

Table 4.7.16.1-A. Interpretation of tag values as types

tag Item Type value Item Constant Type
B byte const_value_index CONSTANT_Integer
C char const_value_index CONSTANT_Integer
D double const_value_index CONSTANT_Double
F float const_value_index CONSTANT_Float
I int const_value_index CONSTANT_Integer
J long const_value_index CONSTANT_Long
S short const_value_index CONSTANT_Integer
Z boolean const_value_index CONSTANT_Integer
s String const_value_index CONSTANT_Utf8
e Enum type enum_const_value Not applicable
c Class class_info_index Not applicable
@ Annotation type annotation_value Not applicable
[ Array type array_value Not applicable

The value item represents the value of an element-value pair. The item is a union, whose own items are as follows:

const_value_index

The const_value_index item denotes either a primitive constant value or a String literal as the value of this element-value pair.

The value of the const_value_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be of a type appropriate to the tag item, as specified in the fourth column of Table 4.7.16.1-A.

enum_const_value

The enum_const_value item denotes an enum constant as the value of this element-value pair.

The enum_const_value item consists of the following two items:

type_name_index

The value of the type_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing a field descriptor (§4.3.2). The constant_pool entry gives the internal form of the binary name of the type of the enum constant represented by this element_value structure (§4.2.1).

const_name_index

The value of the const_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7). The constant_pool entry gives the simple name of the enum constant represented by this element_value structure.

class_info_index

The class_info_index item denotes a class literal as the value of this element-value pair.

The class_info_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_info structure (§4.4.7) representing a return descriptor (§4.3.3). The return descriptor gives the type corresponding to the class literal represented by this element_value structure. Types correspond to class literals as follows:

  • For a class literal C.class, where C is the name of a class, interface, or array type, the corresponding type is C. The return descriptor in the constant_pool will be an ObjectType or an ArrayType.

  • For a class literal p.class, where p is the name of a primitive type, the corresponding type is p. The return descriptor in the constant_pool will be a BaseType character.

  • For a class literal void.class, the corresponding type is void. The return descriptor in the constant_pool will be V.

For example, the class literal Object.class corresponds to the type Object, so the constant_pool entry is Ljava/lang/Object;, whereas the class literal int.classcorresponds to the type int, so the constant_pool entry is I.

The class literal void.class corresponds to void, so the constant_pool entry is V, whereas the class literal Void.class corresponds to the type Void, so the constant_pool entry is Ljava/lang/Void;.

annotation_value

The annotation_value item denotes a "nested" annotation as the value of this element-value pair.

The value of the annotation_value item is an annotation structure (§4.7.16) that gives the annotation represented by this element_value structure.

array_value

The array_value item denotes an array as the value of this element-value pair.

The array_value item consists of the following two items:

num_values

The value of the num_values item gives the number of elements in the array represented by this element_value structure.

values[]

Each value in the values table gives the corresponding element of the array represented by this element_value structure.

4.7.17. The RuntimeInvisibleAnnotations Attribute

The RuntimeInvisibleAnnotations attribute is a variable-length attribute in the attributes table of a ClassFilefield_info, or method_info structure (§4.1§4.5§4.6). TheRuntimeInvisibleAnnotations attribute records run-time invisible annotations on the declaration of the corresponding class, method, or field.

There may be at most one RuntimeInvisibleAnnotations attribute in the attributes table of a ClassFilefield_info, or method_info structure.

The RuntimeInvisibleAnnotations attribute is similar to the RuntimeVisibleAnnotations attribute (§4.7.16), except that the annotations represented by a RuntimeInvisibleAnnotations attribute must not be made available for return by reflective APIs, unless the Java Virtual Machine has been instructed to retain these annotations via some implementation-specific mechanism such as a command line flag. In the absence of such instructions, the Java Virtual Machine ignores this attribute.

The RuntimeInvisibleAnnotations attribute has the following format:

RuntimeInvisibleAnnotations_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 num_annotations;
annotation annotations[num_annotations];
}

The items of the RuntimeInvisibleAnnotations_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "RuntimeInvisibleAnnotations".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

num_annotations

The value of the num_annotations item gives the number of run-time invisible annotations represented by the structure.

annotations[]

Each entry in the annotations table represents a single run-time invisible annotation on a declaration. The annotation structure is specified in §4.7.16.

4.7.18. The RuntimeVisibleParameterAnnotations Attribute

The RuntimeVisibleParameterAnnotations attribute is a variable-length attribute in the attributes table of the method_info structure (§4.6). TheRuntimeVisibleParameterAnnotations attribute records run-time visible annotations on the declarations of formal parameters of the corresponding method.

There may be at most one RuntimeVisibleParameterAnnotations attribute in the attributes table of a method_info structure.

The RuntimeVisibleParameterAnnotations attribute has the following format:

RuntimeVisibleParameterAnnotations_attribute {
u2 attribute_name_index;
u4 attribute_length;
u1 num_parameters;
{ u2 num_annotations;
annotation annotations[num_annotations];
} parameter_annotations[num_parameters];
}

The items of the RuntimeVisibleParameterAnnotations_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "RuntimeVisibleParameterAnnotations".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

num_parameters

The value of the num_parameters item gives the number of run-time visible parameter annotations represented by this structure.

There is no assurance that this number is the same as the number of parameter descriptors in the method descriptor.

parameter_annotations[]

Each entry in the parameter_annotations table represents all of the run-time visible annotations on the declaration of a single formal parameter. Each parameter_annotationsentry contains the following two items:

num_annotations

The value of the num_annotations item indicates the number of run-time visible annotations on the declaration of the formal parameter corresponding to the parameter_annotations entry.

annotations[]

Each entry in the annotations table represents a single run-time visible annotation on the declaration of the formal parameter corresponding to the parameter_annotationsentry. The annotation structure is specified in §4.7.16.

The i'th entry in the parameter_annotations table may, but is not required to, correspond to the i'th parameter descriptor in the method descriptor (§4.3.3).

For example, a compiler may choose to create entries in the table corresponding only to those parameter descriptors which represent explicitly declared parameters in source code. In the Java programming language, a constructor of an inner class is specified to have an implicitly declared parameter before its explicitly declared parameters (JLS §8.8.1), so the corresponding <init> method in a class file has a parameter descriptor representing the implicitly declared parameter before any parameter descriptors representing explicitly declared parameters. If the first explicitly declared parameter is annotated in source code, then a compiler may create parameter_annotations[0] to store annotations corresponding to the second parameter descriptor.

4.7.19. The RuntimeInvisibleParameterAnnotations Attribute

The RuntimeInvisibleParameterAnnotations attribute is a variable-length attribute in the attributes table of a method_info structure (§4.6). TheRuntimeInvisibleParameterAnnotations attribute records run-time invisible annotations on the declarations of formal parameters of the corresponding method.

There may be at most one RuntimeInvisibleParameterAnnotations attribute in the attributes table of a method_info structure.

The RuntimeInvisibleParameterAnnotations attribute is similar to the RuntimeVisibleParameterAnnotations attribute (§4.7.18), except that the annotations represented by a RuntimeInvisibleParameterAnnotations attribute must not be made available for return by reflective APIs, unless the Java Virtual Machine has specifically been instructed to retain these annotations via some implementation-specific mechanism such as a command line flag. In the absence of such instructions, the Java Virtual Machine ignores this attribute.

The RuntimeInvisibleParameterAnnotations attribute has the following format:

RuntimeInvisibleParameterAnnotations_attribute {
u2 attribute_name_index;
u4 attribute_length;
u1 num_parameters;
{ u2 num_annotations;
annotation annotations[num_annotations];
} parameter_annotations[num_parameters];
}

The items of the RuntimeInvisibleParameterAnnotations_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "RuntimeInvisibleParameterAnnotations".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

num_parameters

The value of the num_parameters item gives the number of run-time invisible parameter annotations represented by this structure.

There is no assurance that this number is the same as the number of parameter descriptors in the method descriptor.

parameter_annotations[]

Each entry in the parameter_annotations table represents all of the run-time invisible annotations on the declaration of a single formal parameter. Each parameter_annotationsentry contains the following two items:

num_annotations

The value of the num_annotations item indicates the number of run-time invisible annotations on the declaration of the formal parameter corresponding to the parameter_annotations entry.

annotations[]

Each entry in the annotations table represents a single run-time invisible annotation on the declaration of the formal parameter corresponding to the parameter_annotationsentry. The annotation structure is specified in §4.7.16.

The i'th entry in the parameter_annotations table may, but is not required to, correspond to the i'th parameter descriptor in the method descriptor (§4.3.3).

See the note in §4.7.18 for an example of when parameter_annotations[0] does not correspond to the first parameter descriptor in the method descriptor.

4.7.20. The RuntimeVisibleTypeAnnotations Attribute

The RuntimeVisibleTypeAnnotations attribute is an variable-length attribute in the attributes table of a ClassFilefield_info, or method_info structure, or Code attribute (§4.1§4.5§4.6§4.7.3). The RuntimeVisibleTypeAnnotations attribute records run-time visible annotations on types used in the declaration of the corresponding class, field, or method, or in an expression in the corresponding method body. The RuntimeVisibleTypeAnnotations attribute also records run-time visible annotations on type parameter declarations of generic classes, interfaces, methods, and constructors.

There may be at most one RuntimeVisibleTypeAnnotations attribute in the attributes table of a ClassFilefield_info, or method_info structure, or Code attribute.

An attributes table contains a RuntimeVisibleTypeAnnotations attribute only if types are annotated in kinds of declaration or expression that correspond to the parent structure or attribute of the attributes table.

For example, all annotations on types in the implements clause of a class declaration are recorded in the RuntimeVisibleTypeAnnotations attribute of the class's ClassFile structure. Meanwhile, all annotations on the type in a field declaration are recorded in the RuntimeVisibleTypeAnnotations attribute of the field's field_info structure.

The RuntimeVisibleTypeAnnotations attribute has the following format:

RuntimeVisibleTypeAnnotations_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 num_annotations;
type_annotation annotations[num_annotations];
}

The items of the RuntimeVisibleTypeAnnotations_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure representing the string "RuntimeVisibleTypeAnnotations".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

num_annotations

The value of the num_annotations item gives the number of run-time visible type annotations represented by the structure.

annotations[]

Each entry in the annotations table represents a single run-time visible annotation on a type used in a declaration or expression. The type_annotation structure has the following format:

type_annotation {
u1 target_type;
union {
type_parameter_target;
supertype_target;
type_parameter_bound_target;
empty_target;
formal_parameter_target;
throws_target;
localvar_target;
catch_target;
offset_target;
type_argument_target;
} target_info;
type_path target_path;
u2 type_index;
u2 num_element_value_pairs;
{ u2 element_name_index;
element_value value;
} element_value_pairs[num_element_value_pairs];
}

The first three items - target_typetarget_info, and target_path - specify the precise location of the annotated type. The last three items - type_index,num_element_value_pairs, and element_value_pairs[] - specify the annotation's own type and element-value pairs.

The items of the type_annotation structure are as follows:

target_type

The value of the target_type item denotes the kind of target on which the annotation appears. The various kinds of target correspond to the type contexts of the Java programming language where types are used in declarations and expressions (JLS §4.11).

The legal values of target_type are specified in Table 4.7.20-A and Table 4.7.20-B. Each value is a one-byte tag indicating which item of the target_info union follows the target_type item to give more information about the target.

The kinds of target in Table 4.7.20-A and Table 4.7.20-B correspond to the type contexts in JLS §4.11. Namely, target_type values 0x10-0x17 and 0x40-0x42 correspond to type contexts 1-10, while target_type values 0x43-0x4B correspond to type contexts 11-16.

The value of the target_type item determines whether the type_annotation structure appears in a RuntimeVisibleTypeAnnotations attribute in a ClassFile structure, a field_info structure, a method_info structure, or a Code attribute. Table 4.7.20-C gives the location of the RuntimeVisibleTypeAnnotations attribute for a type_annotation structure with each legal target_type value.

target_info

The value of the target_info item denotes precisely which type in a declaration or expression is annotated.

The items of the target_info union are specified in §4.7.20.1.

target_path

The value of the target_path item denotes precisely which part of the type indicated by target_info is annotated.

The format of the type_path structure is specified in §4.7.20.2.

type_index, num_element_value_pairs, element_value_pairs[]

The meaning of these items in the type_annotation structure is the same as their meaning in the annotation structure (§4.7.16).

Table 4.7.20-A. Interpretation of target_type values (Part 1)

Value Kind of target target_info item
0x00 type parameter declaration of generic class or interface type_parameter_target
0x01 type parameter declaration of generic method or constructor type_parameter_target
0x10 type in extends or implements clause of class declaration (including the direct superclass or direct superinterface of an anonymous class declaration), or in extends clause of interface declaration supertype_target
0x11 type in bound of type parameter declaration of generic class or interface type_parameter_bound_target
0x12 type in bound of type parameter declaration of generic method or constructor type_parameter_bound_target
0x13 type in field declaration empty_target
0x14 return type of method, or type of newly constructed object empty_target
0x15 receiver type of method or constructor empty_target
0x16 type in formal parameter declaration of method, constructor, or lambda expression formal_parameter_target
0x17 type in throws clause of method or constructor throws_target

Table 4.7.20-B. Interpretation of target_type values (Part 2)

Value Kind of target target_info item
0x40 type in local variable declaration localvar_target
0x41 type in resource variable declaration localvar_target
0x42 type in exception parameter declaration catch_target
0x43 type in instanceof expression offset_target
0x44 type in new expression offset_target
0x45 type in method reference expression using ::new offset_target
0x46 type in method reference expression using ::Identifier offset_target
0x47 type in cast expression type_argument_target
0x48 type argument for generic constructor in new expression or explicit constructor invocation statement type_argument_target
0x49 type argument for generic method in method invocation expression type_argument_target
0x4A type argument for generic constructor in method reference expression using ::new type_argument_target
0x4B type argument for generic method in method reference expression using ::Identifier type_argument_target

Table 4.7.20-C. Location of enclosing attribute for target_type values

Value Kind of target Location
0x00 type parameter declaration of generic class or interface ClassFile
0x01 type parameter declaration of generic method or constructor method_info
0x10 type in extends clause of class or interface declaration, or in implements clause of interface declaration ClassFile
0x11 type in bound of type parameter declaration of generic class or interface ClassFile
0x12 type in bound of type parameter declaration of generic method or constructor method_info
0x13 type in field declaration field_info
0x14 return type of method or constructor method_info
0x15 receiver type of method or constructor method_info
0x16 type in formal parameter declaration of method, constructor, or lambda expression method_info
0x17 type in throws clause of method or constructor method_info
0x40-0x4B types in local variable declarations, resource variable declarations, exception parameter declarations, expressions Code

4.7.20.1. The target_info union

The items of the target_info union (except for the first) specify precisely which type in a declaration or expression is annotated. The first item specifies not which type, but rather which declaration of a type parameter is annotated. The items are as follows:

  • The type_parameter_target item indicates that an annotation appears on the declaration of the i'th type parameter of a generic class, generic interface, generic method, or generic constructor.

    type_parameter_target {
    u1 type_parameter_index;
    }

    The value of the type_parameter_index item specifies which type parameter declaration is annotated. A type_parameter_index value of 0 specifies the first type parameter declaration.

  • The supertype_target item indicates that an annotation appears on a type in the extends or implements clause of a class or interface declaration.

    supertype_target {
    u2 supertype_index;
    }

    supertype_index value of 65535 specifies that the annotation appears on the superclass in an extends clause of a class declaration.

    Any other supertype_index value is an index into the interfaces array of the enclosing ClassFile structure, and specifies that the annotation appears on that superinterface in either the implements clause of a class declaration or the extends clause of an interface declaration.

  • The type_parameter_bound_target item indicates that an annotation appears on the i'th bound of the j'th type parameter declaration of a generic class, interface, method, or constructor.

    type_parameter_bound_target {
    u1 type_parameter_index;
    u1 bound_index;
    }

    The value of the of type_parameter_index item specifies which type parameter declaration has an annotated bound. A type_parameter_index value of 0 specifies the first type parameter declaration.

    The value of the bound_index item specifies which bound of the type parameter declaration indicated by type_parameter_index is annotated. A bound_index value of 0 specifies the first bound of a type parameter declaration.

    The type_parameter_bound_target item records that a bound is annotated, but does not record the type which constitutes the bound. The type may be found by inspecting the class signature or method signature stored in the appropriate Signature attribute.

  • The empty_target item indicates that an annotation appears on either the type in a field declaration, the return type of a method, the type of a newly constructed object, or the receiver type of a method or constructor.

    empty_target {
    }

    Only one type appears in each of these locations, so there is no per-type information to represent in the target_info union.

  • The formal_parameter_target item indicates that an annotation appears on the type in a formal parameter declaration of a method, constructor, or lambda expression.

    formal_parameter_target {
    u1 formal_parameter_index;
    }

    The value of the formal_parameter_index item specifies which formal parameter declaration has an annotated type. A formal_parameter_index value of i may, but is not required to, correspond to the i'th parameter descriptor in the method descriptor (§4.3.3).

    The formal_parameter_target item records that a formal parameter's type is annotated, but does not record the type itself. The type may be found by inspecting the method descriptor, although a formal_parameter_index value of 0 does not always indicate the first parameter descriptor in the method descriptor; see the note in §4.7.18 for a similar situation involving the parameter_annotations table.

  • The throws_target item indicates that an annotation appears on the i'th type in the throws clause of a method or constructor declaration.

    throws_target {
    u2 throws_type_index;
    }

    The value of the throws_type_index item is an index into the exception_index_table array of the Exceptions attribute of the method_info structure enclosing the RuntimeVisibleTypeAnnotations attribute.

  • The localvar_target item indicates that an annotation appears on the type in a local variable declaration, including a variable declared as a resource in a try-with-resources statement.

    localvar_target {
    u2 table_length;
    { u2 start_pc;
    u2 length;
    u2 index;
    } table[table_length];
    }

    The value of the table_length item gives the number of entries in the table array. Each entry indicates a range of code array offsets within which a local variable has a value. It also indicates the index into the local variable array of the current frame at which that local variable can be found. Each entry contains the following three items:

    start_pc, length

    The given local variable has a value at indices into the code array in the interval [start_pcstart_pc + length), that is, between start_pc inclusive and start_pc + lengthexclusive.

    index

    The given local variable must be at index in the local variable array of the current frame.

    If the local variable at index is of type double or long, it occupies both index and index + 1.

    A table is needed to fully specify the local variable whose type is annotated, because a single local variable may be represented with different local variable indices over multiple live ranges. The start_pclength, and index items in each table entry specify the same information as a LocalVariableTable attribute.

    The localvar_target item records that a local variable's type is annotated, but does not record the type itself. The type may be found by inspecting the appropriate LocalVariableTableattribute.

  • The catch_target item indicates that an annotation appears on the i'th type in an exception parameter declaration.

    catch_target {
    u2 exception_table_index;
    }

    The value of the exception_table_index item is an index into the exception_table array of the Code attribute enclosing the RuntimeVisibleTypeAnnotations attribute.

    The possibility of more than one type in an exception parameter declaration arises from the multi-catch clause of the try statement, where the type of the exception parameter is a union of types (JLS §14.20). A compiler usually creates one exception_table entry for each type in the union, which allows the catch_target item to distinguish them. This preserves the correspondence between a type and its annotations.

  • The offset_target item indicates that an annotation appears on either the type in an instanceof expression or a new expression, or the type before the :: in a method reference expression.

    offset_target {
    u2 offset;
    }

    The value of the offset item specifies the code array offset of either the bytecode instruction corresponding to the instanceof expression, the new bytecode instruction corresponding to the new expression, or the bytecode instruction corresponding to the method reference expression.

  • The type_argument_target item indicates that an annotation appears either on the i'th type in a cast expression, or on the i'th type argument in the explicit type argument list for any of the following: a new expression, an explicit constructor invocation statement, a method invocation expression, or a method reference expression.

    type_argument_target {
    u2 offset;
    u1 type_argument_index;
    }

    The value of the offset item specifies the code array offset of either the bytecode instruction corresponding to the cast expression, the new bytecode instruction corresponding to the newexpression, the bytecode instruction corresponding to the explicit constructor invocation statement, the bytecode instruction corresponding to the method invocation expression, or the bytecode instruction corresponding to the method reference expression.

    For a cast expression, the value of the type_argument_index item specifies which type in the cast operator is annotated. A type_argument_index value of 0 specifies the first (or only) type in the cast operator.

    The possibility of more than one type in a cast expression arises from a cast to an intersection type.

    For an explicit type argument list, the value of the type_argument_index item specifies which type argument is annotated. A type_argument_index value of 0 specifies the first type argument.

4.7.20.2. The type_path structure

Wherever a type is used in a declaration or expression, the type_path structure identifies which part of the type is annotated. An annotation may appear on the type itself, but if the type is a reference type, then there are additional locations where an annotation may appear:

  • If an array type T[] is used in a declaration or expression, then an annotation may appear on any component type of the array type, including the element type.

  • If a nested type T1.T2 is used in a declaration or expression, then an annotation may appear on the name of the top level type or any member type.

  • If a parameterized type T<A> or T<? extends A> or T<? super A> is used in a declaration or expression, then an annotation may appear on any type argument or on the bound of any wildcard type argument.

For example, consider the different parts of String[][] that are annotated in:

@Foo String[][]   // Annotates the class type String
String @Foo [][] // Annotates the array type String[][]
String[] @Foo [] // Annotates the array type String[]

or the different parts of the nested type Outer.Middle.Inner that are annotated in:

@Foo Outer.Middle.Inner
Outer.@Foo Middle.Inner
Outer.Middle.@Foo Inner

or the different parts of the parameterized types Map<String,Object> and List<...> that are annotated in:

@Foo Map<String,Object>
Map<@Foo String,Object>
Map<String,@Foo Object> List<@Foo ? extends String>
List<? extends @Foo String>

The type_path structure has the following format:

type_path {
u1 path_length;
{ u1 type_path_kind;
u1 type_argument_index;
} path[path_length];
}

The value of the path_length item gives the number of entries in the path array:

  • If the value of path_length is 0, then the annotation appears directly on the type itself.

  • If the value of path_length is non-zero, then each entry in the path array represents an iterative, left-to-right step towards the precise location of the annotation in an array type, nested type, or parameterized type. (In an array type, the iteration visits the array type itself, then its component type, then the component type of that component type, and so on, until the element type is reached.) Each entry contains the following two items:

    type_path_kind

    The legal values for the type_path_kind item are listed in Table 4.7.20.2-A.

    Table 4.7.20.2-A. Interpretation of type_path_kind values

    Value Interpretation
    0 Annotation is deeper in an array type
    1 Annotation is deeper in a nested type
    2 Annotation is on the bound of a wildcard type argument of a parameterized type
    3 Annotation is on a type argument of a parameterized type
    type_argument_index

    If the value of the type_path_kind item is 01, or 2, then the value of the type_argument_index item is 0.

    If the value of the type_path_kind item is 3, then the value of the type_argument_index item specifies which type argument of a parameterized type is annotated, where 0indicates the first type argument of a parameterized type.

Table 4.7.20.2-B. type_path structures for @A Map<@B ? extends @C String, @D List<@E Object>>

Annotation path_length path
@A 0 []
@B 1 [{type_path_kind: 3; type_argument_index: 0}]
@C 2 [{type_path_kind: 3; type_argument_index: 0}, {type_path_kind: 2; type_argument_index: 0}]
@D 1 [{type_path_kind: 3; type_argument_index: 1}]
@E 2 [{type_path_kind: 3; type_argument_index: 1}, {type_path_kind: 3; type_argument_index: 0}]

Table 4.7.20.2-C. type_path structures for @I String @F [] @G [] @H []

Annotation path_length path
@F 0 []
@G 1 [{type_path_kind: 0; type_argument_index: 0}]
@H 2 [{type_path_kind: 0; type_argument_index: 0}, {type_path_kind: 0; type_argument_index: 0}]
@I 3 [{type_path_kind: 0; type_argument_index: 0}, {type_path_kind: 0; type_argument_index: 0}, {type_path_kind: 0; type_argument_index: 0}]

Table 4.7.20.2-D. type_path structures for @A List<@B Comparable<@F Object @C [] @D [] @E []>>

Annotation path_length path
@A 0 []
@B 1 [{type_path_kind: 3; type_argument_index: 0}]
@C 2 [{type_path_kind: 3; type_argument_index: 0}, {type_path_kind: 3; type_argument_index: 0}]
@D 3 [{type_path_kind: 3; type_argument_index: 0}, {type_path_kind: 3; type_argument_index: 0}, {type_path_kind: 0; type_argument_index: 0}]
@E 4 [{type_path_kind: 3; type_argument_index: 0}, {type_path_kind: 3; type_argument_index: 0}, {type_path_kind: 0; type_argument_index: 0}, {type_path_kind: 0; type_argument_index: 0}]
@F 5 [{type_path_kind: 3; type_argument_index: 0}, {type_path_kind: 3; type_argument_index: 0}, {type_path_kind: 0; type_argument_index: 0}, {type_path_kind: 0; type_argument_index: 0}, {type_path_kind: 0; type_argument_index: 0}]

Table 4.7.20.2-E. type_path structures for @C Outer . @B Middle . @A Inner

Annotation path_length path
@A 2 [{type_path_kind: 1; type_argument_index: 0}, {type_path_kind: 1; type_argument_index: 0}]
@B 1 [{type_path_kind: 1; type_argument_index: 0}]
@C 0 []

Table 4.7.20.2-F. type_path structures for Outer . Middle<@D Foo . @C Bar> . Inner<@B String @A []>

Annotation path_length path
@A 3 [{type_path_kind: 1; type_argument_index: 0}, {type_path_kind: 1; type_argument_index: 0}, {type_path_kind: 3; type_argument_index: 0}]
@B 4 [{type_path_kind: 1; type_argument_index: 0}, {type_path_kind: 1; type_argument_index: 0}, {type_path_kind: 3; type_argument_index: 0}, {type_path_kind: 0; type_argument_index: 0}]
@C 3 [{type_path_kind: 1; type_argument_index: 0}, {type_path_kind: 3; type_argument_index: 0}, {type_path_kind: 1; type_argument_index: 0}]
@D 2 [{type_path_kind: 1; type_argument_index: 0}, {type_path_kind: 3; type_argument_index: 0}]

4.7.21. The RuntimeInvisibleTypeAnnotations Attribute

The RuntimeInvisibleTypeAnnotations attribute is an variable-length attribute in the attributes table of a ClassFilefield_info, or method_info structure, or Code attribute (§4.1§4.5§4.6§4.7.3). The RuntimeInvisibleTypeAnnotations attribute records run-time invisible annotations on types used in the corresponding declaration of a class, field, or method, or in an expression in the corresponding method body. The RuntimeInvisibleTypeAnnotations attribute also records annotations on type parameter declarations of generic classes, interfaces, methods, and constructors.

There may be at most one RuntimeInvisibleTypeAnnotations attribute in the attributes table of a ClassFilefield_info, or method_info structure, or Code attribute.

An attributes table contains a RuntimeInvisibleTypeAnnotations attribute only if types are annotated in kinds of declaration or expression that correspond to the parent structure or attribute of the attributes table.

The RuntimeInvisibleTypeAnnotations attribute has the following format:

RuntimeInvisibleTypeAnnotations_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 num_annotations;
type_annotation annotations[num_annotations];
}

The items of the RuntimeInvisibleTypeAnnotations_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure representing the string "RuntimeInvisibleTypeAnnotations".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

num_annotations

The value of the num_annotations item gives the number of run-time invisible type annotations represented by the structure.

annotations[]

Each entry in the annotations table represents a single run-time invisible annotation on a type used in a declaration or expression. The type_annotation structure is specified in§4.7.20.

4.7.22. The AnnotationDefault Attribute

The AnnotationDefault attribute is a variable-length attribute in the attributes table of certain method_info structures (§4.6), namely those representing elements of annotation types (JLS §9.6.1). The AnnotationDefault attribute records the default value (JLS §9.6.2) for the element represented by the method_info structure.

There may be at most one AnnotationDefault attribute in the attributes table of a method_info structure which represents an element of an annotation type.

The AnnotationDefault attribute has the following format:

AnnotationDefault_attribute {
u2 attribute_name_index;
u4 attribute_length;
element_value default_value;
}

The items of the AnnotationDefault_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "AnnotationDefault".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

default_value

The default_value item represents the default value of the annotation type element represented by the method_info structure enclosing this AnnotationDefault attribute.

4.7.23. The BootstrapMethods Attribute

The BootstrapMethods attribute is a variable-length attribute in the attributes table of a ClassFile structure (§4.1). The BootstrapMethods attribute records bootstrap method specifiers referenced by invokedynamic instructions (§invokedynamic).

There must be exactly one BootstrapMethods attribute in the attributes table of a ClassFile structure if the constant_pool table of the ClassFile structure has at least oneCONSTANT_InvokeDynamic_info entry (§4.4.10).

There may be at most one BootstrapMethods attribute in the attributes table of a ClassFile structure.

The BootstrapMethods attribute has the following format:

BootstrapMethods_attribute {
u2 attribute_name_index;
u4 attribute_length;
u2 num_bootstrap_methods;
{ u2 bootstrap_method_ref;
u2 num_bootstrap_arguments;
u2 bootstrap_arguments[num_bootstrap_arguments];
} bootstrap_methods[num_bootstrap_methods];
}

The items of the BootstrapMethods_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure (§4.4.7) representing the string "BootstrapMethods".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

The value of the attribute_length item is thus dependent on the number of invokedynamic instructions in this ClassFile structure.

num_bootstrap_methods

The value of the num_bootstrap_methods item determines the number of bootstrap method specifiers in the bootstrap_methods array.

bootstrap_methods[]

Each entry in the bootstrap_methods table contains an index to a CONSTANT_MethodHandle_info structure which specifies a bootstrap method, and a sequence (perhaps empty) of indexes to static arguments for the bootstrap method.

Each bootstrap_methods entry must contain the following three items:

bootstrap_method_ref

The value of the bootstrap_method_ref item must be a valid index into the constant_pool table. The constant_pool entry at that index must be aCONSTANT_MethodHandle_info structure (§4.4.8).

The form of the method handle is driven by the continuing resolution of the call site specifier in §invokedynamic, where execution of the invoke method of java.lang.invoke.MethodHandle requires that the bootstrap method handle be adjustable to the actual arguments being passed, as if by invocation of the asType method of java.lang.invoke.MethodHandle. Accordingly, the reference_kind item of the CONSTANT_MethodHandle_info structure should have the value 6 or 8 (§5.4.3.5), and the reference_index item should specify a static method or constructor that takes three arguments of type java.lang.invoke.MethodHandles.LookupString, andjava.lang.invoke.MethodType, in that order. Otherwise, invocation of the bootstrap method handle during call site specifier resolution will complete abruptly.

num_bootstrap_arguments

The value of the num_bootstrap_arguments item gives the number of items in the bootstrap_arguments array.

bootstrap_arguments[]

Each entry in the bootstrap_arguments array must be a valid index into the constant_pool table. The constant_pool entry at that index must be aCONSTANT_String_infoCONSTANT_Class_infoCONSTANT_Integer_infoCONSTANT_Long_infoCONSTANT_Float_infoCONSTANT_Double_info,CONSTANT_MethodHandle_info, or CONSTANT_MethodType_info structure (§4.4.3§4.4.1§4.4.4§4.4.5§4.4.8§4.4.9).

4.7.24. The MethodParameters Attribute

The MethodParameters attribute is a variable-length attribute in the attributes table of a method_info structure (§4.6). A MethodParameters attribute records information about the formal parameters of a method, such as their names.

There may be at most one MethodParameters attribute in the attributes table of a method_info structure.

The MethodParameters attribute has the following format:

MethodParameters_attribute {
u2 attribute_name_index;
u4 attribute_length;
u1 parameters_count;
{ u2 name_index;
u2 access_flags;
} parameters[parameters_count];
}

The items of the MethodParameters_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure representing the string "MethodParameters".

attribute_length

The value of the attribute_length item indicates the length of the attribute, excluding the initial six bytes.

parameters_count

The value of the parameters_count item indicates the number of parameter descriptors in the method descriptor (§4.3.3) referenced by the descriptor_index of the attribute's enclosing method_info structure.

This is not a constraint which a Java Virtual Machine implementation must enforce during format checking (§4.8). The task of matching parameter descriptors in a method descriptor against the items in the parameters array below is done by the reflection libraries of the Java SE Platform.

parameters[]

Each entry in the parameters array contains the following pair of items:

name_index

The value of the name_index item must either be zero or a valid index into the constant_pool table.

If the value of the name_index item is zero, then this parameters element indicates a formal parameter with no name.

If the value of the name_index item is nonzero, the constant_pool entry at that index must be a CONSTANT_Utf8_info structure representing a valid unqualified name denoting a formal parameter (§4.2.2).

access_flags

The value of the access_flags item is as follows:

0x0010 (ACC_FINAL)

Indicates that the formal parameter was declared final.

0x1000 (ACC_SYNTHETIC)

Indicates that the formal parameter was not explicitly or implicitly declared in source code, according to the specification of the language in which the source code was written (JLS §13.1). (The formal parameter is an implementation artifact of the compiler which produced this class file.)

0x8000 (ACC_MANDATED)

Indicates that the formal parameter was implicitly declared in source code, according to the specification of the language in which the source code was written (JLS §13.1). (The formal parameter is mandated by a language specification, so all compilers for the language must emit it.)

The i'th entry in the parameters array corresponds to the i'th parameter descriptor in the enclosing method's descriptor. (The parameters_count item is one byte because a method descriptor is limited to 255 parameters.) Effectively, this means the parameters array stores information for all the parameters of the method. One could imagine other schemes, where entries in the parameters array specify their corresponding parameter descriptors, but it would unduly complicate the MethodParameters attribute.

The i'th entry in the parameters array may or may not correspond to the i'th type in the enclosing method's Signature attribute (if present), or to the i'th annotation in the enclosing method's parameter annotations.

4.7.25. The Module Attribute

The Module attribute is a variable-length attribute in the attributes table of a ClassFile structure. The Module attribute indicates the modules required by a module; the packages exported and opened by a module; and the services used and provided by a module.

There may be at most one Module attribute in the attributes table of a ClassFile structure.

The Module attribute has the following format:

Module_attribute {
u2 attribute_name_index;
u4 attribute_length; u2 module_name_index;
u2 module_flags;
u2 module_version_index; u2 requires_count;
{ u2 requires_index;
u2 requires_flags;
u2 requires_version_index;
} requires[requires_count]; u2 exports_count;
{ u2 exports_index;
u2 exports_flags;
u2 exports_to_count;
u2 exports_to_index[exports_to_count];
} exports[exports_count]; u2 opens_count;
{ u2 opens_index;
u2 opens_flags;
u2 opens_to_count;
u2 opens_to_index[opens_to_count];
} opens[opens_count]; u2 uses_count;
u2 uses_index[uses_count]; u2 provides_count;
{ u2 provides_index;
u2 provides_with_count;
u2 provides_with_index[provides_with_count];
} provides[provides_count];
}

The items of the Module_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure representing the string "Module".

attribute_length

The value of the attribute_length item is the length of the attribute excluding the initial six bytes.

module_name_index

The value of the module_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Module_info structure denoting the current module.

module_flags

The value of the module_flags item is as follows:

0x0020 (ACC_OPEN)

Indicates that this module is open.

0x1000 (ACC_SYNTHETIC)

Indicates that this module was not explicitly or implicitly declared.

0x8000 (ACC_MANDATED)

Indicates that this module was implicitly declared.

module_version_index

The value of the module_version_index item must be either zero or a valid index into the constant_pool table. If the value of the item is zero, then no version information about the current module is present. If the value of the item is nonzero, then the constant_pool entry at that index must be a CONSTANT_Utf8_info structure representing the version of the current module.

requires_count

The value of the requires_count item indicates the number of entries in the requires table.

If the current module is java.base, then requires_count must be zero.

If the current module is not java.base, then requires_count must be at least one.

requires[]

Each entry in the requires table specifies a dependence of the current module. The items in each entry are as follows:

requires_index

The value of the requires_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Module_infostructure denoting a module on which the current module depends.

At most one entry in the requires table may specify a module of a given name with its requires_index item.

requires_flags

The value of the requires_flags item is as follows:

0x0020 (ACC_TRANSITIVE)

Indicates that any module which depends on the current module, implicitly declares a dependence on the module indicated by this entry.

0x0040 (ACC_STATIC_PHASE)

Indicates that this dependence is mandatory in the static phase, i.e., at compile time, but is optional in the dynamic phase, i.e., at run time.

0x1000 (ACC_SYNTHETIC)

Indicates that this dependence was not explicitly or implicitly declared in the source of the module declaration.

0x8000 (ACC_MANDATED)

Indicates that this dependence was implicitly declared in the source of the module declaration.

requires_version_index

The value of the requires_version_index item must be either zero or a valid index into the constant_pool table. If the value of the item is zero, then no version information about the dependence is present. If the value of the item is nonzero, then the constant_pool entry at that index must be a CONSTANT_Utf8_info structure representing the version of the module specified by requires_index.

Unless the current module is java.base, exactly one entry in the requires table must have both a requires_index item which indicates java.base and a requires_flags item which has the ACC_SYNTHETIC flag not set.

exports_count

The value of the exports_count item indicates the number of entries in the exports table.

exports[]

Each entry in the exports table specifies a package exported by the current module, such that public and protected types in the package, and their public and protectedmembers, may be accessed from outside the current module, possibly from a limited set of "friend" modules.

The items in each entry are as follows:

exports_index

The value of the exports_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Package_infostructure representing a package exported by the current module.

At most one entry in the exports table may specify a package of a given name with its exports_index item.

exports_flags

The value of the exports_flags item is as follows:

0x1000 (ACC_SYNTHETIC)

Indicates that this export was not explicitly or implicitly declared in the source of the module declaration.

0x8000 (ACC_MANDATED)

Indicates that this export was implicitly declared in the source of the module declaration.

exports_to_count

The value of the exports_to_count indicates the number of entries in the exports_to_index table.

If exports_to_count is zero, then this package is exported by the current module in an unqualified fashion; code in any other module may access the types and members in the package.

If exports_to_count is nonzero, then this package is exported by the current module in a qualified fashion; only code in the modules listed in the exports_to_index table may access the types and members in the package.

exports_to_index[]

The value of each entry in the exports_to_index table must be a valid index into the constant_pool table. The constant_pool entry at that index must be aCONSTANT_Module_info structure denoting a module whose code can access the types and members in this exported package.

For each entry in the exports table, at most one entry in its exports_to_index table may specify a module of a given name.

opens_count

The value of the opens_count item indicates the number of entries in the opens table.

opens_count must be zero if the current module is open.

opens[]

Each entry in the opens table specifies a package opened by the current module, such that all types in the package, and all their members, may be accessed from outside the current module via the reflection libraries of the Java SE Platform, possibly from a limited set of "friend" modules.

The items in each entry are as follows:

opens_index

The value of the opens_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Package_infostructure representing a package opened by the current module.

At most one entry in the opens table may specify a package of a given name with its opens_index item.

opens_flags

The value of the opens_flags item is as follows:

0x1000 (ACC_SYNTHETIC)

Indicates that this opening was not explicitly or implicitly declared in the source of the module declaration.

0x8000 (ACC_MANDATED)

Indicates that this opening was implicitly declared in the source of the module declaration.

opens_to_count

The value of the opens_to_count indicates the number of entries in the opens_to_index table.

If opens_to_count is zero, then this package is opened by the current module in an unqualified fashion; code in any other module may reflectively access the types and members in the package.

If opens_to_count is nonzero, then this package is opened by the current module in a qualified fashion; only code in the modules listed in the exports_to_index table may reflectively access the types and members in the package.

opens_to_index[]

The value of each entry in the opens_to_index table must be a valid index into the constant_pool table. The constant_pool entry at that index must be aCONSTANT_Module_info structure denoting a module whose code can access the types and members in this opened package.

For each entry in the opens table, at most one entry in its opens_to_index table may specify a module of a given name.

uses_count

The value of the uses_count item indicates the number of entries in the uses_index table.

uses_index[]

The value of each entry in the uses_index table must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Class_infostructure representing a service interface which the current module may discover via java.util.ServiceLoader.

At most one entry in the uses_index table may specify a service interface of a given name.

provides_count

The value of the provides_count item indicates the number of entries in the provides table.

provides[]

Each entry in the provides table represents a service implementation for a given service interface.

The items in each entry are as follows:

provides_index

The value of the provides_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Class_infostructure representing a service interface for which the current module provides a service implementation.

At most one entry in the provides table may specify a service interface of a given name with its provides_index item.

provides_with_count

The value of the provides_with_count indicates the number of entries in the provides_with_index table.

provides_with_count must be nonzero.

provides_with_index[]

The value of each entry in the provides_with_index table must be a valid index into the constant_pool table. The constant_pool entry at that index must be aCONSTANT_Class_info structure representing a service implementation for the service interface specified by provides_index.

For each entry in the provides table, at most one entry in its provides_with_index table may specify a service implementation of a given name.

4.7.26. The ModulePackages Attribute

The ModulePackages attribute is a variable-length attribute in the attributes table of a ClassFile structure. The ModulePackages attribute indicates all the packages of a module that are exported or opened by the Module attribute, as well as all the packages of the service implementations recorded in the Module attribute. The ModulePackages attribute may also indicate packages in the module that are neither exported nor opened nor contain service implementations.

There may be at most one ModulePackages attribute in the attributes table of a ClassFile structure.

The ModulePackages attribute has the following format:

ModulePackages_attribute {
u2 attribute_name_index;
u4 attribute_length; u2 package_count;
u2 package_index[package_count];
}

The items of the ModulePackages_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure representing the string "ModulePackages".

attribute_length

The value of the attribute_length item is the length of the attribute excluding the initial six bytes.

package_count

The value of the package_count item indicates the number of entries in the package_index table.

package_index[]

The value of each entry in the package_index table must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Package_info structure representing a package in the current module.

At most one entry in the package_index table may specify a package of a given name.

4.7.27. The ModuleMainClass Attribute

The ModuleMainClass attribute is a fixed-length attribute in the attributes table of a ClassFile structure. The ModuleMainClass attribute indicates the main class of a module.

There may be at most one ModuleMainClass attribute in the attributes table of a ClassFile structure.

The ModuleMainClass attribute has the following format:

ModuleMainClass_attribute {
u2 attribute_name_index;
u4 attribute_length; u2 main_class_index;
}

The items of the ModuleMainClass_attribute structure are as follows:

attribute_name_index

The value of the attribute_name_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Utf8_infostructure representing the string "ModuleMainClass".

attribute_length

The value of the attribute_length item is the length of the attribute excluding the initial six bytes.

main_class_index

The value of the main_class_index item must be a valid index into the constant_pool table. The constant_pool entry at that index must be a CONSTANT_Class_info structure representing the binary name of the main class of the current module.

4.8. Format Checking

When a prospective class file is loaded by the Java Virtual Machine (§5.3), the Java Virtual Machine first ensures that the file has the basic format of a class file (§4.1). This process is known as format checking. The checks are as follows:

  • The first four bytes must contain the right magic number.

  • All predefined attributes (§4.7) must be of the proper length, except for StackMapTableRuntimeVisibleAnnotationsRuntimeInvisibleAnnotations,RuntimeVisibleParameterAnnotationsRuntimeInvisibleParameterAnnotationsRuntimeVisibleTypeAnnotationsRuntimeInvisibleTypeAnnotations, andAnnotationDefault.

  • The class file must not be truncated or have extra bytes at the end.

  • The constant pool must satisfy the constraints documented throughout §4.4.

    For example, each CONSTANT_Class_info structure in the constant pool must contain in its name_index item a valid constant pool index for a CONSTANT_Utf8_info structure.

  • All field references and method references in the constant pool must have valid names, valid classes, and valid descriptors (§4.3).

    Format checking does not ensure that the given field or method actually exists in the given class, nor that the descriptors given refer to real classes. Format checking ensures only that these items are well formed. More detailed checking is performed when the bytecodes themselves are verified, and during resolution.

These checks for basic class file integrity are necessary for any interpretation of the class file contents. Format checking is distinct from bytecode verification, although historically they have been confused because both are a form of integrity check.

4.9. Constraints on Java Virtual Machine Code

The code for a method, instance initialization method (§2.9.1), or class or interface initialization method (§2.9.2) is stored in the code array of the Code attribute of a method_info structure of a class file (§4.7.3). This section describes the constraints associated with the contents of the Code_attribute structure.

4.9.1. Static Constraints

The static constraints on a class file are those defining the well-formedness of the file. These constraints have been given in the previous sections, except for static constraints on the code in theclass file. The static constraints on the code in a class file specify how Java Virtual Machine instructions must be laid out in the code array and what the operands of individual instructions must be.

The static constraints on the instructions in the code array are as follows:

  • Only instances of the instructions documented in §6.5 may appear in the code array. Instances of instructions using the reserved opcodes (§6.2) or any opcodes not documented in this specification must not appear in the code array.

    If the class file version number is 51.0 or above, then neither the jsr opcode or the jsr_w opcode may appear in the code array.

  • The opcode of the first instruction in the code array begins at index 0.

  • For each instruction in the code array except the last, the index of the opcode of the next instruction equals the index of the opcode of the current instruction plus the length of that instruction, including all its operands.

    The wide instruction is treated like any other instruction for these purposes; the opcode specifying the operation that a wide instruction is to modify is treated as one of the operands of that wideinstruction. That opcode must never be directly reachable by the computation.

  • The last byte of the last instruction in the code array must be the byte at index code_length - 1.

The static constraints on the operands of instructions in the code array are as follows:

  • The target of each jump and branch instruction (jsrjsr_wgotogoto_wifeqifneifleifltifgeifgtifnullifnonnullif_icmpeqif_icmpneif_icmpleif_icmpltif_icmpgeif_icmpgtif_acmpeq,if_acmpne) must be the opcode of an instruction within this method.

    The target of a jump or branch instruction must never be the opcode used to specify the operation to be modified by a wide instruction; a jump or branch target may be the wide instruction itself.

  • Each target, including the default, of each tableswitch instruction must be the opcode of an instruction within this method.

    Each tableswitch instruction must have a number of entries in its jump table that is consistent with the value of its low and high jump table operands, and its low value must be less than or equal to its high value.

    No target of a tableswitch instruction may be the opcode used to specify the operation to be modified by a wide instruction; a tableswitch target may be a wide instruction itself.

  • Each target, including the default, of each lookupswitch instruction must be the opcode of an instruction within this method.

    Each lookupswitch instruction must have a number of match-offset pairs that is consistent with the value of its npairs operand. The match-offset pairs must be sorted in increasing numerical order by signed match value.

    No target of a lookupswitch instruction may be the opcode used to specify the operation to be modified by a wide instruction; a lookupswitch target may be a wide instruction itself.

  • The operand of each ldc instruction and each ldc_w instruction must be a valid index into the constant_pool table. The constant pool entry referenced by that index must be of type:

    • CONSTANT_IntegerCONSTANT_Float, or CONSTANT_String if the class file version number is less than 49.0.

    • CONSTANT_IntegerCONSTANT_FloatCONSTANT_String, or CONSTANT_Class if the class file version number is 49.0 or 50.0.

    • CONSTANT_IntegerCONSTANT_FloatCONSTANT_StringCONSTANT_ClassCONSTANT_MethodType, or CONSTANT_MethodHandle if the class file version number is 51.0 or above.

  • The operands of each ldc2_w instruction must represent a valid index into the constant_pool table. The constant pool entry referenced by that index must be of type CONSTANT_Long orCONSTANT_Double.

    The subsequent constant pool index must also be a valid index into the constant pool, and the constant pool entry at that index must not be used.

  • The operands of each getfieldputfieldgetstatic, and putstatic instruction must represent a valid index into the constant_pool table. The constant pool entry referenced by that index must be of type CONSTANT_Fieldref.

  • The indexbyte operands of each invokevirtual instruction must represent a valid index into the constant_pool table. The constant pool entry referenced by that index must be of type CONSTANT_Methodref.

  • The indexbyte operands of each invokespecial and invokestatic instruction must represent a valid index into the constant_pool table. If the class file version number is less than 52.0, the constant pool entry referenced by that index must be of type CONSTANT_Methodref; if the class file version number is 52.0 or above, the constant pool entry referenced by that index must be of type CONSTANT_Methodref or CONSTANT_InterfaceMethodref.

  • The indexbyte operands of each invokeinterface instruction must represent a valid index into the constant_pool table. The constant pool entry referenced by that index must be of type CONSTANT_InterfaceMethodref.

    The value of the count operand of each invokeinterface instruction must reflect the number of local variables necessary to store the arguments to be passed to the interface method, as implied by the descriptor of the CONSTANT_NameAndType_info structure referenced by the CONSTANT_InterfaceMethodref constant pool entry.

    The fourth operand byte of each invokeinterface instruction must have the value zero.

  • The indexbyte operands of each invokedynamic instruction must represent a valid index into the constant_pool table. The constant pool entry referenced by that index must be of type CONSTANT_InvokeDynamic.

    The third and fourth operand bytes of each invokedynamic instruction must have the value zero.

  • Only the invokespecial instruction is allowed to invoke an instance initialization method (§2.9.1).

    No other method whose name begins with the character '<' ('\u003c') may be called by the method invocation instructions. In particular, the class or interface initialization method specially named <clinit> is never called explicitly from Java Virtual Machine instructions, but only implicitly by the Java Virtual Machine itself.

  • The operands of each instanceofcheckcastnew, and anewarray instruction, and the indexbyte operands of each multianewarray instruction, must represent a valid index into the constant_pool table. The constant pool entry referenced by that index must be of type CONSTANT_Class.

  • No new instruction may reference a constant pool entry of type CONSTANT_Class that represents an array type (§4.3.2). The new instruction cannot be used to create an array.

  • No anewarray instruction may be used to create an array of more than 255 dimensions.

  • multianewarray instruction must be used only to create an array of a type that has at least as many dimensions as the value of its dimensions operand. That is, while a multianewarrayinstruction is not required to create all of the dimensions of the array type referenced by its indexbyte operands, it must not attempt to create more dimensions than are in the array type.

    The dimensions operand of each multianewarray instruction must not be zero.

  • The atype operand of each newarray instruction must take one of the values T_BOOLEAN (4), T_CHAR (5), T_FLOAT (6), T_DOUBLE (7), T_BYTE (8), T_SHORT (9), T_INT (10), or T_LONG (11).

  • The index operand of each iloadfloadaloadistorefstoreastoreiinc, and ret instruction must be a non-negative integer no greater than max_locals - 1.

    The implicit index of each iload_<n>fload_<n>aload_<n>istore_<n>fstore_<n>, and astore_<n> instruction must be no greater than max_locals - 1.

  • The index operand of each lloaddloadlstore, and dstore instruction must be no greater than max_locals - 2.

    The implicit index of each lload_<n>dload_<n>lstore_<n>, and dstore_<n> instruction must be no greater than max_locals - 2.

  • The indexbyte operands of each wide instruction modifying an iloadfloadaloadistorefstoreastoreiinc, or ret instruction must represent a non-negative integer no greater thanmax_locals - 1.

    The indexbyte operands of each wide instruction modifying an lloaddloadlstore, or dstore instruction must represent a non-negative integer no greater than max_locals - 2.

4.9.2. Structural Constraints

The structural constraints on the code array specify constraints on relationships between Java Virtual Machine instructions. The structural constraints are as follows:

  • Each instruction must only be executed with the appropriate type and number of arguments in the operand stack and local variable array, regardless of the execution path that leads to its invocation.

    An instruction operating on values of type int is also permitted to operate on values of type booleanbytechar, and short.

    As noted in §2.3.4 and §2.11.1, the Java Virtual Machine internally converts values of types booleanbyteshort, and char to type int.)

  • If an instruction can be executed along several different execution paths, the operand stack must have the same depth (§2.6.2) prior to the execution of the instruction, regardless of the path taken.

  • At no point during execution can the operand stack grow to a depth greater than that implied by the max_stack item.

  • At no point during execution can more values be popped from the operand stack than it contains.

  • At no point during execution can the order of the local variable pair holding a value of type long or double be reversed or the pair split up. At no point can the local variables of such a pair be operated on individually.

  • No local variable (or local variable pair, in the case of a value of type long or double) can be accessed before it is assigned a value.

  • Each invokespecial instruction must name one of the following:

    • an instance initialization method (§2.9.1)

    • a method in the current class or interface

    • a method in a superclass of the current class

    • a method in a direct superinterface of the current class or interface

    • a method in Object

    If an invokespecial instruction names an instance initialization method, then the target reference on the operand stack must be an uninitialized class instance. An instance initialization method must never be invoked on an initialized class instance. In addition:

    • If the target reference on the operand stack is an uninitialized class instance for the current class, then invokespecial must name an instance initialization method from the current class or its direct superclass.

    • If an invokespecial instruction names an instance initialization method and the target reference on the operand stack is a class instance created by an earlier new instruction, theninvokespecial must name an instance initialization method from the class of that class instance.

    If an invokespecial instruction names a method which is not an instance initialization method, then the target reference on the operand stack must be a class instance whose type is assignment compatible with the current class (JLS §5.2).

    The general rule for invokespecial is that the class or interface named by invokespecial must be be "above" the caller class or interface, while the receiver object targeted by invokespecial must be "at" or "below" the caller class or interface. The latter clause is especially important: a class or interface can only perform invokespecial on its own objects. See §invokespecial for an explanation of how the latter clause is implemented in Prolog.

  • Each instance initialization method, except for the instance initialization method derived from the constructor of class Object, must call either another instance initialization method of this or an instance initialization method of its direct superclass super before its instance members are accessed.

    However, instance fields of this that are declared in the current class may be assigned by putfield before calling any instance initialization method.

  • When any instance method is invoked or when any instance variable is accessed, the class instance that contains the instance method or instance variable must already be initialized.

  • If there is an uninitialized class instance in a local variable in code protected by an exception handler, then i) if the handler is inside an <init> method, the handler must throw an exception or loop forever, and ii) if the handler is not inside an <init> method, the uninitialized class instance must remain uninitialized.

  • There must never be an uninitialized class instance on the operand stack or in a local variable when a jsr or jsr_w instruction is executed.

  • The type of every class instance that is the target of a method invocation instruction (that is, the type of the target reference on the operand stack) must be assignment compatible with the class or interface type specified in the instruction.

  • The types of the arguments to each method invocation must be method invocation compatible with the method descriptor (JLS §5.3, §4.3.3).

  • Each return instruction must match its method's return type:

    • If the method returns a booleanbytecharshort, or int, only the ireturn instruction may be used.

    • If the method returns a floatlong, or double, only an freturnlreturn, or dreturn instruction, respectively, may be used.

    • If the method returns a reference type, only an areturn instruction may be used, and the type of the returned value must be assignment compatible with the return descriptor of the method (§4.3.3).

    • All instance initialization methods, class or interface initialization methods, and methods declared to return void must use only the return instruction.

  • The type of every class instance accessed by a getfield instruction or modified by a putfield instruction (that is, the type of the target reference on the operand stack) must be assignment compatible with the class type specified in the instruction.

  • The type of every value stored by a putfield or putstatic instruction must be compatible with the descriptor of the field (§4.3.2) of the class instance or class being stored into:

    • If the descriptor type is booleanbytecharshort, or int, then the value must be an int.

    • If the descriptor type is floatlong, or double, then the value must be a floatlong, or double, respectively.

    • If the descriptor type is a reference type, then the value must be of a type that is assignment compatible with the descriptor type.

  • The type of every value stored into an array by an aastore instruction must be a reference type.

    The component type of the array being stored into by the aastore instruction must also be a reference type.

  • Each athrow instruction must throw only values that are instances of class Throwable or of subclasses of Throwable.

    Each class mentioned in a catch_type item of the exception_table array of the method's Code_attribute structure must be Throwable or a subclass of Throwable.

  • If getfield or putfield is used to access a protected field declared in a superclass that is a member of a different run-time package than the current class, then the type of the class instance being accessed (that is, the type of the target reference on the operand stack) must be assignment compatible with the current class.

    If invokevirtual or invokespecial is used to access a protected method declared in a superclass that is a member of a different run-time package than the current class, then the type of the class instance being accessed (that is, the type of the target reference on the operand stack) must be assignment compatible with the current class.

  • Execution never falls off the bottom of the code array.

  • No return address (a value of type returnAddress) may be loaded from a local variable.

  • The instruction following each jsr or jsr_w instruction may be returned to only by a single ret instruction.

  • No jsr or jsr_w instruction that is returned to may be used to recursively call a subroutine if that subroutine is already present in the subroutine call chain. (Subroutines can be nested when using try-finally constructs from within a finally clause.)

  • Each instance of type returnAddress can be returned to at most once.

    If a ret instruction returns to a point in the subroutine call chain above the ret instruction corresponding to a given instance of type returnAddress, then that instance can never be used as a return address.

4.10. Verification of class Files

Even though a compiler for the Java programming language must only produce class files that satisfy all the static and structural constraints in the previous sections, the Java Virtual Machine has no guarantee that any file it is asked to load was generated by that compiler or is properly formed. Applications such as web browsers do not download source code, which they then compile; these applications download already-compiled class files. The browser needs to determine whether the class file was produced by a trustworthy compiler or by an adversary attempting to exploit the Java Virtual Machine.

An additional problem with compile-time checking is version skew. A user may have successfully compiled a class, say PurchaseStockOptions, to be a subclass of TradingClass. But the definition of TradingClass might have changed since the time the class was compiled in a way that is not compatible with pre-existing binaries. Methods might have been deleted or had their return types or modifiers changed. Fields might have changed types or changed from instance variables to class variables. The access modifiers of a method or variable may have changed from public to private. For a discussion of these issues, see Chapter 13, "Binary Compatibility," in The Java Language Specification, Java SE 9 Edition.

Because of these potential problems, the Java Virtual Machine needs to verify for itself that the desired constraints are satisfied by the class files it attempts to incorporate. A Java Virtual Machine implementation verifies that each class file satisfies the necessary constraints at linking time (§5.4).

Link-time verification enhances the performance of the run-time interpreter. Expensive checks that would otherwise have to be performed to verify constraints at run time for each interpreted instruction can be eliminated. The Java Virtual Machine can assume that these checks have already been performed. For example, the Java Virtual Machine will already know the following:

  • There are no operand stack overflows or underflows.

  • All local variable uses and stores are valid.

  • The arguments to all the Java Virtual Machine instructions are of valid types.

There are two strategies that Java Virtual Machine implementations may use for verification:

  • Verification by type checking must be used to verify class files whose version number is greater than or equal to 50.0.

  • Verification by type inference must be supported by all Java Virtual Machine implementations, except those conforming to the Java ME CLDC and Java Card profiles, in order to verify classfiles whose version number is less than 50.0.

    Verification on Java Virtual Machine implementations supporting the Java ME CLDC and Java Card profiles is governed by their respective specifications.

In both strategies, verification is mainly concerned with enforcing the static and structural constraints from §4.9 on the code array of the Code attribute (§4.7.3). However, there are three additional checks outside the Code attribute which must be performed during verification:

  • Ensuring that final classes are not subclassed.

  • Ensuring that final methods are not overridden (§5.4.5).

  • Checking that every class (except Object) has a direct superclass.

4.10.1. Verification by Type Checking

class file whose version number is 50.0 or above (§4.1) must be verified using the type checking rules given in this section.

If, and only if, a class file's version number equals 50.0, then if the type checking fails, a Java Virtual Machine implementation may choose to attempt to perform verification by type inference (§4.10.2).

This is a pragmatic adjustment, designed to ease the transition to the new verification discipline. Many tools that manipulate class files may alter the bytecodes of a method in a manner that requires adjustment of the method's stack map frames. If a tool does not make the necessary adjustments to the stack map frames, type checking may fail even though the bytecode is in principle valid (and would consequently verify under the old type inference scheme). To allow implementors time to adapt their tools, Java Virtual Machine implementations may fall back to the older verification discipline, but only for a limited time.

In cases where type checking fails but type inference is invoked and succeeds, a certain performance penalty is expected. Such a penalty is unavoidable. It also should serve as a signal to tool vendors that their output needs to be adjusted, and provides vendors with additional incentive to make these adjustments.

In summary, failover to verification by type inference supports both the gradual addition of stack map frames to the Java SE Platform (if they are not present in a version 50.0 class file, failover is allowed) and the gradual removal of the jsr and jsr_w instructions from the Java SE Platform (if they are present in a version 50.0 class file, failover is allowed).

If a Java Virtual Machine implementation ever attempts to perform verification by type inference on version 50.0 class files, it must do so in all cases where verification by type checking fails.

This means that a Java Virtual Machine implementation cannot choose to resort to type inference in once case and not in another. It must either reject class files that do not verify via type checking, or else consistently failover to the type inferencing verifier whenever type checking fails.

The type checker enforces type rules that are specified by means of Prolog clauses. English language text is used to describe the type rules in an informal way, while the Prolog clauses provide a formal specification.

The type checker requires a list of stack map frames for each method with a Code attribute (§4.7.3). A list of stack map frames is given by the StackMapTable attribute (§4.7.4) of a Codeattribute. The intent is that a stack map frame must appear at the beginning of each basic block in a method. The stack map frame specifies the verification type of each operand stack entry and of each local variable at the start of each basic block. The type checker reads the stack map frames for each method with a Code attribute and uses these maps to generate a proof of the type safety of the instructions in the Code attribute.

A class is type safe if all its methods are type safe, and it does not subclass a final class.

classIsTypeSafe(Class) :-
classClassName(Class, Name),
classDefiningLoader(Class, L),
superclassChain(Name, L, Chain),
Chain \= [],
classSuperClassName(Class, SuperclassName),
loadedClass(SuperclassName, L, Superclass),
classIsNotFinal(Superclass),
classMethods(Class, Methods),
checklist(methodIsTypeSafe(Class), Methods).
classIsTypeSafe(Class) :-
classClassName(Class, 'java/lang/Object'),
classDefiningLoader(Class, L),
isBootstrapLoader(L),
classMethods(Class, Methods),
checklist(methodIsTypeSafe(Class), Methods).

The Prolog predicate classIsTypeSafe assumes that Class is a Prolog term representing a binary class that has been successfully parsed and loaded. This specification does not mandate the precise structure of this term, but does require that certain predicates be defined upon it.

For example, we assume a predicate classMethods(Class, Methods) that, given a term representing a class as described above as its first argument, binds its second argument to a list comprising all the methods of the class, represented in a convenient form described later.

Iff the predicate classIsTypeSafe is not true, the type checker must throw the exception VerifyError to indicate that the class file is malformed. Otherwise, the class file has type checked successfully and bytecode verification has completed successfully.

The rest of this section explains the process of type checking in detail:

  • First, we give Prolog predicates for core Java Virtual Machine artifacts like classes and methods (§4.10.1.1).

  • Second, we specify the type system known to the type checker (§4.10.1.2).

  • Third, we specify the Prolog representation of instructions and stack map frames (§4.10.1.3§4.10.1.4).

  • Fourth, we specify how a method is type checked, for methods without code (§4.10.1.5) and methods with code (§4.10.1.6).

  • Fifth, we discuss type checking issues common to all load and store instructions (§4.10.1.7), and also issues of access to protected members (§4.10.1.8).

  • Finally, we specify the rules to type check each instruction (§4.10.1.9).

4.10.1.1. Accessors for Java Virtual Machine Artifacts

We stipulate the existence of 28 Prolog predicates ("accessors") that have certain expected behavior but whose formal definitions are not given in this specification.

classClassName(Class, ClassName)

Extracts the name, ClassName, of the class Class.

classIsInterface(Class)

True iff the class, Class, is an interface.

classIsNotFinal(Class)

True iff the class, Class, is not a final class.

classSuperClassName(Class, SuperClassName)

Extracts the name, SuperClassName, of the superclass of class Class.

classInterfaces(Class, Interfaces)

Extracts a list, Interfaces, of the direct superinterfaces of the class Class.

classMethods(Class, Methods)

Extracts a list, Methods, of the methods declared in the class Class.

classAttributes(Class, Attributes)

Extracts a list, Attributes, of the attributes of the class Class.

Each attribute is represented as a functor application of the form attribute(AttributeName, AttributeContents), where AttributeName is the name of the attribute. The format of the attribute's contents is unspecified.

classDefiningLoader(Class, Loader)

Extracts the defining class loader, Loader, of the class Class.

isBootstrapLoader(Loader)

True iff the class loader Loader is the bootstrap class loader.

loadedClass(Name, InitiatingLoader, ClassDefinition)

True iff there exists a class named Name whose representation (in accordance with this specification) when loaded by the class loader InitiatingLoader is ClassDefinition.

methodName(Method, Name)

Extracts the name, Name, of the method Method.

methodAccessFlags(Method, AccessFlags)

Extracts the access flags, AccessFlags, of the method Method.

methodDescriptor(Method, Descriptor)

Extracts the descriptor, Descriptor, of the method Method.

methodAttributes(Method, Attributes)

Extracts a list, Attributes, of the attributes of the method Method.

isInit(Method)

True iff Method (regardless of class) is <init>.

isNotInit(Method)

True iff Method (regardless of class) is not <init>.

isNotFinal(Method, Class)

True iff Method in class Class is not final.

isStatic(Method, Class)

True iff Method in class Class is static.

isNotStatic(Method, Class)

True iff Method in class Class is not static.

isPrivate(Method, Class)

True iff Method in class Class is private.

isNotPrivate(Method, Class)

True iff Method in class Class is not private.

isProtected(MemberClass, MemberName, MemberDescriptor)

True iff there is a member named MemberName with descriptor MemberDescriptor in the class MemberClass and it is protected.

isNotProtected(MemberClass, MemberName, MemberDescriptor)

True iff there is a member named MemberName with descriptor MemberDescriptor in the class MemberClass and it is not protected.

parseFieldDescriptor(Descriptor, Type)

Converts a field descriptor, Descriptor, into the corresponding verification type Type (§4.10.1.2).

parseMethodDescriptor(Descriptor, ArgTypeList, ReturnType)

Converts a method descriptor, Descriptor, into a list of verification types, ArgTypeList, corresponding to the method argument types, and a verification type, ReturnType, corresponding to the return type.

parseCodeAttribute(Class, Method, FrameSize, MaxStack, ParsedCode, Handlers, StackMap)

Extracts the instruction stream, ParsedCode, of the method Method in Class, as well as the maximum operand stack size, MaxStack, the maximal number of local variables, FrameSize, the exception handlers, Handlers, and the stack map StackMap.

The representation of the instruction stream and stack map attribute must be as specified in §4.10.1.3 and §4.10.1.4.

samePackageName(Class1, Class2)

True iff the package names of Class1 and Class2 are the same.

differentPackageName(Class1, Class2)

True iff the package names of Class1 and Class2 are different.

When type checking a method's body, it is convenient to access information about the method. For this purpose, we define an environment, a six-tuple consisting of:

  • a class

  • a method

  • the declared return type of the method

  • the instructions in a method

  • the maximal size of the operand stack

  • a list of exception handlers

We specify accessors to extract information from the environment.

allInstructions(Environment, Instructions) :-
Environment = environment(_Class, _Method, _ReturnType,
Instructions, _, _). exceptionHandlers(Environment, Handlers) :-
Environment = environment(_Class, _Method, _ReturnType,
_Instructions, _, Handlers). maxOperandStackLength(Environment, MaxStack) :-
Environment = environment(_Class, _Method, _ReturnType,
_Instructions, MaxStack, _Handlers). thisClass(Environment, class(ClassName, L)) :-
Environment = environment(Class, _Method, _ReturnType,
_Instructions, _, _),
classDefiningLoader(Class, L),
classClassName(Class, ClassName). thisMethodReturnType(Environment, ReturnType) :-
Environment = environment(_Class, _Method, ReturnType,
_Instructions, _, _).

We specify additional predicates to extract higher-level information from the environment.

offsetStackFrame(Environment, Offset, StackFrame) :-
allInstructions(Environment, Instructions),
member(stackMap(Offset, StackFrame), Instructions). currentClassLoader(Environment, Loader) :-
thisClass(Environment, class(_, Loader)).

Finally, we specify a general predicate used throughout the type rules:

notMember(_, []).
notMember(X, [A | More]) :- X \= A, notMember(X, More).

The principle guiding the determination as to which accessors are stipulated and which are fully specified is that we do not want to over-specify the representation of the class file. Providing specific accessors to the Class or Method term would force us to completely specify the format for a Prolog term representing the class file.

4.10.1.2. Verification Type System

The type checker enforces a type system based upon a hierarchy of verification types, illustrated below.

Verification type hierarchy:

                             top
____________/\____________
/ \
/ \
oneWord twoWord
/ | \ / \
/ | \ / \
int float reference long double
/ \
/ \_____________
/ \
/ \
uninitialized +------------------+
/ \ | Java reference |
/ \ | type hierarchy |
uninitializedThis uninitialized(Offset) +------------------+
|
|
null

Most verification types have a direct correspondence with the primitive and reference types represented by field descriptors in Table 4.3-A:

  • The primitive types doublefloatint, and long (field descriptors DFIJ) each correspond to the verification type of the same name.

  • The primitive types bytecharshort, and boolean (field descriptors BCSZ) all correspond to the verification type int.

  • Class and interface types (field descriptors beginning L) correspond to verification types that use the functor class. The verification type class(NL) represents the class whose binary name is N as loaded by the loader L. Note that L is an initiating loader (§5.3) of the class represented by class(NL) and may, or may not, be the class's defining loader.

    For example, the class type Object would be represented as class('java/lang/Object', BL), where BL is the bootstrap loader.

  • Array types (field descriptors beginning [) correspond to verification types that use the functor arrayOf. Note that the primitive types bytecharshort, and boolean do not correspond to verification types, but an array type whose element type is bytecharshort, or boolean does correspond to a verification type; such verification types support the baloadbastorecaloadcastoresaloadsastore, and newarray instructions.

    • The verification type arrayOf(T) represents the array type whose component type is the verification type T.

    • The verification type arrayOf(byte) represents the array type whose element type is byte.

    • The verification type arrayOf(char) represents the array type whose element type is char.

    • The verification type arrayOf(short) represents the array type whose element type is short.

    • The verification type arrayOf(boolean) represents the array type whose element type is boolean.

    For example, the array types int[] and Object[] would be represented by the verification types arrayOf(int) and arrayOf(class('java/lang/Object', BL)) respectively. The array types byte[] and boolean[][] would be represented by the verification types arrayOf(byte) and arrayOf(arrayOf(boolean)) respectively.

The remaining verification types are described as follows:

  • The verification types toponeWordtwoWord, and reference are represented in Prolog as atoms whose name denotes the verification type in question.

  • The verification type uninitialized(Offset) is represented by applying the functor uninitialized to an argument representing the numerical value of the Offset.

The subtyping rules for verification types are as follows.

Subtyping is reflexive.

isAssignable(X, X).

The verification types which are not reference types in the Java programming language have subtype rules of the form:

isAssignable(v, X) :- isAssignable(the_direct_supertype_of_v, X).

That is, v is a subtype of X if the direct supertype of v is a subtype of X. The rules are:

isAssignable(oneWord, top).
isAssignable(twoWord, top). isAssignable(int, X) :- isAssignable(oneWord, X).
isAssignable(float, X) :- isAssignable(oneWord, X).
isAssignable(long, X) :- isAssignable(twoWord, X).
isAssignable(double, X) :- isAssignable(twoWord, X). isAssignable(reference, X) :- isAssignable(oneWord, X).
isAssignable(class(_, _), X) :- isAssignable(reference, X).
isAssignable(arrayOf(_), X) :- isAssignable(reference, X). isAssignable(uninitialized, X) :- isAssignable(reference, X).
isAssignable(uninitializedThis, X) :- isAssignable(uninitialized, X).
isAssignable(uninitialized(_), X) :- isAssignable(uninitialized, X). isAssignable(null, class(_, _)).
isAssignable(null, arrayOf(_)).
isAssignable(null, X) :- isAssignable(class('java/lang/Object', BL), X),
isBootstrapLoader(BL).

These subtype rules are not necessarily the most obvious formulation of subtyping. There is a clear split between subtyping rules for reference types in the Java programming language, and rules for the remaining verification types. The split allows us to state general subtyping relations between Java programming language reference types and other verification types. These relations hold independently of a Java reference type's position in the type hierarchy, and help to prevent excessive class loading by a Java Virtual Machine implementation. For example, we do not want to start climbing the Java superclass hierarchy in response to a query of the form class(foo, L) <: twoWord.

We also have a rule that says subtyping is reflexive, so together these rules cover most verification types that are not reference types in the Java programming language.

Subtype rules for the reference types in the Java programming language are specified recursively with isJavaAssignable.

isAssignable(class(X, Lx), class(Y, Ly)) :-
isJavaAssignable(class(X, Lx), class(Y, Ly)). isAssignable(arrayOf(X), class(Y, L)) :-
isJavaAssignable(arrayOf(X), class(Y, L)). isAssignable(arrayOf(X), arrayOf(Y)) :-
isJavaAssignable(arrayOf(X), arrayOf(Y)).

For assignments, interfaces are treated like Object.

isJavaAssignable(class(_, _), class(To, L)) :-
loadedClass(To, L, ToClass),
classIsInterface(ToClass). isJavaAssignable(From, To) :-
isJavaSubclassOf(From, To).

Array types are subtypes of Object. The intent is also that array types are subtypes of Cloneable and java.io.Serializable.

isJavaAssignable(arrayOf(_), class('java/lang/Object', BL)) :-
isBootstrapLoader(BL). isJavaAssignable(arrayOf(_), X) :-
isArrayInterface(X). isArrayInterface(class('java/lang/Cloneable', BL)) :-
isBootstrapLoader(BL). isArrayInterface(class('java/io/Serializable', BL)) :-
isBootstrapLoader(BL).

Subtyping between arrays of primitive type is the identity relation.

isJavaAssignable(arrayOf(X), arrayOf(Y)) :-
atom(X),
atom(Y),
X = Y.

Subtyping between arrays of reference type is covariant.

isJavaAssignable(arrayOf(X), arrayOf(Y)) :-
compound(X), compound(Y), isJavaAssignable(X, Y).

Subclassing is reflexive.

isJavaSubclassOf(class(SubclassName, L), class(SubclassName, L)).
isJavaSubclassOf(class(SubclassName, LSub), class(SuperclassName, LSuper)) :-
superclassChain(SubclassName, LSub, Chain),
member(class(SuperclassName, L), Chain),
loadedClass(SuperclassName, L, Sup),
loadedClass(SuperclassName, LSuper, Sup). superclassChain(ClassName, L, [class(SuperclassName, Ls) | Rest]) :-
loadedClass(ClassName, L, Class),
classSuperClassName(Class, SuperclassName),
classDefiningLoader(Class, Ls),
superclassChain(SuperclassName, Ls, Rest). superclassChain('java/lang/Object', L, []) :-
loadedClass('java/lang/Object', L, Class),
classDefiningLoader(Class, BL),
isBootstrapLoader(BL).

4.10.1.3. Instruction Representation

Individual bytecode instructions are represented in Prolog as terms whose functor is the name of the instruction and whose arguments are its parsed operands.

For example, an aload instruction is represented as the term aload(N), which includes the index N that is the operand of the instruction.

The instructions as a whole are represented as a list of terms of the form:

instruction(Offset, AnInstruction)

For example, instruction(21, aload(1)).

The order of instructions in this list must be the same as in the class file.

A few instructions have operands that are constant pool entries representing fields, methods, and dynamic call sites. In the constant pool, a field is represented by a CONSTANT_Fieldref_infostructure, a method is represented by a CONSTANT_InterfaceMethodref_info structure (for an interface's method) or a CONSTANT_Methodref_info structure (for a class's method), and a dynamic call site is represented by a CONSTANT_InvokeDynamic_info structure (§4.4.2§4.4.10). Such structures are represented as functor applications of the form:

  • field(FieldClassName, FieldName, FieldDescriptor) for a field, where FieldClassName is the name of the class referenced by the class_index item in theCONSTANT_Fieldref_info structure, and FieldName and FieldDescriptor correspond to the name and field descriptor referenced by the name_and_type_index item of theCONSTANT_Fieldref_info structure.

  • imethod(MethodIntfName, MethodName, MethodDescriptor) for an interface's method, where MethodIntfName is the name of the interface referenced by the class_index item of the CONSTANT_InterfaceMethodref_info structure, and MethodName and MethodDescriptor correspond to the name and method descriptor referenced by the name_and_type_index item of the CONSTANT_InterfaceMethodref_info structure;

  • method(MethodClassName, MethodName, MethodDescriptor) for a class's method, where MethodClassName is the name of the class referenced by the class_index item of the CONSTANT_Methodref_info structure, and MethodName and MethodDescriptor correspond to the name and method descriptor referenced by the name_and_type_index item of theCONSTANT_Methodref_info structure; and

  • dmethod(CallSiteName, MethodDescriptor) for a dynamic call site, where CallSiteName and MethodDescriptor correspond to the name and method descriptor referenced by the name_and_type_index item of the CONSTANT_InvokeDynamic_info structure.

For clarity, we assume that field and method descriptors (§4.3.2§4.3.3) are mapped into more readable names: the leading L and trailing ; are dropped from class names, and the BaseTypecharacters used for primitive types are mapped to the names of those types.

For example, a getfield instruction whose operand was an index into the constant pool that refers to a field foo of type F in class Bar would be represented as getfield(field('Bar', 'foo', 'F')).

Constant pool entries that refer to constant values, such as CONSTANT_StringCONSTANT_IntegerCONSTANT_FloatCONSTANT_LongCONSTANT_Double, and CONSTANT_Class, are encoded via the functors whose names are stringintfloatlongdouble, and classConstant respectively.

For example, an ldc instruction for loading the integer 91 would be encoded as ldc(int(91)).

4.10.1.4. Stack Map Frames and Type Transitions

Stack map frames are represented in Prolog as a list of terms of the form:

stackMap(Offset, TypeState)

where:

  • Offset is an integer indicating the bytecode offset at which the stack map frame applies (§4.7.4).

    The order of bytecode offsets in this list must be the same as in the class file.

  • TypeState is the expected incoming type state for the instruction at Offset.

type state is a mapping from locations in the operand stack and local variables of a method to verification types. It has the form:

frame(Locals, OperandStack, Flags)

where:

  • Locals is a list of verification types, such that the i'th element of the list (with 0-based indexing) represents the type of local variable i.

    Types of size 2 (long and double) are represented by two local variables (§2.6.1), with the first local variable being the type itself and the second local variable being top (§4.10.1.7).

  • OperandStack is a list of verification types, such that the first element of the list represents the type of the top of the operand stack, and the types of stack entries below the top follow in the list in the appropriate order.

    Types of size 2 (long and double) are represented by two stack entries, with the first entry being top and the second entry being the type itself.

    For example, a stack with a double value, an int value, and a long value is represented in a type state as a stack with five entries: top and double entries for the double value, an intentry for the int value, and top and long entries for the long value. Accordingly, OperandStack is the list [top, double, int, top, long].

  • Flags is a list which may either be empty or have the single element flagThisUninit.

    If any local variable in Locals has the type uninitializedThis, then Flags has the single element flagThisUninit, otherwise Flags is an empty list.

    flagThisUninit is used in constructors to mark type states where initialization of this has not yet been completed. In such type states, it is illegal to return from the method.

Subtyping of verification types is extended pointwise to type states. The local variable array of a method has a fixed length by construction (see methodInitialStackFrame in §4.10.1.6), but the operand stack grows and shrinks, so we require an explicit check on the length of the operand stacks whose assignability is desired for subtyping.

frameIsAssignable(frame(Locals1, StackMap1, Flags1),
frame(Locals2, StackMap2, Flags2)) :-
length(StackMap1, StackMapLength),
length(StackMap2, StackMapLength),
maplist(isAssignable, Locals1, Locals2),
maplist(isAssignable, StackMap1, StackMap2),
subset(Flags1, Flags2).

Most of the type rules for individual instructions (§4.10.1.9) depend on the notion of a valid type transition. A type transition is valid if one can pop a list of expected types off the incoming type state's operand stack and replace them with an expected result type, resulting in a new type state where the length of the operand stack does not exceed its declared maximum size.

validTypeTransition(Environment, ExpectedTypesOnStack, ResultType,
frame(Locals, InputOperandStack, Flags),
frame(Locals, NextOperandStack, Flags)) :-
popMatchingList(InputOperandStack, ExpectedTypesOnStack,
InterimOperandStack),
pushOperandStack(InterimOperandStack, ResultType, NextOperandStack),
operandStackHasLegalLength(Environment, NextOperandStack).

Pop a list of types off the stack.

popMatchingList(OperandStack, [], OperandStack).
popMatchingList(OperandStack, [P | Rest], NewOperandStack) :-
popMatchingType(OperandStack, P, TempOperandStack, _ActualType),
popMatchingList(TempOperandStack, Rest, NewOperandStack).

Pop an individual type off the stack. The exact behavior depends on the stack contents. If the logical top of the stack is some subtype of the specified type, Type, then pop it. If a type occupies two stack entries, then the logical top of the stack is really the type just below the top, and the top of the stack is the unusable type top.

popMatchingType([ActualType | OperandStack],
Type, OperandStack, ActualType) :-
sizeOf(Type, 1),
isAssignable(ActualType, Type). popMatchingType([top, ActualType | OperandStack],
Type, OperandStack, ActualType) :-
sizeOf(Type, 2),
isAssignable(ActualType, Type). sizeOf(X, 2) :- isAssignable(X, twoWord).
sizeOf(X, 1) :- isAssignable(X, oneWord).
sizeOf(top, 1).

Push a logical type onto the stack. The exact behavior varies with the size of the type. If the pushed type is of size 1, we just push it onto the stack. If the pushed type is of size 2, we push it, and then push top.

pushOperandStack(OperandStack, 'void', OperandStack).
pushOperandStack(OperandStack, Type, [Type | OperandStack]) :-
sizeOf(Type, 1).
pushOperandStack(OperandStack, Type, [top, Type | OperandStack]) :-
sizeOf(Type, 2).

The length of the operand stack must not exceed the declared maximum size.

operandStackHasLegalLength(Environment, OperandStack) :-
length(OperandStack, Length),
maxOperandStackLength(Environment, MaxStack),
Length =< MaxStack.

The dup instructions pop expected types off the incoming type state's operand stack and replace them with predefined result types, resulting in a new type state. However, these instructions are not defined in terms of type transitions because there is no need to match types by means of the subtyping relation. Instead, the dup instructions manipulate the operand stack entirely in terms of the category of types on the stack (§2.11.1).

Category 1 types occupy a single stack entry. Popping a logical type of category 1, Type, off the stack is possible if the top of the stack is Type and Type is not top (otherwise it could denote the upper half of a category 2 type). The result is the incoming stack, with the top entry popped off.

popCategory1([Type | Rest], Type, Rest) :-
Type \= top,
sizeOf(Type, 1).

Category 2 types occupy two stack entries. Popping a logical type of category 2, Type, off the stack is possible if the top of the stack is type top, and the entry directly below it is Type. The result is the incoming stack, with the top two entries popped off.

popCategory2([top, Type | Rest], Type, Rest) :-
sizeOf(Type, 2).

The dup instructions push a list of types onto the stack in essentially the same way as when a type is pushed for a valid type transition.

canSafelyPush(Environment, InputOperandStack, Type, OutputOperandStack) :-
pushOperandStack(InputOperandStack, Type, OutputOperandStack),
operandStackHasLegalLength(Environment, OutputOperandStack). canSafelyPushList(Environment, InputOperandStack, Types,
OutputOperandStack) :-
canPushList(InputOperandStack, Types, OutputOperandStack),
operandStackHasLegalLength(Environment, OutputOperandStack). canPushList(InputOperandStack, [], InputOperandStack).
canPushList(InputOperandStack, [Type | Rest], OutputOperandStack) :-
pushOperandStack(InputOperandStack, Type, InterimOperandStack),
canPushList(InterimOperandStack, Rest, OutputOperandStack).

Many of the type rules for individual instructions use the following clause to easily pop a list of types off the stack.

canPop(frame(Locals, OperandStack, Flags), Types,
frame(Locals, PoppedOperandStack, Flags)) :-
popMatchingList(OperandStack, Types, PoppedOperandStack).

Finally, certain array instructions (§aaload§arraylength§baload§bastore) peek at types on the operand stack in order to check they are array types. The following clause accesses the i'th element of the operand stack from a type state.

nth1OperandStackIs(i, frame(_Locals, OperandStack, _Flags), Element) :-
nth1(i, OperandStack, Element).

4.10.1.5. Type Checking Abstract and Native Methods

abstract methods and native methods are considered to be type safe if they do not override a final method.

methodIsTypeSafe(Class, Method) :-
doesNotOverrideFinalMethod(Class, Method),
methodAccessFlags(Method, AccessFlags),
member(abstract, AccessFlags). methodIsTypeSafe(Class, Method) :-
doesNotOverrideFinalMethod(Class, Method),
methodAccessFlags(Method, AccessFlags),
member(native, AccessFlags).

private methods and static methods are orthogonal to dynamic method dispatch, so they never override other methods (§5.4.5).

doesNotOverrideFinalMethod(class('java/lang/Object', L), Method) :-
isBootstrapLoader(L). doesNotOverrideFinalMethod(Class, Method) :-
isPrivate(Method, Class). doesNotOverrideFinalMethod(Class, Method) :-
isStatic(Method, Class). doesNotOverrideFinalMethod(Class, Method) :-
isNotPrivate(Method, Class),
isNotStatic(Method, Class),
doesNotOverrideFinalMethodOfSuperclass(Class, Method). doesNotOverrideFinalMethodOfSuperclass(Class, Method) :-
classSuperClassName(Class, SuperclassName),
classDefiningLoader(Class, L),
loadedClass(SuperclassName, L, Superclass),
classMethods(Superclass, SuperMethodList),
finalMethodNotOverridden(Method, Superclass, SuperMethodList).

final methods that are private and/or static are unusual, as private methods and static methods cannot be overridden per se. Therefore, if a final private method or a finalstatic method is found, it was logically not overridden by another method.

finalMethodNotOverridden(Method, Superclass, SuperMethodList) :-
methodName(Method, Name),
methodDescriptor(Method, Descriptor),
member(method(_, Name, Descriptor), SuperMethodList),
isFinal(Method, Superclass),
isPrivate(Method, Superclass). finalMethodNotOverridden(Method, Superclass, SuperMethodList) :-
methodName(Method, Name),
methodDescriptor(Method, Descriptor),
member(method(_, Name, Descriptor), SuperMethodList),
isFinal(Method, Superclass),
isStatic(Method, Superclass).

If a non-final private method or a non-final static method is found, skip over it because it is orthogonal to overriding.

finalMethodNotOverridden(Method, Superclass, SuperMethodList) :-
methodName(Method, Name),
methodDescriptor(Method, Descriptor),
member(method(_, Name, Descriptor), SuperMethodList),
isNotFinal(Method, Superclass),
isPrivate(Method, Superclass),
doesNotOverrideFinalMethodOfSuperclass(Superclass, Method). finalMethodNotOverridden(Method, Superclass, SuperMethodList) :-
methodName(Method, Name),
methodDescriptor(Method, Descriptor),
member(method(_, Name, Descriptor), SuperMethodList),
isNotFinal(Method, Superclass),
isStatic(Method, Superclass),
doesNotOverrideFinalMethodOfSuperclass(Superclass, Method).

If a non-final, non-private, non-static method is found, then indeed a final method was not overridden. Otherwise, recurse upwards.

finalMethodNotOverridden(Method, Superclass, SuperMethodList) :-
methodName(Method, Name),
methodDescriptor(Method, Descriptor),
member(method(_, Name, Descriptor), SuperMethodList),
isNotFinal(Method, Superclass),
isNotStatic(Method, Superclass),
isNotPrivate(Method, Superclass). finalMethodNotOverridden(Method, Superclass, SuperMethodList) :-
methodName(Method, Name),
methodDescriptor(Method, Descriptor),
notMember(method(_, Name, Descriptor), SuperMethodList),
doesNotOverrideFinalMethodOfSuperclass(Superclass, Method).

4.10.1.6. Type Checking Methods with Code

Non-abstract, non-native methods are type correct if they have code and the code is type correct.

methodIsTypeSafe(Class, Method) :-
doesNotOverrideFinalMethod(Class, Method),
methodAccessFlags(Method, AccessFlags),
methodAttributes(Method, Attributes),
notMember(native, AccessFlags),
notMember(abstract, AccessFlags),
member(attribute('Code', _), Attributes),
methodWithCodeIsTypeSafe(Class, Method).

A method with code is type safe if it is possible to merge the code and the stack map frames into a single stream such that each stack map frame precedes the instruction it corresponds to, and the merged stream is type correct. The method's exception handlers, if any, must also be legal.

methodWithCodeIsTypeSafe(Class, Method) :-
parseCodeAttribute(Class, Method, FrameSize, MaxStack,
ParsedCode, Handlers, StackMap),
mergeStackMapAndCode(StackMap, ParsedCode, MergedCode),
methodInitialStackFrame(Class, Method, FrameSize, StackFrame, ReturnType),
Environment = environment(Class, Method, ReturnType, MergedCode,
MaxStack, Handlers),
handlersAreLegal(Environment),
mergedCodeIsTypeSafe(Environment, MergedCode, StackFrame).

Let us consider exception handlers first.

An exception handler is represented by a functor application of the form:

handler(Start, End, Target, ClassName)

whose arguments are, respectively, the start and end of the range of instructions covered by the handler, the first instruction of the handler code, and the name of the exception class that this handler is designed to handle.

An exception handler is legal if its start (Start) is less than its end (End), there exists an instruction whose offset is equal to Start, there exists an instruction whose offset equals End, and the handler's exception class is assignable to the class Throwable. The exception class of a handler is Throwable if the handler's class entry is 0, otherwise it is the class named in the handler.

An additional requirement exists for a handler inside an <init> method if one of the instructions covered by the handler is invokespecial of an <init> method. In this case, the fact that a handler is running means the object under construction is likely broken, so it is important that the handler does not swallow the exception and allow the enclosing <init> method to return normally to the caller. Accordingly, the handler is required to either complete abruptly by throwing an exception to the caller of the enclosing <init> method, or to loop forever.

handlersAreLegal(Environment) :-
exceptionHandlers(Environment, Handlers),
checklist(handlerIsLegal(Environment), Handlers). handlerIsLegal(Environment, Handler) :-
Handler = handler(Start, End, Target, _),
Start < End,
allInstructions(Environment, Instructions),
member(instruction(Start, _), Instructions),
offsetStackFrame(Environment, Target, _),
instructionsIncludeEnd(Instructions, End),
currentClassLoader(Environment, CurrentLoader),
handlerExceptionClass(Handler, ExceptionClass, CurrentLoader),
isBootstrapLoader(BL),
isAssignable(ExceptionClass, class('java/lang/Throwable', BL)),
initHandlerIsLegal(Environment, Handler). instructionsIncludeEnd(Instructions, End) :-
member(instruction(End, _), Instructions).
instructionsIncludeEnd(Instructions, End) :-
member(endOfCode(End), Instructions). handlerExceptionClass(handler(_, _, _, 0),
class('java/lang/Throwable', BL), _) :-
isBootstrapLoader(BL). handlerExceptionClass(handler(_, _, _, Name),
class(Name, L), L) :-
Name \= 0.
initHandlerIsLegal(Environment, Handler) :-
notInitHandler(Environment, Handler). notInitHandler(Environment, Handler) :-
Environment = environment(_Class, Method, _, Instructions, _, _),
isNotInit(Method). notInitHandler(Environment, Handler) :-
Environment = environment(_Class, Method, _, Instructions, _, _),
isInit(Method),
member(instruction(_, invokespecial(CP)), Instructions),
CP = method(MethodClassName, MethodName, Descriptor),
MethodName \= '<init>'. initHandlerIsLegal(Environment, Handler) :-
isInitHandler(Environment, Handler),
sublist(isApplicableInstruction(Target), Instructions,
HandlerInstructions),
noAttemptToReturnNormally(HandlerInstructions). isInitHandler(Environment, Handler) :-
Environment = environment(_Class, Method, _, Instructions, _, _),
isInit(Method).
member(instruction(_, invokespecial(CP)), Instructions),
CP = method(MethodClassName, '<init>', Descriptor). isApplicableInstruction(HandlerStart, instruction(Offset, _)) :-
Offset >= HandlerStart. noAttemptToReturnNormally(Instructions) :-
notMember(instruction(_, return), Instructions). noAttemptToReturnNormally(Instructions) :-
member(instruction(_, athrow), Instructions).

Let us now turn to the stream of instructions and stack map frames.

Merging instructions and stack map frames into a single stream involves four cases:

  • Merging an empty StackMap and a list of instructions yields the original list of instructions.

    mergeStackMapAndCode([], CodeList, CodeList).
    
  • Given a list of stack map frames beginning with the type state for the instruction at Offset, and a list of instructions beginning at Offset, the merged list is the head of the stack map frame list, followed by the head of the instruction list, followed by the merge of the tails of the two lists.

    mergeStackMapAndCode([stackMap(Offset, Map) | RestMap],
    [instruction(Offset, Parse) | RestCode],
    [stackMap(Offset, Map),
    instruction(Offset, Parse) | RestMerge]) :-
    mergeStackMapAndCode(RestMap, RestCode, RestMerge).
  • Otherwise, given a list of stack map frames beginning with the type state for the instruction at OffsetM, and a list of instructions beginning at OffsetP, then, if OffsetP < OffsetM, the merged list consists of the head of the instruction list, followed by the merge of the stack map frame list and the tail of the instruction list.

    mergeStackMapAndCode([stackMap(OffsetM, Map) | RestMap],
    [instruction(OffsetP, Parse) | RestCode],
    [instruction(OffsetP, Parse) | RestMerge]) :-
    OffsetP < OffsetM,
    mergeStackMapAndCode([stackMap(OffsetM, Map) | RestMap],
    RestCode, RestMerge).
  • Otherwise, the merge of the two lists is undefined. Since the instruction list has monotonically increasing offsets, the merge of the two lists is not defined unless every stack map frame offset has a corresponding instruction offset and the stack map frames are in monotonically increasing order.

To determine if the merged stream for a method is type correct, we first infer the method's initial type state.

The initial type state of a method consists of an empty operand stack and local variable types derived from the type of this and the arguments, as well as the appropriate flag, depending on whether this is an <init> method.

methodInitialStackFrame(Class, Method, FrameSize, frame(Locals, [], Flags),
ReturnType):-
methodDescriptor(Method, Descriptor),
parseMethodDescriptor(Descriptor, RawArgs, ReturnType),
expandTypeList(RawArgs, Args),
methodInitialThisType(Class, Method, ThisList),
flags(ThisList, Flags),
append(ThisList, Args, ThisArgs),
expandToLength(ThisArgs, FrameSize, top, Locals).

Given a list of types, the following clause produces a list where every type of size 2 has been substituted by two entries: one for itself, and one top entry. The result then corresponds to the representation of the list as 32-bit words in the Java Virtual Machine.

expandTypeList([], []).
expandTypeList([Item | List], [Item | Result]) :-
sizeOf(Item, 1),
expandTypeList(List, Result).
expandTypeList([Item | List], [Item, top | Result]) :-
sizeOf(Item, 2),
expandTypeList(List, Result).
flags([uninitializedThis], [flagThisUninit]).
flags(X, []) :- X \= [uninitializedThis]. expandToLength(List, Size, _Filler, List) :-
length(List, Size).
expandToLength(List, Size, Filler, Result) :-
length(List, ListLength),
ListLength < Size,
Delta is Size - ListLength,
length(Extra, Delta),
checklist(=(Filler), Extra),
append(List, Extra, Result).

For the initial type state of an instance method, we compute the type of this and put it in a list. The type of this in the <init> method of Object is Object; in other <init> methods, the type of this is uninitializedThis; otherwise, the type of this in an instance method is class(N, L) where N is the name of the class containing the method and L is its defining class loader.

For the initial type state of a static method, this is irrelevant, so the list is empty.

methodInitialThisType(_Class, Method, []) :-
methodAccessFlags(Method, AccessFlags),
member(static, AccessFlags),
methodName(Method, MethodName),
MethodName \= '<init>'. methodInitialThisType(Class, Method, [This]) :-
methodAccessFlags(Method, AccessFlags),
notMember(static, AccessFlags),
instanceMethodInitialThisType(Class, Method, This). instanceMethodInitialThisType(Class, Method, class('java/lang/Object', L)) :-
methodName(Method, '<init>'),
classDefiningLoader(Class, L),
isBootstrapLoader(L),
classClassName(Class, 'java/lang/Object'). instanceMethodInitialThisType(Class, Method, uninitializedThis) :-
methodName(Method, '<init>'),
classClassName(Class, ClassName),
classDefiningLoader(Class, CurrentLoader),
superclassChain(ClassName, CurrentLoader, Chain),
Chain \= []. instanceMethodInitialThisType(Class, Method, class(ClassName, L)) :-
methodName(Method, MethodName),
MethodName \= '<init>',
classDefiningLoader(Class, L),
classClassName(Class, ClassName).

We now compute whether the merged stream for a method is type correct, using the method's initial type state:

  • If we have a stack map frame and an incoming type state, the type state must be assignable to the one in the stack map frame. We may then proceed to type check the rest of the stream with the type state given in the stack map frame.

    mergedCodeIsTypeSafe(Environment, [stackMap(Offset, MapFrame) | MoreCode],
    frame(Locals, OperandStack, Flags)) :-
    frameIsAssignable(frame(Locals, OperandStack, Flags), MapFrame),
    mergedCodeIsTypeSafe(Environment, MoreCode, MapFrame).
  • A merged code stream is type safe relative to an incoming type state T if it begins with an instruction I that is type safe relative to T, and I satisfies its exception handlers (see below), and the tail of the stream is type safe given the type state following that execution of I.

    NextStackFrame indicates what falls through to the following instruction. For an unconditional branch instruction, it will have the special value afterGotoExceptionStackFrame indicates what is passed to exception handlers.

    mergedCodeIsTypeSafe(Environment, [instruction(Offset, Parse) | MoreCode],
    frame(Locals, OperandStack, Flags)) :-
    instructionIsTypeSafe(Parse, Environment, Offset,
    frame(Locals, OperandStack, Flags),
    NextStackFrame, ExceptionStackFrame),
    instructionSatisfiesHandlers(Environment, Offset, ExceptionStackFrame),
    mergedCodeIsTypeSafe(Environment, MoreCode, NextStackFrame).
  • After an unconditional branch (indicated by an incoming type state of afterGoto), if we have a stack map frame giving the type state for the following instructions, we can proceed and type check them using the type state provided by the stack map frame.

    mergedCodeIsTypeSafe(Environment, [stackMap(Offset, MapFrame) | MoreCode],
    afterGoto) :-
    mergedCodeIsTypeSafe(Environment, MoreCode, MapFrame).
  • It is illegal to have code after an unconditional branch without a stack map frame being provided for it.

    mergedCodeIsTypeSafe(_Environment, [instruction(_, _) | _MoreCode],
    afterGoto) :-
    write_ln('No stack frame after unconditional branch'),
    fail.
  • If we have an unconditional branch at the end of the code, stop.

    mergedCodeIsTypeSafe(_Environment, [endOfCode(Offset)],
    afterGoto).

Branching to a target is type safe if the target has an associated stack frame, Frame, and the current stack frame, StackFrame, is assignable to Frame.

targetIsTypeSafe(Environment, StackFrame, Target) :-
offsetStackFrame(Environment, Target, Frame),
frameIsAssignable(StackFrame, Frame).

An instruction satisfies its exception handlers if it satisfies every exception handler that is applicable to the instruction.

instructionSatisfiesHandlers(Environment, Offset, ExceptionStackFrame) :-
exceptionHandlers(Environment, Handlers),
sublist(isApplicableHandler(Offset), Handlers, ApplicableHandlers),
checklist(instructionSatisfiesHandler(Environment, ExceptionStackFrame),
ApplicableHandlers).

An exception handler is applicable to an instruction if the offset of the instruction is greater or equal to the start of the handler's range and less than the end of the handler's range.

isApplicableHandler(Offset, handler(Start, End, _Target, _ClassName)) :-
Offset >= Start,
Offset < End.

An instruction satisfies an exception handler if the instructions's outgoing type state is ExcStackFrame, and the handler's target (the initial instruction of the handler code) is type safe assuming an incoming type state T. The type state T is derived from ExcStackFrame by replacing the operand stack with a stack whose sole element is the handler's exception class.

instructionSatisfiesHandler(Environment, ExcStackFrame, Handler) :-
Handler = handler(_, _, Target, _),
currentClassLoader(Environment, CurrentLoader),
handlerExceptionClass(Handler, ExceptionClass, CurrentLoader),
/* The stack consists of just the exception. */
ExcStackFrame = frame(Locals, _, Flags),
TrueExcStackFrame = frame(Locals, [ ExceptionClass ], Flags),
operandStackHasLegalLength(Environment, TrueExcStackFrame),
targetIsTypeSafe(Environment, TrueExcStackFrame, Target).

4.10.1.7. Type Checking Load and Store Instructions

All load instructions are variations on a common pattern, varying the type of the value that the instruction loads.

Loading a value of type Type from local variable Index is type safe, if the type of that local variable is ActualTypeActualType is assignable to Type, and pushing ActualType onto the incoming operand stack is a valid type transition (§4.10.1.4) that yields a new type state NextStackFrame. After execution of the load instruction, the type state will be NextStackFrame.

loadIsTypeSafe(Environment, Index, Type, StackFrame, NextStackFrame) :-
StackFrame = frame(Locals, _OperandStack, _Flags),
nth0(Index, Locals, ActualType),
isAssignable(ActualType, Type),
validTypeTransition(Environment, [], ActualType, StackFrame,
NextStackFrame).

All store instructions are variations on a common pattern, varying the type of the value that the instruction stores.

In general, a store instruction is type safe if the local variable it references is of a type that is a supertype of Type, and the top of the operand stack is of a subtype of Type, where Type is the type the instruction is designed to store.

More precisely, the store is type safe if one can pop a type ActualType that "matches" Type (that is, is a subtype of Type) off the operand stack (§4.10.1.4), and then legally assign that type the local variable LIndex.

storeIsTypeSafe(_Environment, Index, Type,
frame(Locals, OperandStack, Flags),
frame(NextLocals, NextOperandStack, Flags)) :-
popMatchingType(OperandStack, Type, NextOperandStack, ActualType),
modifyLocalVariable(Index, ActualType, Locals, NextLocals).

Given local variables Locals, modifying Index to have type Type results in the local variable list NewLocals. The modifications are somewhat involved, because some values (and their corresponding types) occupy two local variables. Hence, modifying LN may require modifying LN+1 (because the type will occupy both the N and N+1 slots) or LN-1 (because local N used to be the upper half of the two word value/type starting at local N-1, and so local N-1 must be invalidated), or both. This is described further below. We start at L0 and count up.

modifyLocalVariable(Index, Type, Locals, NewLocals) :-
modifyLocalVariable(0, Index, Type, Locals, NewLocals).

Given LocalsRest, the suffix of the local variable list starting at index I, modifying local variable Index to have type Type results in the local variable list suffix NextLocalsRest.

If I < Index-1, just copy the input to the output and recurse forward. If I = Index-1, the type of local I may change. This can occur if LI has a type of size 2. Once we set LI+1 to the new type (and the corresponding value), the type/value of LI will be invalidated, as its upper half will be trashed. Then we recurse forward.

modifyLocalVariable(I, Index, Type,
[Locals1 | LocalsRest],
[Locals1 | NextLocalsRest] ) :-
I < Index - 1,
I1 is I + 1,
modifyLocalVariable(I1, Index, Type, LocalsRest, NextLocalsRest). modifyLocalVariable(I, Index, Type,
[Locals1 | LocalsRest],
[NextLocals1 | NextLocalsRest] ) :-
I =:= Index - 1,
modifyPreIndexVariable(Locals1, NextLocals1),
modifyLocalVariable(Index, Index, Type, LocalsRest, NextLocalsRest).

When we find the variable, and it only occupies one word, we change it to Type and we're done. When we find the variable, and it occupies two words, we change its type to Type and the next word to top.

modifyLocalVariable(Index, Index, Type,
[_ | LocalsRest], [Type | LocalsRest]) :-
sizeOf(Type, 1). modifyLocalVariable(Index, Index, Type,
[_, _ | LocalsRest], [Type, top | LocalsRest]) :-
sizeOf(Type, 2).

We refer to a local whose index immediately precedes a local whose type will be modified as a pre-index variable. The future type of a pre-index variable of type InputType is Result. If the type, Type, of the pre-index local is of size 1, it doesn't change. If the type of the pre-index local, Type, is 2, we need to mark the lower half of its two word value as unusable, by setting its type to top.

modifyPreIndexVariable(Type, Type) :- sizeOf(Type, 1).
modifyPreIndexVariable(Type, top) :- sizeOf(Type, 2).

4.10.1.8. Type Checking for protected Members

All instructions that access members must contend with the rules concerning protected members. This section describes the protected check that corresponds to JLS §6.6.2.1.

The protected check applies only to protected members of superclasses of the current class. protected members in other classes will be caught by the access checking done at resolution (§5.4.4). There are four cases:

  • If the name of a class is not the name of any superclass, it cannot be a superclass, and so it can safely be ignored.

    passesProtectedCheck(Environment, MemberClassName, MemberName,
    MemberDescriptor, StackFrame) :-
    thisClass(Environment, class(CurrentClassName, CurrentLoader)),
    superclassChain(CurrentClassName, CurrentLoader, Chain),
    notMember(class(MemberClassName, _), Chain).
  • If the MemberClassName is the same as the name of a superclass, the class being resolved may indeed be a superclass. In this case, if no superclass named MemberClassName in a different run-time package has a protected member named MemberName with descriptor MemberDescriptor, the protected check does not apply.

    This is because the actual class being resolved will either be one of these superclasses, in which case we know that it is either in the same run-time package, and the access is legal; or the member in question is not protected and the check does not apply; or it will be a subclass, in which case the check would succeed anyway; or it will be some other class in the same run-time package, in which case the access is legal and the check need not take place; or the verifier need not flag this as a problem, since it will be caught anyway because resolution will per force fail.

    passesProtectedCheck(Environment, MemberClassName, MemberName,
    MemberDescriptor, StackFrame) :-
    thisClass(Environment, class(CurrentClassName, CurrentLoader)),
    superclassChain(CurrentClassName, CurrentLoader, Chain),
    member(class(MemberClassName, _), Chain),
    classesInOtherPkgWithProtectedMember(
    class(CurrentClassName, CurrentLoader),
    MemberName, MemberDescriptor, MemberClassName, Chain, []).
  • If there does exist a protected superclass member in a different run-time package, then load MemberClassName; if the member in question is not protected, the check does not apply. (Using a superclass member that is not protected is trivially correct.)

    passesProtectedCheck(Environment, MemberClassName, MemberName,
    MemberDescriptor,
    frame(_Locals, [Target | Rest], _Flags)) :-
    thisClass(Environment, class(CurrentClassName, CurrentLoader)),
    superclassChain(CurrentClassName, CurrentLoader, Chain),
    member(class(MemberClassName, _), Chain),
    classesInOtherPkgWithProtectedMember(
    class(CurrentClassName, CurrentLoader),
    MemberName, MemberDescriptor, MemberClassName, Chain, List),
    List \= [],
    loadedClass(MemberClassName, CurrentLoader, ReferencedClass),
    isNotProtected(ReferencedClass, MemberName, MemberDescriptor).
  • Otherwise, use of a member of an object of type Target requires that Target be assignable to the type of the current class.

    passesProtectedCheck(Environment, MemberClassName, MemberName,
    MemberDescriptor,
    frame(_Locals, [Target | Rest], _Flags)) :-
    thisClass(Environment, class(CurrentClassName, CurrentLoader)),
    superclassChain(CurrentClassName, CurrentLoader, Chain),
    member(class(MemberClassName, _), Chain),
    classesInOtherPkgWithProtectedMember(
    class(CurrentClassName, CurrentLoader),
    MemberName, MemberDescriptor, MemberClassName, Chain, List),
    List \= [],
    loadedClass(MemberClassName, CurrentLoader, ReferencedClass),
    isProtected(ReferencedClass, MemberName, MemberDescriptor),
    isAssignable(Target, class(CurrentClassName, CurrentLoader)).

The predicate classesInOtherPkgWithProtectedMember(Class, MemberName, MemberDescriptor, MemberClassName, Chain, List) is true if List is the set of classes in Chain with name MemberClassName that are in a different run-time package than Class which have a protected member named MemberName with descriptor MemberDescriptor.

classesInOtherPkgWithProtectedMember(_, _, _, _, [], []).

classesInOtherPkgWithProtectedMember(Class, MemberName,
MemberDescriptor, MemberClassName,
[class(MemberClassName, L) | Tail],
[class(MemberClassName, L) | T]) :-
differentRuntimePackage(Class, class(MemberClassName, L)),
loadedClass(MemberClassName, L, Super),
isProtected(Super, MemberName, MemberDescriptor),
classesInOtherPkgWithProtectedMember(
Class, MemberName, MemberDescriptor, MemberClassName, Tail, T). classesInOtherPkgWithProtectedMember(Class, MemberName,
MemberDescriptor, MemberClassName,
[class(MemberClassName, L) | Tail],
T) :-
differentRuntimePackage(Class, class(MemberClassName, L)),
loadedClass(MemberClassName, L, Super),
isNotProtected(Super, MemberName, MemberDescriptor),
classesInOtherPkgWithProtectedMember(
Class, MemberName, MemberDescriptor, MemberClassName, Tail, T). classesInOtherPkgWithProtectedMember(Class, MemberName,
MemberDescriptor, MemberClassName,
[class(MemberClassName, L) | Tail],
T] :-
sameRuntimePackage(Class, class(MemberClassName, L)),
classesInOtherPkgWithProtectedMember(
Class, MemberName, MemberDescriptor, MemberClassName, Tail, T). sameRuntimePackage(Class1, Class2) :-
classDefiningLoader(Class1, L),
classDefiningLoader(Class2, L),
samePackageName(Class1, Class2). differentRuntimePackage(Class1, Class2) :-
classDefiningLoader(Class1, L1),
classDefiningLoader(Class2, L2),
L1 \= L2. differentRuntimePackage(Class1, Class2) :-
differentPackageName(Class1, Class2).

4.10.1.9. Type Checking Instructions

In general, the type rule for an instruction is given relative to an environment Environment that defines the class and method in which the instruction occurs (§4.10.1.1), and the offset Offsetwithin the method at which the instruction occurs. The rule states that if the incoming type state StackFrame fulfills certain requirements, then:

  • The instruction is type safe.

  • It is provable that the type state after the instruction completes normally has a particular form given by NextStackFrame, and that the type state after the instruction completes abruptly is given by ExceptionStackFrame.

    The type state after an instruction completes abruptly is the same as the incoming type state, except that the operand stack is empty.

    exceptionStackFrame(StackFrame, ExceptionStackFrame) :-
    StackFrame = frame(Locals, _OperandStack, Flags),
    ExceptionStackFrame = frame(Locals, [], Flags).

Many instructions have type rules that are completely isomorphic to the rules for other instructions. If an instruction b1 is isomorphic to another instruction b2, then the type rule for b1 is the same as the type rule for b2.

instructionIsTypeSafe(Instruction, Environment, Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
instructionHasEquivalentTypeRule(Instruction, IsomorphicInstruction),
instructionIsTypeSafe(IsomorphicInstruction, Environment, Offset,
StackFrame, NextStackFrame,
ExceptionStackFrame).

The English language description of each rule is intended to be readable, intuitive, and concise. As such, the description avoids repeating all the contextual assumptions given above. In particular:

  • The description does not explicitly mention the environment.

  • When the description speaks of the operand stack or local variables in the following, it is referring to the operand stack and local variable components of a type state: either the incoming type state or the outgoing one.

  • The type state after the instruction completes abruptly is almost always identical to the incoming type state. The description only discusses the type state after the instruction completes abruptly when that is not the case.

  • The description speaks of popping and pushing types onto the operand stack, and does not explicitly discuss issues of stack underflow or overflow. The description assumes these operations can be completed successfully, but the Prolog clauses for operand stack manipulation ensure that the necessary checks are made.

  • The description discusses only the manipulation of logical types. In practice, some types take more than one word. The description abstracts from these representation details, but the Prolog clauses that manipulate data do not.

Any ambiguities can be resolved by referring to the formal Prolog clauses.

aaload

An aaload instruction is type safe iff one can validly replace types matching int and an array type with component type ComponentType where ComponentType is a subtype of Object, with ComponentType yielding the outgoing type state.

instructionIsTypeSafe(aaload, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
nth1OperandStackIs(2, StackFrame, ArrayType),
arrayComponentType(ArrayType, ComponentType),
isBootstrapLoader(BL),
validTypeTransition(Environment,
[int, arrayOf(class('java/lang/Object', BL))],
ComponentType, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The component type of an array of X is X. We define the component type of null to be null.

arrayComponentType(arrayOf(X), X).
arrayComponentType(null, null).
aastore

An aastore instruction is type safe iff one can validly pop types matching Objectint, and an array of Object off the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(aastore, _Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
isBootstrapLoader(BL),
canPop(StackFrame,
[class('java/lang/Object', BL),
int,
arrayOf(class('java/lang/Object', BL))],
NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
aconst_null

An aconst_null instruction is type safe if one can validly push the type null onto the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(aconst_null, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [], null, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
aloadaload_<n>

An aload instruction with operand Index is type safe and yields an outgoing type state NextStackFrame, if a load instruction with operand Index and type reference is type safe and yields an outgoing type state NextStackFrame.

instructionIsTypeSafe(aload(Index), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
loadIsTypeSafe(Environment, Index, reference, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The instructions aload_<n>, for 0 ≤ n ≤ 3, are type safe iff the equivalent aload instruction is type safe.

instructionHasEquivalentTypeRule(aload_0, aload(0)).
instructionHasEquivalentTypeRule(aload_1, aload(1)).
instructionHasEquivalentTypeRule(aload_2, aload(2)).
instructionHasEquivalentTypeRule(aload_3, aload(3)).
anewarray

An anewarray instruction with operand CP is type safe iff CP refers to a constant pool entry denoting a class, interface, or array type, and one can legally replace a type matching int on the incoming operand stack with an array with component type CP yielding the outgoing type state.

instructionIsTypeSafe(anewarray(CP), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
(CP = class(_, _) ; CP = arrayOf(_)),
validTypeTransition(Environment, [int], arrayOf(CP),
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
areturn

An areturn instruction is type safe iff the enclosing method has a declared return type, ReturnType, that is a reference type, and one can validly pop a type matching ReturnType off the incoming operand stack.

instructionIsTypeSafe(areturn, Environment, _Offset, StackFrame,
afterGoto, ExceptionStackFrame) :-
thisMethodReturnType(Environment, ReturnType),
isAssignable(ReturnType, reference),
canPop(StackFrame, [ReturnType], _PoppedStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
arraylength

An arraylength instruction is type safe iff one can validly replace an array type on the incoming operand stack with the type int yielding the outgoing type state.

instructionIsTypeSafe(arraylength, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
nth1OperandStackIs(1, StackFrame, ArrayType),
arrayComponentType(ArrayType, _),
validTypeTransition(Environment, [top], int, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
astoreastore_<n>

An astore instruction with operand Index is type safe and yields an outgoing type state NextStackFrame, if a store instruction with operand Index and type reference is type safe and yields an outgoing type state NextStackFrame.

instructionIsTypeSafe(astore(Index), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
storeIsTypeSafe(Environment, Index, reference, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The instructions astore_<n>, for 0 ≤ n ≤ 3, are type safe iff the equivalent astore instruction is type safe.

instructionHasEquivalentTypeRule(astore_0, astore(0)).
instructionHasEquivalentTypeRule(astore_1, astore(1)).
instructionHasEquivalentTypeRule(astore_2, astore(2)).
instructionHasEquivalentTypeRule(astore_3, astore(3)).
athrow

An athrow instruction is type safe iff the top of the operand stack matches Throwable.

instructionIsTypeSafe(athrow, _Environment, _Offset, StackFrame,
afterGoto, ExceptionStackFrame) :-
isBootstrapLoader(BL),
canPop(StackFrame, [class('java/lang/Throwable', BL)], _PoppedStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
baload

baload instruction is type safe iff one can validly replace types matching int and a small array type on the incoming operand stack with int yielding the outgoing type state.

instructionIsTypeSafe(baload, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :
nth1OperandStackIs(2, StackFrame, ArrayType),
isSmallArray(ArrayType),
validTypeTransition(Environment, [int, top], int,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

An array type is a small array type if it is an array of byte, an array of boolean, or a subtype thereof (null).

isSmallArray(arrayOf(byte)).
isSmallArray(arrayOf(boolean)).
isSmallArray(null).
bastore

bastore instruction is type safe iff one can validly pop types matching intint and a small array type off the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(bastore, _Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
nth1OperandStackIs(3, StackFrame, ArrayType),
isSmallArray(ArrayType),
canPop(StackFrame, [int, int, top], NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
bipush

bipush instruction is type safe iff the equivalent sipush instruction is type safe.

instructionHasEquivalentTypeRule(bipush(Value), sipush(Value)).
caload

caload instruction is type safe iff one can validly replace types matching int and array of char on the incoming operand stack with int yielding the outgoing type state.

instructionIsTypeSafe(caload, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [int, arrayOf(char)], int,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
castore

castore instruction is type safe iff one can validly pop types matching intint and array of char off the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(castore, _Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
canPop(StackFrame, [int, int, arrayOf(char)], NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
checkcast

checkcast instruction with operand CP is type safe iff CP refers to a constant pool entry denoting either a class or an array, and one can validly replace the type Object on top of the incoming operand stack with the type denoted by CP yielding the outgoing type state.

instructionIsTypeSafe(checkcast(CP), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
(CP = class(_, _) ; CP = arrayOf(_)),
isBootstrapLoader(BL),
validTypeTransition(Environment, [class('java/lang/Object', BL)], CP,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
d2fd2id2l

d2f instruction is type safe if one can validly pop double off the incoming operand stack and replace it with float, yielding the outgoing type state.

instructionIsTypeSafe(d2f, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [double], float,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

d2i instruction is type safe if one can validly pop double off the incoming operand stack and replace it with int, yielding the outgoing type state.

instructionIsTypeSafe(d2i, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [double], int,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

d2l instruction is type safe if one can validly pop double off the incoming operand stack and replace it with long, yielding the outgoing type state.

instructionIsTypeSafe(d2l, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [double], long,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
dadd

dadd instruction is type safe iff one can validly replace types matching double and double on the incoming operand stack with double yielding the outgoing type state.

instructionIsTypeSafe(dadd, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [double, double], double,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
daload

daload instruction is type safe iff one can validly replace types matching int and array of double on the incoming operand stack with double yielding the outgoing type state.

instructionIsTypeSafe(daload, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [int, arrayOf(double)], double,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
dastore

dastore instruction is type safe iff one can validly pop types matching doubleint and array of double off the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(dastore, _Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
canPop(StackFrame, [double, int, arrayOf(double)], NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
dcmp<op>

dcmpg instruction is type safe iff one can validly replace types matching double and double on the incoming operand stack with int yielding the outgoing type state.

instructionIsTypeSafe(dcmpg, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [double, double], int,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

dcmpl instruction is type safe iff the equivalent dcmpg instruction is type safe.

instructionHasEquivalentTypeRule(dcmpl, dcmpg).
dconst_<d>

dconst_0 instruction is type safe if one can validly push the type double onto the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(dconst_0, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [], double, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

dconst_1 instruction is type safe iff the equivalent dconst_0 instruction is type safe.

instructionHasEquivalentTypeRule(dconst_1, dconst_0).
ddiv

ddiv instruction is type safe iff the equivalent dadd instruction is type safe.

instructionHasEquivalentTypeRule(ddiv, dadd).
dloaddload_<n>

dload instruction with operand Index is type safe and yields an outgoing type state NextStackFrame, if a load instruction with operand Index and type double is type safe and yields an outgoing type state NextStackFrame.

instructionIsTypeSafe(dload(Index), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
loadIsTypeSafe(Environment, Index, double, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The instructions dload_<n>, for 0 ≤ n ≤ 3, are typesafe iff the equivalent dload instruction is type safe.

instructionHasEquivalentTypeRule(dload_0, dload(0)).
instructionHasEquivalentTypeRule(dload_1, dload(1)).
instructionHasEquivalentTypeRule(dload_2, dload(2)).
instructionHasEquivalentTypeRule(dload_3, dload(3)).
dmul

dmul instruction is type safe iff the equivalent dadd instruction is type safe.

instructionHasEquivalentTypeRule(dmul, dadd).
dneg

dneg instruction is type safe iff there is a type matching double on the incoming operand stack. The dneg instruction does not alter the type state.

instructionIsTypeSafe(dneg, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [double], double,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
drem

drem instruction is type safe iff the equivalent dadd instruction is type safe.

instructionHasEquivalentTypeRule(drem, dadd).
dreturn

dreturn instruction is type safe if the enclosing method has a declared return type of double, and one can validly pop a type matching double off the incoming operand stack.

instructionIsTypeSafe(dreturn, Environment, _Offset, StackFrame,
afterGoto, ExceptionStackFrame) :-
thisMethodReturnType(Environment, double),
canPop(StackFrame, [double], _PoppedStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
dstoredstore_<n>

dstore instruction with operand Index is type safe and yields an outgoing type state NextStackFrame, if a store instruction with operand Index and type double is type safe and yields an outgoing type state NextStackFrame.

instructionIsTypeSafe(dstore(Index), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
storeIsTypeSafe(Environment, Index, double, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The instructions dstore_<n>, for 0 ≤ n ≤ 3, are type safe iff the equivalent dstore instruction is type safe.

instructionHasEquivalentTypeRule(dstore_0, dstore(0)).
instructionHasEquivalentTypeRule(dstore_1, dstore(1)).
instructionHasEquivalentTypeRule(dstore_2, dstore(2)).
instructionHasEquivalentTypeRule(dstore_3, dstore(3)).
dsub

dsub instruction is type safe iff the equivalent dadd instruction is type safe.

instructionHasEquivalentTypeRule(dsub, dadd).
dup

dup instruction is type safe iff one can validly replace a category 1 type, Type, with the types TypeType, yielding the outgoing type state.

instructionIsTypeSafe(dup, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
StackFrame = frame(Locals, InputOperandStack, Flags),
popCategory1(InputOperandStack, Type, _),
canSafelyPush(Environment, InputOperandStack, Type, OutputOperandStack),
NextStackFrame = frame(Locals, OutputOperandStack, Flags),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
dup_x1

dup_x1 instruction is type safe iff one can validly replace two category 1 types, Type1, and Type2, on the incoming operand stack with the types Type1Type2Type1, yielding the outgoing type state.

instructionIsTypeSafe(dup_x1, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
StackFrame = frame(Locals, InputOperandStack, Flags),
popCategory1(InputOperandStack, Type1, Stack1),
popCategory1(Stack1, Type2, Rest),
canSafelyPushList(Environment, Rest, [Type1, Type2, Type1],
OutputOperandStack),
NextStackFrame = frame(Locals, OutputOperandStack, Flags),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
dup_x2

dup_x2 instruction is type safe iff it is a type safe form of the dup_x2 instruction.

instructionIsTypeSafe(dup_x2, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
StackFrame = frame(Locals, InputOperandStack, Flags),
dup_x2FormIsTypeSafe(Environment, InputOperandStack, OutputOperandStack),
NextStackFrame = frame(Locals, OutputOperandStack, Flags),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

dup_x2 instruction is a type safe form of the dup_x2 instruction iff it is a type safe form 1 dup_x2 instruction or a type safe form 2 dup_x2 instruction.

dup_x2FormIsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
dup_x2Form1IsTypeSafe(Environment, InputOperandStack, OutputOperandStack). dup_x2FormIsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
dup_x2Form2IsTypeSafe(Environment, InputOperandStack, OutputOperandStack).

dup_x2 instruction is a type safe form 1 dup_x2 instruction iff one can validly replace three category 1 types, Type1Type2Type3 on the incoming operand stack with the types Type1Type2Type3Type1, yielding the outgoing type state.

dup_x2Form1IsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
popCategory1(InputOperandStack, Type1, Stack1),
popCategory1(Stack1, Type2, Stack2),
popCategory1(Stack2, Type3, Rest),
canSafelyPushList(Environment, Rest, [Type1, Type3, Type2, Type1],
OutputOperandStack).

dup_x2 instruction is a type safe form 2 dup_x2 instruction iff one can validly replace a category 1 type, Type1, and a category 2 type, Type2, on the incoming operand stack with the types Type1Type2Type1, yielding the outgoing type state.

dup_x2Form2IsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
popCategory1(InputOperandStack, Type1, Stack1),
popCategory2(Stack1, Type2, Rest),
canSafelyPushList(Environment, Rest, [Type1, Type2, Type1],
OutputOperandStack).
dup2

dup2 instruction is type safe iff it is a type safe form of the dup2 instruction.

instructionIsTypeSafe(dup2, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
StackFrame = frame(Locals, InputOperandStack, Flags),
dup2FormIsTypeSafe(Environment,InputOperandStack, OutputOperandStack),
NextStackFrame = frame(Locals, OutputOperandStack, Flags),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

dup2 instruction is a type safe form of the dup2 instruction iff it is a type safe form 1 dup2 instruction or a type safe form 2 dup2 instruction.

dup2FormIsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
dup2Form1IsTypeSafe(Environment,InputOperandStack, OutputOperandStack). dup2FormIsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
dup2Form2IsTypeSafe(Environment,InputOperandStack, OutputOperandStack).

dup2 instruction is a type safe form 1 dup2 instruction iff one can validly replace two category 1 types, Type1 and Type2 on the incoming operand stack with the types Type1Type2Type1Type2, yielding the outgoing type state.

dup2Form1IsTypeSafe(Environment, InputOperandStack, OutputOperandStack):-
popCategory1(InputOperandStack, Type1, TempStack),
popCategory1(TempStack, Type2, _),
canSafelyPushList(Environment, InputOperandStack, [Type1, Type2],
OutputOperandStack).

dup2 instruction is a type safe form 2 dup2 instruction iff one can validly replace a category 2 type, Type on the incoming operand stack with the types TypeType, yielding the outgoing type state.

dup2Form2IsTypeSafe(Environment, InputOperandStack, OutputOperandStack):-
popCategory2(InputOperandStack, Type, _),
canSafelyPush(Environment, InputOperandStack, Type, OutputOperandStack).
dup2_x1

dup2_x1 instruction is type safe iff it is a type safe form of the dup2_x1 instruction.

instructionIsTypeSafe(dup2_x1, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
StackFrame = frame(Locals, InputOperandStack, Flags),
dup2_x1FormIsTypeSafe(Environment, InputOperandStack, OutputOperandStack),
NextStackFrame = frame(Locals, OutputOperandStack, Flags),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

dup2_x1 instruction is a type safe form of the dup2_x1 instruction iff it is a type safe form 1 dup2_x1 instruction or a type safe form 2 dup_x2 instruction.

dup2_x1FormIsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
dup2_x1Form1IsTypeSafe(Environment, InputOperandStack, OutputOperandStack). dup2_x1FormIsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
dup2_x1Form2IsTypeSafe(Environment, InputOperandStack, OutputOperandStack).

dup2_x1 instruction is a type safe form 1 dup2_x1 instruction iff one can validly replace three category 1 types, Type1Type2Type3, on the incoming operand stack with the types Type1Type2Type3Type1Type2, yielding the outgoing type state.

dup2_x1Form1IsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
popCategory1(InputOperandStack, Type1, Stack1),
popCategory1(Stack1, Type2, Stack2),
popCategory1(Stack2, Type3, Rest),
canSafelyPushList(Environment, Rest, [Type2, Type1, Type3, Type2, Type1],
OutputOperandStack).

dup2_x1 instruction is a type safe form 2 dup2_x1 instruction iff one can validly replace a category 2 type, Type1, and a category 1 type, Type2, on the incoming operand stack with the types Type1Type2Type1, yielding the outgoing type state.

dup2_x1Form2IsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
popCategory2(InputOperandStack, Type1, Stack1),
popCategory1(Stack1, Type2, Rest),
canSafelyPushList(Environment, Rest, [Type1, Type2, Type1],
OutputOperandStack).
dup2_x2

dup2_x2 instruction is type safe iff it is a type safe form of the dup2_x2 instruction.

instructionIsTypeSafe(dup2_x2, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
StackFrame = frame(Locals, InputOperandStack, Flags),
dup2_x2FormIsTypeSafe(Environment, InputOperandStack, OutputOperandStack),
NextStackFrame = frame(Locals, OutputOperandStack, Flags),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

dup2_x2 instruction is a type safe form of the dup2_x2 instruction iff one of the following holds:

  • it is a type safe form 1 dup2_x2 instruction.

  • it is a type safe form 2 dup2_x2 instruction.

  • it is a type safe form 3 dup2_x2 instruction.

  • it is a type safe form 4 dup2_x2 instruction.

dup2_x2FormIsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
dup2_x2Form1IsTypeSafe(Environment, InputOperandStack, OutputOperandStack). dup2_x2FormIsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
dup2_x2Form2IsTypeSafe(Environment, InputOperandStack, OutputOperandStack). dup2_x2FormIsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
dup2_x2Form3IsTypeSafe(Environment, InputOperandStack, OutputOperandStack). dup2_x2FormIsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
dup2_x2Form4IsTypeSafe(Environment, InputOperandStack, OutputOperandStack).

dup2_x2 instruction is a type safe form 1 dup2_x2 instruction iff one can validly replace four category 1 types, Type1Type2Type3Type4, on the incoming operand stack with the types Type1Type2Type3Type4Type1Type2, yielding the outgoing type state.

dup2_x2Form1IsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
popCategory1(InputOperandStack, Type1, Stack1),
popCategory1(Stack1, Type2, Stack2),
popCategory1(Stack2, Type3, Stack3),
popCategory1(Stack3, Type4, Rest),
canSafelyPushList(Environment, Rest,
[Type2, Type1, Type4, Type3, Type2, Type1],
OutputOperandStack).

dup2_x2 instruction is a type safe form 2 dup2_x2 instruction iff one can validly replace a category 2 type, Type1, and two category 1 types, Type2Type3, on the incoming operand stack with the types Type1Type2Type3Type1, yielding the outgoing type state.

dup2_x2Form2IsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
popCategory2(InputOperandStack, Type1, Stack1),
popCategory1(Stack1, Type2, Stack2),
popCategory1(Stack2, Type3, Rest),
canSafelyPushList(Environment, Rest,
[Type1, Type3, Type2, Type1],
OutputOperandStack).

dup2_x2 instruction is a type safe form 3 dup2_x2 instruction iff one can validly replace two category 1 types, Type1Type2, and a category 2 type, Type3, on the incoming operand stack with the types Type1Type2Type3Type1Type2, yielding the outgoing type state.

dup2_x2Form3IsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
popCategory1(InputOperandStack, Type1, Stack1),
popCategory1(Stack1, Type2, Stack2),
popCategory2(Stack2, Type3, Rest),
canSafelyPushList(Environment, Rest,
[Type2, Type1, Type3, Type2, Type1],
OutputOperandStack).

dup2_x2 instruction is a type safe form 4 dup2_x2 instruction iff one can validly replace two category 2 types, Type1Type2, on the incoming operand stack with the types Type1Type2Type1, yielding the outgoing type state.

dup2_x2Form4IsTypeSafe(Environment, InputOperandStack, OutputOperandStack) :-
popCategory2(InputOperandStack, Type1, Stack1),
popCategory2(Stack1, Type2, Rest),
canSafelyPushList(Environment, Rest, [Type1, Type2, Type1],
OutputOperandStack).
f2df2if2l

An f2d instruction is type safe if one can validly pop float off the incoming operand stack and replace it with double, yielding the outgoing type state.

instructionIsTypeSafe(f2d, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [float], double,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

An f2i instruction is type safe if one can validly pop float off the incoming operand stack and replace it with int, yielding the outgoing type state.

instructionIsTypeSafe(f2i, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [float], int,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

An f2l instruction is type safe if one can validly pop float off the incoming operand stack and replace it with long, yielding the outgoing type state.

instructionIsTypeSafe(f2l, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [float], long,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
fadd

An fadd instruction is type safe iff one can validly replace types matching float and float on the incoming operand stack with float yielding the outgoing type state.

instructionIsTypeSafe(fadd, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [float, float], float,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
faload

An faload instruction is type safe iff one can validly replace types matching int and array of float on the incoming operand stack with float yielding the outgoing type state.

instructionIsTypeSafe(faload, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [int, arrayOf(float)], float,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
fastore

An fastore instruction is type safe iff one can validly pop types matching floatint and array of float off the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(fastore, _Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
canPop(StackFrame, [float, int, arrayOf(float)], NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
fcmp<op>

An fcmpg instruction is type safe iff one can validly replace types matching float and float on the incoming operand stack with int yielding the outgoing type state.

instructionIsTypeSafe(fcmpg, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [float, float], int,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

An fcmpl instruction is type safe iff the equivalent fcmpg instruction is type safe.

instructionHasEquivalentTypeRule(fcmpl, fcmpg).
fconst_<f>

An fconst_0 instruction is type safe if one can validly push the type float onto the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(fconst_0, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [], float, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The rules for the other variants of fconst are equivalent.

instructionHasEquivalentTypeRule(fconst_1, fconst_0).
instructionHasEquivalentTypeRule(fconst_2, fconst_0).
fdiv

An fdiv instruction is type safe iff the equivalent fadd instruction is type safe.

instructionHasEquivalentTypeRule(fdiv, fadd).
floadfload_<n>

An fload instruction with operand Index is type safe and yields an outgoing type state NextStackFrame, if a load instruction with operand Index and type float is type safe and yields an outgoing type state NextStackFrame.

instructionIsTypeSafe(fload(Index), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
loadIsTypeSafe(Environment, Index, float, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The instructions fload_<n>, for 0 ≤ n ≤ 3, are typesafe iff the equivalent fload instruction is type safe.

instructionHasEquivalentTypeRule(fload_0, fload(0)).
instructionHasEquivalentTypeRule(fload_1, fload(1)).
instructionHasEquivalentTypeRule(fload_2, fload(2)).
instructionHasEquivalentTypeRule(fload_3, fload(3)).
fmul

An fmul instruction is type safe iff the equivalent fadd instruction is type safe.

instructionHasEquivalentTypeRule(fmul, fadd).
fneg

An fneg instruction is type safe iff there is a type matching float on the incoming operand stack. The fneg instruction does not alter the type state.

instructionIsTypeSafe(fneg, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [float], float,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
frem

An frem instruction is type safe iff the equivalent fadd instruction is type safe.

instructionHasEquivalentTypeRule(frem, fadd).
freturn

An freturn instruction is type safe if the enclosing method has a declared return type of float, and one can validly pop a type matching float off the incoming operand stack.

instructionIsTypeSafe(freturn, Environment, _Offset, StackFrame,
afterGoto, ExceptionStackFrame) :-
thisMethodReturnType(Environment, float),
canPop(StackFrame, [float], _PoppedStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
fstorefstore_<n>

An fstore instruction with operand Index is type safe and yields an outgoing type state NextStackFrame, if a store instruction with operand Index and type float is type safe and yields an outgoing type state NextStackFrame.

instructionIsTypeSafe(fstore(Index), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
storeIsTypeSafe(Environment, Index, float, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The instructions fstore_<n>, for 0 ≤ n ≤ 3, are typesafe iff the equivalent fstore instruction is type safe.

instructionHasEquivalentTypeRule(fstore_0, fstore(0)).
instructionHasEquivalentTypeRule(fstore_1, fstore(1)).
instructionHasEquivalentTypeRule(fstore_2, fstore(2)).
instructionHasEquivalentTypeRule(fstore_3, fstore(3)).
fsub

An fsub instruction is type safe iff the equivalent fadd instruction is type safe.

instructionHasEquivalentTypeRule(fsub, fadd).
getfield

getfield instruction with operand CP is type safe iff CP refers to a constant pool entry denoting a field whose declared type is FieldType, declared in a class FieldClassName, and one can validly replace a type matching FieldClassName with type FieldType on the incoming operand stack yielding the outgoing type state. FieldClassName must not be an array type. protected fields are subject to additional checks (§4.10.1.8).

instructionIsTypeSafe(getfield(CP), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
CP = field(FieldClassName, FieldName, FieldDescriptor),
parseFieldDescriptor(FieldDescriptor, FieldType),
passesProtectedCheck(Environment, FieldClassName, FieldName,
FieldDescriptor, StackFrame),
currentClassLoader(Environment, CurrentLoader),
validTypeTransition(Environment,
[class(FieldClassName, CurrentLoader)], FieldType,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
getstatic

getstatic instruction with operand CP is type safe iff CP refers to a constant pool entry denoting a field whose declared type is FieldType, and one can validly push FieldType on the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(getstatic(CP), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
CP = field(_FieldClassName, _FieldName, FieldDescriptor),
parseFieldDescriptor(FieldDescriptor, FieldType),
validTypeTransition(Environment, [], FieldType,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
gotogoto_w

goto instruction is type safe iff its target operand is a valid branch target.

instructionIsTypeSafe(goto(Target), Environment, _Offset, StackFrame,
afterGoto, ExceptionStackFrame) :-
targetIsTypeSafe(Environment, StackFrame, Target),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

goto_w instruction is type safe iff the equivalent goto instruction is type safe.

instructionHasEquivalentTypeRule(goto_w(Target), goto(Target)).
i2bi2ci2di2fi2li2s

An i2b instruction is type safe iff the equivalent ineg instruction is type safe.

instructionHasEquivalentTypeRule(i2b, ineg).

An i2c instruction is type safe iff the equivalent ineg instruction is type safe.

instructionHasEquivalentTypeRule(i2c, ineg).

An i2d instruction is type safe if one can validly pop int off the incoming operand stack and replace it with double, yielding the outgoing type state.

instructionIsTypeSafe(i2d, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [int], double,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

An i2f instruction is type safe if one can validly pop int off the incoming operand stack and replace it with float, yielding the outgoing type state.

instructionIsTypeSafe(i2f, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [int], float,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

An i2l instruction is type safe if one can validly pop int off the incoming operand stack and replace it with long, yielding the outgoing type state.

instructionIsTypeSafe(i2l, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [int], long,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

An i2s instruction is type safe iff the equivalent ineg instruction is type safe.

instructionHasEquivalentTypeRule(i2s, ineg).
iadd

An iadd instruction is type safe iff one can validly replace types matching int and int on the incoming operand stack with int yielding the outgoing type state.

instructionIsTypeSafe(iadd, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [int, int], int,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
iaload

An iaload instruction is type safe iff one can validly replace types matching int and array of int on the incoming operand stack with int yielding the outgoing type state.

instructionIsTypeSafe(iaload, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [int, arrayOf(int)], int,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
iand

An iand instruction is type safe iff the equivalent iadd instruction is type safe.

instructionHasEquivalentTypeRule(iand, iadd).
iastore

An iastore instruction is type safe iff one can validly pop types matching intint and array of int off the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(iastore, _Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
canPop(StackFrame, [int, int, arrayOf(int)], NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
iconst_<i>

An iconst_m1 instruction is type safe if one can validly push the type int onto the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(iconst_m1, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [], int, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The rules for the other variants of iconst are equivalent.

instructionHasEquivalentTypeRule(iconst_0, iconst_m1).
instructionHasEquivalentTypeRule(iconst_1, iconst_m1).
instructionHasEquivalentTypeRule(iconst_2, iconst_m1).
instructionHasEquivalentTypeRule(iconst_3, iconst_m1).
instructionHasEquivalentTypeRule(iconst_4, iconst_m1).
instructionHasEquivalentTypeRule(iconst_5, iconst_m1).
idiv

An idiv instruction is type safe iff the equivalent iadd instruction is type safe.

instructionHasEquivalentTypeRule(idiv, iadd).
if_acmp<cond>

An if_acmpeq instruction is type safe iff one can validly pop types matching reference and reference on the incoming operand stack yielding the outgoing type state NextStackFrame, and the operand of the instruction, Target, is a valid branch target assuming an incoming type state of NextStackFrame.

instructionIsTypeSafe(if_acmpeq(Target), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
canPop(StackFrame, [reference, reference], NextStackFrame),
targetIsTypeSafe(Environment, NextStackFrame, Target),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The rule for if_acmpne is identical.

instructionHasEquivalentTypeRule(if_acmpne(Target), if_acmpeq(Target)).
if_icmp<cond>

An if_icmpeq instruction is type safe iff one can validly pop types matching int and int on the incoming operand stack yielding the outgoing type state NextStackFrame, and the operand of the instruction, Target, is a valid branch target assuming an incoming type state of NextStackFrame.

instructionIsTypeSafe(if_icmpeq(Target), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
canPop(StackFrame, [int, int], NextStackFrame),
targetIsTypeSafe(Environment, NextStackFrame, Target),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The rules for all other variants of the if_icmp<cond> instruction are identical.

instructionHasEquivalentTypeRule(if_icmpge(Target), if_icmpeq(Target)).
instructionHasEquivalentTypeRule(if_icmpgt(Target), if_icmpeq(Target)).
instructionHasEquivalentTypeRule(if_icmple(Target), if_icmpeq(Target)).
instructionHasEquivalentTypeRule(if_icmplt(Target), if_icmpeq(Target)).
instructionHasEquivalentTypeRule(if_icmpne(Target), if_icmpeq(Target)).
if<cond>

An ifeq instruction is type safe iff one can validly pop a type matching int off the incoming operand stack yielding the outgoing type state NextStackFrame, and the operand of the instruction, Target, is a valid branch target assuming an incoming type state of NextStackFrame.

instructionIsTypeSafe(ifeq(Target), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
canPop(StackFrame, [int], NextStackFrame),
targetIsTypeSafe(Environment, NextStackFrame, Target),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The rules for all other variations of the if<cond> instruction are identical.

instructionHasEquivalentTypeRule(ifge(Target), ifeq(Target)).
instructionHasEquivalentTypeRule(ifgt(Target), ifeq(Target)).
instructionHasEquivalentTypeRule(ifle(Target), ifeq(Target)).
instructionHasEquivalentTypeRule(iflt(Target), ifeq(Target)).
instructionHasEquivalentTypeRule(ifne(Target), ifeq(Target)).
ifnonnullifnull

An ifnonnull instruction is type safe iff one can validly pop a type matching reference off the incoming operand stack yielding the outgoing type state NextStackFrame, and the operand of the instruction, Target, is a valid branch target assuming an incoming type state of NextStackFrame.

instructionIsTypeSafe(ifnonnull(Target), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
canPop(StackFrame, [reference], NextStackFrame),
targetIsTypeSafe(Environment, NextStackFrame, Target),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

An ifnull instruction is type safe iff the equivalent ifnonnull instruction is type safe.

instructionHasEquivalentTypeRule(ifnull(Target), ifnonnull(Target)).
iinc

An iinc instruction with first operand Index is type safe iff LIndex has type int. The iinc instruction does not change the type state.

instructionIsTypeSafe(iinc(Index, _Value), _Environment, _Offset,
StackFrame, StackFrame, ExceptionStackFrame) :-
StackFrame = frame(Locals, _OperandStack, _Flags),
nth0(Index, Locals, int),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
iloadiload_<n>

An iload instruction with operand Index is type safe and yields an outgoing type state NextStackFrame, if a load instruction with operand Index and type int is type safe and yields an outgoing type state NextStackFrame.

instructionIsTypeSafe(iload(Index), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
loadIsTypeSafe(Environment, Index, int, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The instructions iload_<n>, for 0 ≤ n ≤ 3, are typesafe iff the equivalent iload instruction is type safe.

instructionHasEquivalentTypeRule(iload_0, iload(0)).
instructionHasEquivalentTypeRule(iload_1, iload(1)).
instructionHasEquivalentTypeRule(iload_2, iload(2)).
instructionHasEquivalentTypeRule(iload_3, iload(3)).
imul

An imul instruction is type safe iff the equivalent iadd instruction is type safe.

instructionHasEquivalentTypeRule(imul, iadd).
ineg

An ineg instruction is type safe iff there is a type matching int on the incoming operand stack. The ineg instruction does not alter the type state.

instructionIsTypeSafe(ineg, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [int], int, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
instanceof

An instanceof instruction with operand CP is type safe iff CP refers to a constant pool entry denoting either a class or an array, and one can validly replace the type Object on top of the incoming operand stack with type int yielding the outgoing type state.

instructionIsTypeSafe(instanceof(CP), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
(CP = class(_, _) ; CP = arrayOf(_)),
isBootstrapLoader(BL),
validTypeTransition(Environment, [class('java/lang/Object', BL)], int,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
invokedynamic

An invokedynamic instruction is type safe iff all of the following are true:

  • Its first operand, CP, refers to a constant pool entry denoting an dynamic call site with name CallSiteName with descriptor Descriptor.

  • CallSiteName is not <init>.

  • CallSiteName is not <clinit>.

  • One can validly replace types matching the argument types given in Descriptor on the incoming operand stack with the return type given in Descriptor, yielding the outgoing type state.

instructionIsTypeSafe(invokedynamic(CP,0,0), Environment, _Offset,
StackFrame, NextStackFrame, ExceptionStackFrame) :-
CP = dmethod(CallSiteName, Descriptor),
CallSiteName \= '<init>',
CallSiteName \= '<clinit>',
parseMethodDescriptor(Descriptor, OperandArgList, ReturnType),
reverse(OperandArgList, StackArgList),
validTypeTransition(Environment, StackArgList, ReturnType,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
invokeinterface

An invokeinterface instruction is type safe iff all of the following are true:

  • Its first operand, CP, refers to a constant pool entry denoting an interface method named MethodName with descriptor Descriptor that is a member of an interface MethodIntfName.

  • MethodName is not <init>.

  • MethodName is not <clinit>.

  • Its second operand, Count, is a valid count operand (see below).

  • One can validly replace types matching the type MethodIntfName and the argument types given in Descriptor on the incoming operand stack with the return type given in Descriptor, yielding the outgoing type state.

instructionIsTypeSafe(invokeinterface(CP, Count, 0), Environment, _Offset,
StackFrame, NextStackFrame, ExceptionStackFrame) :-
CP = imethod(MethodIntfName, MethodName, Descriptor),
MethodName \= '<init>',
MethodName \= '<clinit>',
parseMethodDescriptor(Descriptor, OperandArgList, ReturnType),
currentClassLoader(Environment, CurrentLoader),
reverse([class(MethodIntfName, CurrentLoader) | OperandArgList],
StackArgList),
canPop(StackFrame, StackArgList, TempFrame),
validTypeTransition(Environment, [], ReturnType,
TempFrame, NextStackFrame),
countIsValid(Count, StackFrame, TempFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The Count operand of an invokeinterface instruction is valid if it equals the size of the arguments to the instruction. This is equal to the difference between the size of InputFrame and OutputFrame.

countIsValid(Count, InputFrame, OutputFrame) :-
InputFrame = frame(_Locals1, OperandStack1, _Flags1),
OutputFrame = frame(_Locals2, OperandStack2, _Flags2),
length(OperandStack1, Length1),
length(OperandStack2, Length2),
Count =:= Length1 - Length2.
invokespecial

An invokespecial instruction is type safe iff all of the following are true:

  • Its first operand, CP, refers to a constant pool entry denoting a method named MethodName with descriptor Descriptor that is a member of a class MethodClassName.

  • Either:

    • MethodName is not <init>.

    • MethodName is not <clinit>.

    • One can validly replace types matching the current class and the argument types given in Descriptor on the incoming operand stack with the return type given in Descriptor, yielding the outgoing type state.

    • One can validly replace types matching the class MethodClassName and the argument types given in Descriptor on the incoming operand stack with the return type given in Descriptor.

instructionIsTypeSafe(invokespecial(CP), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
CP = method(MethodClassName, MethodName, Descriptor),
MethodName \= '<init>',
MethodName \= '<clinit>',
parseMethodDescriptor(Descriptor, OperandArgList, ReturnType),
thisClass(Environment, class(CurrentClassName, CurrentLoader)),
isAssignable(class(CurrentClassName, CurrentLoader),
class(MethodClassName, CurrentLoader)),
reverse([class(CurrentClassName, CurrentLoader) | OperandArgList],
StackArgList),
validTypeTransition(Environment, StackArgList, ReturnType,
StackFrame, NextStackFrame),
reverse([class(MethodClassName, CurrentLoader) | OperandArgList],
StackArgList2),
validTypeTransition(Environment, StackArgList2, ReturnType,
StackFrame, _ResultStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The isAssignable clause enforces the structural constraint that invokespecial, for other than an instance initialization method, must name a method in the current class/interface or a superclass/superinterface.

The first validTypeTransition clause enforces the structural constraint that invokespecial, for other than an instance initialization method, targets a receiver object of the current class or deeper. To see why, consider that StackArgList simulates the list of types on the operand stack expected by the method, starting with the current class (the class performing invokespecial). The actual types on the operand stack are in StackFrame. The effect of validTypeTransition is to pop the first type from the operand stack in StackFrame and check it is a subtype of the first term of StackArgList, namely the current class. Thus, the actual receiver type is compatible with the current class.

A sharp-eyed reader might notice that enforcing this structural constraint supercedes the structural constraint pertaining to invokespecial of a protected method. Thus, the Prolog code above makes no reference to passesProtectedCheck (§4.10.1.8), whereas the Prolog code for invokespecial of an instance initialization method uses passesProtectedCheck to ensure the actual receiver type is compatible with the current class when certain protected instance initialization methods are named.

The second validTypeTransition clause enforces the structural constraint that any method invocation instruction must target a receiver object whose type is compatible with the type named by the instruction. To see why, consider that StackArgList2 simulates the list of types on the operand stack expected by the method, starting with the type named by the instruction. Again, the actual types on the operand stack are in StackFrame, and the effect of validTypeTransition is to check the actual receiver type in StackFrame is compatible with the type named by the instruction in StackArgList2.

  • Or:

    • MethodName is <init>.

    • Descriptor specifies a void return type.

    • One can validly pop types matching the argument types given in Descriptor and an uninitialized type, UninitializedArg, off the incoming operand stack, yielding OperandStack.

    • The outgoing type state is derived from the incoming type state by first replacing the incoming operand stack with OperandStack and then replacing all instances of UninitializedArgwith the type of instance being initialized.

    • If the instruction calls an instance initialization method on a class instance created by an earlier new instruction, and the method is protected, the usage conforms to the special rules governing access to protected members (§4.10.1.8).

instructionIsTypeSafe(invokespecial(CP), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
CP = method(MethodClassName, '<init>', Descriptor),
parseMethodDescriptor(Descriptor, OperandArgList, void),
reverse(OperandArgList, StackArgList),
canPop(StackFrame, StackArgList, TempFrame),
TempFrame = frame(Locals, [uninitializedThis | OperandStack], Flags),
currentClassLoader(Environment, CurrentLoader),
rewrittenUninitializedType(uninitializedThis, Environment,
class(MethodClassName, CurrentLoader), This),
rewrittenInitializationFlags(uninitializedThis, Flags, NextFlags),
substitute(uninitializedThis, This, OperandStack, NextOperandStack),
substitute(uninitializedThis, This, Locals, NextLocals),
NextStackFrame = frame(NextLocals, NextOperandStack, NextFlags),
ExceptionStackFrame = frame(Locals, [], Flags).
instructionIsTypeSafe(invokespecial(CP), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
CP = method(MethodClassName, '<init>', Descriptor),
parseMethodDescriptor(Descriptor, OperandArgList, void),
reverse(OperandArgList, StackArgList),
canPop(StackFrame, StackArgList, TempFrame),
TempFrame = frame(Locals, [uninitialized(Address) | OperandStack], Flags),
currentClassLoader(Environment, CurrentLoader),
rewrittenUninitializedType(uninitialized(Address), Environment,
class(MethodClassName, CurrentLoader), This),
rewrittenInitializationFlags(uninitialized(Address), Flags, NextFlags),
substitute(uninitialized(Address), This, OperandStack, NextOperandStack),
substitute(uninitialized(Address), This, Locals, NextLocals),
NextStackFrame = frame(NextLocals, NextOperandStack, NextFlags),
ExceptionStackFrame = frame(Locals, [], Flags),
passesProtectedCheck(Environment, MethodClassName, '<init>',
Descriptor, NextStackFrame).

To compute what type the uninitialized argument's type needs to be rewritten to, there are two cases:

  • If we are initializing an object within its constructor, its type is initially uninitializedThis. This type will be rewritten to the type of the class of the <init> method.

  • The second case arises from initialization of an object created by new. The uninitialized arg type is rewritten to MethodClass, the type of the method holder of <init>. We check whether there really is a new instruction at Address.

rewrittenUninitializedType(uninitializedThis, Environment,
MethodClass, MethodClass) :-
MethodClass = class(MethodClassName, CurrentLoader),
thisClass(Environment, MethodClass). rewrittenUninitializedType(uninitializedThis, Environment,
MethodClass, MethodClass) :-
MethodClass = class(MethodClassName, CurrentLoader),
thisClass(Environment, class(thisClassName, thisLoader)),
superclassChain(thisClassName, thisLoader, [MethodClass | Rest]). rewrittenUninitializedType(uninitialized(Address), Environment,
MethodClass, MethodClass) :-
allInstructions(Environment, Instructions),
member(instruction(Address, new(MethodClass)), Instructions). rewrittenInitializationFlags(uninitializedThis, _Flags, []).
rewrittenInitializationFlags(uninitialized(_), Flags, Flags). substitute(_Old, _New, [], []).
substitute(Old, New, [Old | FromRest], [New | ToRest]) :-
substitute(Old, New, FromRest, ToRest).
substitute(Old, New, [From1 | FromRest], [From1 | ToRest]) :-
From1 \= Old,
substitute(Old, New, FromRest, ToRest).

The rule for invokespecial of an <init> method is the sole motivation for passing back a distinct exception stack frame. The concern is that when initializing an object within its constructor, invokespecial can cause a superclass <init> method to be invoked, and that invocation could fail, leaving this uninitialized. This situation cannot be created using source code in the Java programming language, but can be created by programming in bytecode directly.

In this situation, the original frame holds an uninitialized object in local variable 0 and has flag flagThisUninit. Normal termination of invokespecial initializes the uninitialized object and turns off the flagThisUninit flag. But if the invocation of an <init> method throws an exception, the uninitialized object might be left in a partially initialized state, and needs to be made permanently unusable. This is represented by an exception frame containing the broken object (the new value of the local) and the flagThisUninit flag (the old flag). There is no way to get from an apparently-initialized object bearing the flagThisUninit flag to a properly initialized object, so the object is permanently unusable.

If not for this situation, the flags of the exception stack frame would always be the same as the flags of the input stack frame.

invokestatic

An invokestatic instruction is type safe iff all of the following are true:

  • Its first operand, CP, refers to a constant pool entry denoting a method named MethodName with descriptor Descriptor.

  • MethodName is not <init>.

  • MethodName is not <clinit>.

  • One can validly replace types matching the argument types given in Descriptor on the incoming operand stack with the return type given in Descriptor, yielding the outgoing type state.

instructionIsTypeSafe(invokestatic(CP), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
CP = method(_MethodClassName, MethodName, Descriptor),
MethodName \= '<init>',
MethodName \= '<clinit>',
parseMethodDescriptor(Descriptor, OperandArgList, ReturnType),
reverse(OperandArgList, StackArgList),
validTypeTransition(Environment, StackArgList, ReturnType,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
invokevirtual

An invokevirtual instruction is type safe iff all of the following are true:

  • Its first operand, CP, refers to a constant pool entry denoting a method named MethodName with descriptor Descriptor that is a member of a class MethodClassName.

  • MethodName is not <init>.

  • MethodName is not <clinit>.

  • One can validly replace types matching the class MethodClassName and the argument types given in Descriptor on the incoming operand stack with the return type given in Descriptor, yielding the outgoing type state.

  • If the method is protected, the usage conforms to the special rules governing access to protected members (§4.10.1.8).

instructionIsTypeSafe(invokevirtual(CP), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
CP = method(MethodClassName, MethodName, Descriptor),
MethodName \= '<init>',
MethodName \= '<clinit>',
parseMethodDescriptor(Descriptor, OperandArgList, ReturnType),
reverse(OperandArgList, ArgList),
currentClassLoader(Environment, CurrentLoader),
reverse([class(MethodClassName, CurrentLoader) | OperandArgList],
StackArgList),
validTypeTransition(Environment, StackArgList, ReturnType,
StackFrame, NextStackFrame),
canPop(StackFrame, ArgList, PoppedFrame),
passesProtectedCheck(Environment, MethodClassName, MethodName,
Descriptor, PoppedFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
iorirem

An ior instruction is type safe iff the equivalent iadd instruction is type safe.

instructionHasEquivalentTypeRule(ior, iadd).

An irem instruction is type safe iff the equivalent iadd instruction is type safe.

instructionHasEquivalentTypeRule(irem, iadd).
ireturn

An ireturn instruction is type safe if the enclosing method has a declared return type of int, and one can validly pop a type matching int off the incoming operand stack.

instructionIsTypeSafe(ireturn, Environment, _Offset, StackFrame,
afterGoto, ExceptionStackFrame) :-
thisMethodReturnType(Environment, int),
canPop(StackFrame, [int], _PoppedStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
ishlishriushr

An ishl instruction is type safe iff the equivalent iadd instruction is type safe.

instructionHasEquivalentTypeRule(ishl, iadd).

An ishr instruction is type safe iff the equivalent iadd instruction is type safe.

instructionHasEquivalentTypeRule(ishr, iadd).

An iushr instruction is type safe iff the equivalent iadd instruction is type safe.

instructionHasEquivalentTypeRule(iushr, iadd).
istoreistore_<n>

An istore instruction with operand Index is type safe and yields an outgoing type state NextStackFrame, if a store instruction with operand Index and type int is type safe and yields an outgoing type state NextStackFrame.

instructionIsTypeSafe(istore(Index), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
storeIsTypeSafe(Environment, Index, int, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The instructions istore_<n>, for 0 ≤ n ≤ 3, are type safe iff the equivalent istore instruction is type safe.

instructionHasEquivalentTypeRule(istore_0, istore(0)).
instructionHasEquivalentTypeRule(istore_1, istore(1)).
instructionHasEquivalentTypeRule(istore_2, istore(2)).
instructionHasEquivalentTypeRule(istore_3, istore(3)).
isubixor

An isub instruction is type safe iff the equivalent iadd instruction is type safe.

instructionHasEquivalentTypeRule(isub, iadd).

An ixor instruction is type safe iff the equivalent iadd instruction is type safe.

instructionHasEquivalentTypeRule(ixor, iadd).
l2dl2fl2i

An l2d instruction is type safe if one can validly pop long off the incoming operand stack and replace it with double, yielding the outgoing type state.

instructionIsTypeSafe(l2d, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [long], double,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

An l2f instruction is type safe if one can validly pop long off the incoming operand stack and replace it with float, yielding the outgoing type state.

instructionIsTypeSafe(l2f, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [long], float,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

An l2i instruction is type safe if one can validly pop long off the incoming operand stack and replace it with int, yielding the outgoing type state.

instructionIsTypeSafe(l2i, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [long], int,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
ladd

An ladd instruction is type safe iff one can validly replace types matching long and long on the incoming operand stack with long yielding the outgoing type state.

instructionIsTypeSafe(ladd, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [long, long], long,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
laload

An laload instruction is type safe iff one can validly replace types matching int and array of long on the incoming operand stack with long yielding the outgoing type state.

instructionIsTypeSafe(laload, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [int, arrayOf(long)], long,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
land

An land instruction is type safe iff the equivalent ladd instruction is type safe.

instructionHasEquivalentTypeRule(land, ladd).
lastore

An lastore instruction is type safe iff one can validly pop types matching longint and array of long off the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(lastore, _Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
canPop(StackFrame, [long, int, arrayOf(long)], NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
lcmp

lcmp instruction is type safe iff one can validly replace types matching long and long on the incoming operand stack with int yielding the outgoing type state.

instructionIsTypeSafe(lcmp, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [long, long], int,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
lconst_<l>

An lconst_0 instruction is type safe if one can validly push the type long onto the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(lconst_0, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [], long, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

An lconst_1 instruction is type safe iff the equivalent lconst_0 instruction is type safe.

instructionHasEquivalentTypeRule(lconst_1, lconst_0).
ldcldc_wldc2_w

An ldc instruction with operand CP is type safe iff CP refers to a constant pool entry denoting an entity of type Type, where Type is either intfloatStringClassjava.lang.invoke.MethodType, or java.lang.invoke.MethodHandle, and one can validly push Type onto the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(ldc(CP), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
functor(CP, Tag, _),
isBootstrapLoader(BL),
member([Tag, Type], [
[int, int],
[float, float],
[string, class('java/lang/String', BL)],
[classConst, class('java/lang/Class', BL)],
[methodTypeConst, class('java/lang/invoke/MethodType', BL)],
[methodHandleConst, class('java/lang/invoke/MethodHandle', BL)],
]),
validTypeTransition(Environment, [], Type, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

An ldc_w instruction is type safe iff the equivalent ldc instruction is type safe.

instructionHasEquivalentTypeRule(ldc_w(CP), ldc(CP))

An ldc2_w instruction with operand CP is type safe iff CP refers to a constant pool entry denoting an entity of type Tag, where Tag is either long or double, and one can validly push Tag onto the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(ldc2_w(CP), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
functor(CP, Tag, _),
member(Tag, [long, double]),
validTypeTransition(Environment, [], Tag, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
ldiv

An ldiv instruction is type safe iff the equivalent ladd instruction is type safe.

instructionHasEquivalentTypeRule(ldiv, ladd).
lloadlload_<n>

An lload instruction with operand Index is type safe and yields an outgoing type state NextStackFrame, if a load instruction with operand Index and type long is type safe and yields an outgoing type state NextStackFrame.

instructionIsTypeSafe(lload(Index), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
loadIsTypeSafe(Environment, Index, long, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The instructions lload_<n>, for 0 ≤ n ≤ 3, are type safe iff the equivalent lload instruction is type safe.

instructionHasEquivalentTypeRule(lload_0, lload(0)).
instructionHasEquivalentTypeRule(lload_1, lload(1)).
instructionHasEquivalentTypeRule(lload_2, lload(2)).
instructionHasEquivalentTypeRule(lload_3, lload(3)).
lmul

An lmul instruction is type safe iff the equivalent ladd instruction is type safe.

instructionHasEquivalentTypeRule(lmul, ladd).
lneg

An lneg instruction is type safe iff there is a type matching long on the incoming operand stack. The lneg instruction does not alter the type state.

instructionIsTypeSafe(lneg, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [long], long,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
lookupswitch

lookupswitch instruction is type safe if its keys are sorted, one can validly pop int off the incoming operand stack yielding a new type state BranchStackFrame, and all of the instruction's targets are valid branch targets assuming BranchStackFrame as their incoming type state.

instructionIsTypeSafe(lookupswitch(Targets, Keys), Environment, _, StackFrame,
afterGoto, ExceptionStackFrame) :-
sort(Keys, Keys),
canPop(StackFrame, [int], BranchStackFrame),
checklist(targetIsTypeSafe(Environment, BranchStackFrame), Targets),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
lorlrem

lor instruction is type safe iff the equivalent ladd instruction is type safe.

instructionHasEquivalentTypeRule(lor, ladd).

An lrem instruction is type safe iff the equivalent ladd instruction is type safe.

instructionHasEquivalentTypeRule(lrem, ladd).
lreturn

An lreturn instruction is type safe if the enclosing method has a declared return type of long, and one can validly pop a type matching long off the incoming operand stack.

instructionIsTypeSafe(lreturn, Environment, _Offset, StackFrame,
afterGoto, ExceptionStackFrame) :-
thisMethodReturnType(Environment, long),
canPop(StackFrame, [long], _PoppedStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
lshllshrlushr

An lshl instruction is type safe if one can validly replace the types int and long on the incoming operand stack with the type long yielding the outgoing type state.

instructionIsTypeSafe(lshl, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [int, long], long,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

An lshr instruction is type safe iff the equivalent lshl instruction is type safe.

instructionHasEquivalentTypeRule(lshr, lshl).

An lushr instruction is type safe iff the equivalent lshl instruction is type safe.

instructionHasEquivalentTypeRule(lushr, lshl).
lstorelstore_<n>

An lstore instruction with operand Index is type safe and yields an outgoing type state NextStackFrame, if a store instruction with operand Index and type long is type safe and yields an outgoing type state NextStackFrame.

instructionIsTypeSafe(lstore(Index), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
storeIsTypeSafe(Environment, Index, long, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The instructions lstore_<n>, for 0 ≤ n ≤ 3, are type safe iff the equivalent lstore instruction is type safe.

instructionHasEquivalentTypeRule(lstore_0, lstore(0)).
instructionHasEquivalentTypeRule(lstore_1, lstore(1)).
instructionHasEquivalentTypeRule(lstore_2, lstore(2)).
instructionHasEquivalentTypeRule(lstore_3, lstore(3)).
lsublxor

An lsub instruction is type safe iff the equivalent ladd instruction is type safe.

instructionHasEquivalentTypeRule(lsub, ladd).

An lxor instruction is type safe iff the equivalent ladd instruction is type safe.

instructionHasEquivalentTypeRule(lxor, ladd).
monitorentermonitorexit

monitorenter instruction is type safe iff one can validly pop a type matching reference off the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(monitorenter, _Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
canPop(StackFrame, [reference], NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

monitorexit instruction is type safe iff the equivalent monitorenter instruction is type safe.

instructionHasEquivalentTypeRule(monitorexit, monitorenter).
multianewarray

multianewarray instruction with operands CP and Dim is type safe iff CP refers to a constant pool entry denoting an array type whose dimension is greater or equal to DimDim is strictly positive, and one can validly replace Dim int types on the incoming operand stack with the type denoted by CP yielding the outgoing type state.

instructionIsTypeSafe(multianewarray(CP, Dim), Environment, _Offset,
StackFrame, NextStackFrame, ExceptionStackFrame) :-
CP = arrayOf(_),
classDimension(CP, Dimension),
Dimension >= Dim,
Dim > 0,
/* Make a list of Dim ints */
findall(int, between(1, Dim, _), IntList),
validTypeTransition(Environment, IntList, CP,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The dimension of an array type whose component type is also an array type is one more than the dimension of its component type.

classDimension(arrayOf(X), Dimension) :-
classDimension(X, Dimension1),
Dimension is Dimension1 + 1. classDimension(_, Dimension) :-
Dimension = 0.
new

new instruction with operand CP at offset Offset is type safe iff CP refers to a constant pool entry denoting a class or interface type, the type uninitialized(Offset) does not appear in the incoming operand stack, and one can validly push uninitialized(Offset) onto the incoming operand stack and replace uninitialized(Offset) with top in the incoming local variables yielding the outgoing type state.

instructionIsTypeSafe(new(CP), Environment, Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
StackFrame = frame(Locals, OperandStack, Flags),
CP = class(_, _),
NewItem = uninitialized(Offset),
notMember(NewItem, OperandStack),
substitute(NewItem, top, Locals, NewLocals),
validTypeTransition(Environment, [], NewItem,
frame(NewLocals, OperandStack, Flags),
NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The substitute predicate is defined in the rule for invokespecial (§invokespecial).

newarray

newarray instruction with operand TypeCode is type safe iff TypeCode corresponds to the primitive type ElementType, and one can validly replace the type int on the incoming operand stack with the type 'array of ElementType', yielding the outgoing type state.

instructionIsTypeSafe(newarray(TypeCode), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
primitiveArrayInfo(TypeCode, _TypeChar, ElementType, _VerifierType),
validTypeTransition(Environment, [int], arrayOf(ElementType),
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

The correspondence between type codes and primitive types is specified by the following predicate:

primitiveArrayInfo(4,  0'Z, boolean, int).
primitiveArrayInfo(5, 0'C, char, int).
primitiveArrayInfo(6, 0'F, float, float).
primitiveArrayInfo(7, 0'D, double, double).
primitiveArrayInfo(8, 0'B, byte, int).
primitiveArrayInfo(9, 0'S, short, int).
primitiveArrayInfo(10, 0'I, int, int).
primitiveArrayInfo(11, 0'J, long, long).
nop

nop instruction is always type safe. The nop instruction does not affect the type state.

instructionIsTypeSafe(nop, _Environment, _Offset, StackFrame,
StackFrame, ExceptionStackFrame) :-
exceptionStackFrame(StackFrame, ExceptionStackFrame).
poppop2

pop instruction is type safe iff one can validly pop a category 1 type off the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(pop, _Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
StackFrame = frame(Locals, [Type | Rest], Flags),
popCategory1([Type | Rest], Type, Rest),
NextStackFrame = frame(Locals, Rest, Flags),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

pop2 instruction is type safe iff it is a type safe form of the pop2 instruction.

instructionIsTypeSafe(pop2, _Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
StackFrame = frame(Locals, InputOperandStack, Flags),
pop2SomeFormIsTypeSafe(InputOperandStack, OutputOperandStack),
NextStackFrame = frame(Locals, OutputOperandStack, Flags),
exceptionStackFrame(StackFrame, ExceptionStackFrame).

pop2 instruction is a type safe form of the pop2 instruction iff it is a type safe form 1 pop2 instruction or a type safe form 2 pop2 instruction.

pop2SomeFormIsTypeSafe(InputOperandStack, OutputOperandStack) :-
pop2Form1IsTypeSafe(InputOperandStack, OutputOperandStack). pop2SomeFormIsTypeSafe(InputOperandStack, OutputOperandStack) :-
pop2Form2IsTypeSafe(InputOperandStack, OutputOperandStack).

pop2 instruction is a type safe form 1 pop2 instruction iff one can validly pop two types of size 1 off the incoming operand stack yielding the outgoing type state.

pop2Form1IsTypeSafe([Type1, Type2 | Rest], Rest) :-
popCategory1([Type1 | Rest], Type1, Rest),
popCategory1([Type2 | Rest], Type2, Rest).

pop2 instruction is a type safe form 2 pop2 instruction iff one can validly pop a type of size 2 off the incoming operand stack yielding the outgoing type state.

pop2Form2IsTypeSafe([top, Type | Rest], Rest) :-
popCategory2([top, Type | Rest], Type, Rest).
putfield

putfield instruction with operand CP is type safe iff all of the following are true:

  • Its first operand, CP, refers to a constant pool entry denoting a field whose declared type is FieldType, declared in a class FieldClassNameFieldClassName must not be an array type.

  • Either:

    • One can validly pop types matching FieldType and FieldClassName off the incoming operand stack yielding the outgoing type state.

    • protected fields are subject to additional checks (§4.10.1.8).

instructionIsTypeSafe(putfield(CP), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
CP = field(FieldClassName, FieldName, FieldDescriptor),
parseFieldDescriptor(FieldDescriptor, FieldType),
canPop(StackFrame, [FieldType], PoppedFrame),
passesProtectedCheck(Environment, FieldClassName, FieldName,
FieldDescriptor, PoppedFrame),
currentClassLoader(Environment, CurrentLoader),
canPop(StackFrame, [FieldType, class(FieldClassName, CurrentLoader)],
NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
  • Or:

    • If the instruction occurs in an instance initialization method of the class FieldClassName, then one can validly pop types matching FieldType and uninitializedThis off the incoming operand stack yielding the outgoing type state. This allows instance fields of this that are declared in the current class to be assigned prior to complete initialization of this.

instructionIsTypeSafe(putfield(CP), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
CP = field(FieldClassName, _FieldName, FieldDescriptor),
parseFieldDescriptor(FieldDescriptor, FieldType),
Environment = environment(CurrentClass, CurrentMethod, _, _, _, _),
CurrentClass = class(FieldClassName, _),
isInit(CurrentMethod),
canPop(StackFrame, [FieldType, uninitializedThis], NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
putstatic

putstatic instruction with operand CP is type safe iff CP refers to a constant pool entry denoting a field whose declared type is FieldType, and one can validly pop a type matching FieldTypeoff the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(putstatic(CP), _Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
CP = field(_FieldClassName, _FieldName, FieldDescriptor),
parseFieldDescriptor(FieldDescriptor, FieldType),
canPop(StackFrame, [FieldType], NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
return

return instruction is type safe if the enclosing method declares a void return type, and either:

  • The enclosing method is not an <init> method, or

  • this has already been completely initialized at the point where the instruction occurs.

instructionIsTypeSafe(return, Environment, _Offset, StackFrame,
afterGoto, ExceptionStackFrame) :-
thisMethodReturnType(Environment, void),
StackFrame = frame(_Locals, _OperandStack, Flags),
notMember(flagThisUninit, Flags),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
saload

An saload instruction is type safe iff one can validly replace types matching int and array of short on the incoming operand stack with int yielding the outgoing type state.

instructionIsTypeSafe(saload, Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [int, arrayOf(short)], int,
StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
sastore

An sastore instruction is type safe iff one can validly pop types matching intint, and array of short off the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(sastore, _Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
canPop(StackFrame, [int, int, arrayOf(short)], NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
sipush

An sipush instruction is type safe iff one can validly push the type int onto the incoming operand stack yielding the outgoing type state.

instructionIsTypeSafe(sipush(_Value), Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
validTypeTransition(Environment, [], int, StackFrame, NextStackFrame),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
swap

swap instruction is type safe iff one can validly replace two category 1 types, Type1 and Type2, on the incoming operand stack with the types Type2 and Type1 yielding the outgoing type state.

instructionIsTypeSafe(swap, _Environment, _Offset, StackFrame,
NextStackFrame, ExceptionStackFrame) :-
StackFrame = frame(_Locals, [Type1, Type2 | Rest], _Flags),
popCategory1([Type1 | Rest], Type1, Rest),
popCategory1([Type2 | Rest], Type2, Rest),
NextStackFrame = frame(_Locals, [Type2, Type1 | Rest], _Flags),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
tableswitch

tableswitch instruction is type safe if its keys are sorted, one can validly pop int off the incoming operand stack yielding a new type state BranchStackFrame, and all of the instruction's targets are valid branch targets assuming BranchStackFrame as their incoming type state.

instructionIsTypeSafe(tableswitch(Targets, Keys), Environment, _Offset,
StackFrame, afterGoto, ExceptionStackFrame) :-
sort(Keys, Keys),
canPop(StackFrame, [int], BranchStackFrame),
checklist(targetIsTypeSafe(Environment, BranchStackFrame), Targets),
exceptionStackFrame(StackFrame, ExceptionStackFrame).
wide

The wide instructions follow the same rules as the instructions they widen.

instructionHasEquivalentTypeRule(wide(WidenedInstruction),
WidenedInstruction).

4.10.2. Verification by Type Inference

class file that does not contain a StackMapTable attribute (which necessarily has a version number of 49.0 or below) must be verified using type inference.

4.10.2.1. The Process of Verification by Type Inference

During linking, the verifier checks the code array of the Code attribute for each method of the class file by performing data-flow analysis on each method. The verifier ensures that at any given point in the program, no matter what code path is taken to reach that point, all of the following are true:

  • The operand stack is always the same size and contains the same types of values.

  • No local variable is accessed unless it is known to contain a value of an appropriate type.

  • Methods are invoked with the appropriate arguments.

  • Fields are assigned only using values of appropriate types.

  • All opcodes have appropriately typed arguments on the operand stack and in the local variable array.

For efficiency reasons, certain tests that could in principle be performed by the verifier are delayed until the first time the code for the method is actually invoked. In so doing, the verifier avoids loading class files unless it has to.

For example, if a method invokes another method that returns an instance of class A, and that instance is assigned only to a field of the same type, the verifier does not bother to check if the class A actually exists. However, if it is assigned to a field of the type B, the definitions of both A and B must be loaded in to ensure that A is a subclass of B.

4.10.2.2. The Bytecode Verifier

The code for each method is verified independently. First, the bytes that make up the code are broken up into a sequence of instructions, and the index into the code array of the start of each instruction is placed in an array. The verifier then goes through the code a second time and parses the instructions. During this pass a data structure is built to hold information about each Java Virtual Machine instruction in the method. The operands, if any, of each instruction are checked to make sure they are valid. For instance:

  • Branches must be within the bounds of the code array for the method.

  • The targets of all control-flow instructions are each the start of an instruction. In the case of a wide instruction, the wide opcode is considered the start of the instruction, and the opcode giving the operation modified by that wide instruction is not considered to start an instruction. Branches into the middle of an instruction are disallowed.

  • No instruction can access or modify a local variable at an index greater than or equal to the number of local variables that its method indicates it allocates.

  • All references to the constant pool must be to an entry of the appropriate type. (For example, the instruction getfield must reference a field.)

  • The code does not end in the middle of an instruction.

  • Execution cannot fall off the end of the code.

  • For each exception handler, the starting and ending point of code protected by the handler must be at the beginning of an instruction or, in the case of the ending point, immediately past the end of the code. The starting point must be before the ending point. The exception handler code must start at a valid instruction, and it must not start at an opcode being modified by the wideinstruction.

For each instruction of the method, the verifier records the contents of the operand stack and the contents of the local variable array prior to the execution of that instruction. For the operand stack, it needs to know the stack height and the type of each value on it. For each local variable, it needs to know either the type of the contents of that local variable or that the local variable contains an unusable or unknown value (it might be uninitialized). The bytecode verifier does not need to distinguish between the integral types (e.g., byteshortchar) when determining the value types on the operand stack.

Next, a data-flow analyzer is initialized. For the first instruction of the method, the local variables that represent parameters initially contain values of the types indicated by the method's type descriptor; the operand stack is empty. All other local variables contain an illegal value. For the other instructions, which have not been examined yet, no information is available regarding the operand stack or local variables.

Finally, the data-flow analyzer is run. For each instruction, a "changed" bit indicates whether this instruction needs to be looked at. Initially, the "changed" bit is set only for the first instruction. The data-flow analyzer executes the following loop:

  1. Select a Java Virtual Machine instruction whose "changed" bit is set. If no instruction remains whose "changed" bit is set, the method has successfully been verified. Otherwise, turn off the "changed" bit of the selected instruction.

  2. Model the effect of the instruction on the operand stack and local variable array by doing the following:

    • If the instruction uses values from the operand stack, ensure that there are a sufficient number of values on the stack and that the top values on the stack are of an appropriate type. Otherwise, verification fails.

    • If the instruction uses a local variable, ensure that the specified local variable contains a value of the appropriate type. Otherwise, verification fails.

    • If the instruction pushes values onto the operand stack, ensure that there is sufficient room on the operand stack for the new values. Add the indicated types to the top of the modeled operand stack.

    • If the instruction modifies a local variable, record that the local variable now contains the new type.

  3. Determine the instructions that can follow the current instruction. Successor instructions can be one of the following:

    • The next instruction, if the current instruction is not an unconditional control transfer instruction (for instance, gotoreturn, or athrow). Verification fails if it is possible to "fall off" the last instruction of the method.

    • The target(s) of a conditional or unconditional branch or switch.

    • Any exception handlers for this instruction.

  4. Merge the state of the operand stack and local variable array at the end of the execution of the current instruction into each of the successor instructions, as follows:

    • If this is the first time the successor instruction has been visited, record that the operand stack and local variable values calculated in step 2 are the state of the operand stack and local variable array prior to executing the successor instruction. Set the "changed" bit for the successor instruction.

    • If the successor instruction has been seen before, merge the operand stack and local variable values calculated in step 2 into the values already there. Set the "changed" bit if there is any modification to the values.

    In the special case of control transfer to an exception handler:

    • Record that a single object, of the exception type indicated by the exception handler, is the state of the operand stack prior to executing the successor instruction. There must be sufficient room on the operand stack for this single value, as if an instruction had pushed it.

    • Record that the local variable values from immediately before step 2 are the state of the local variable array prior to executing the successor instruction. The local variable values calculated in step 2 are irrelevant.

  5. Continue at step 1.

To merge two operand stacks, the number of values on each stack must be identical. Then, corresponding values on the two stacks are compared and the value on the merged stack is computed, as follows:

  • If one value is a primitive type, then the corresponding value must be the same primitive type. The merged value is the primitive type.

  • If one value is a non-array reference type, then the corresponding value must be a reference type (array or non-array). The merged value is a reference to an instance of the first common supertype of the two reference types. (Such a reference type always exists because the type Object is a supertype of all class, interface, and array types.)

    For example, Object and String can be merged; the result is Object. Similarly, Object and String[] can be merged; the result is again Object. Even Object and int[] can be merged, or String and int[]; the result is Object for both.

  • If corresponding values are both array reference types, then their dimensions are examined. If the array types have the same dimensions, then the merged value is a reference to an instance of an array type which is first common supertype of both array types. (If either or both of the array types has a primitive element type, then Object is used as the element type instead.) If the array types have different dimensions, then the merged value is a reference to an instance of an array type whose dimension is the smaller of the two; the element type isCloneable or java.io.Serializable if the smaller array type was Cloneable or java.io.Serializable, and Object otherwise.

    For example, Object[] and String[] can be merged; the result is Object[]Cloneable[] and String[] can be merged, or java.io.Serializable[] and String[]; the result is Cloneable[] and java.io.Serializable[] respectively. Even int[] and String[] can be merged; the result is Object[], because Object is used instead of int when computing the first common supertype.

    Since the array types can have different dimensions, Object[] and String[][] can be merged, or Object[][] and String[]; in both cases the result is Object[]Cloneable[] andString[][] can be merged; the result is Cloneable[]. Finally, Cloneable[][] and String[] can be merged; the result is Object[].

If the operand stacks cannot be merged, verification of the method fails.

To merge two local variable array states, corresponding pairs of local variables are compared. The value of the merged local variable is computed using the rules above, except that the corresponding values are permitted to be different primitive types. In that case, the verifier records that the merged local variable contains an unusable value.

If the data-flow analyzer runs on a method without reporting a verification failure, then the method has been successfully verified by the class file verifier.

Certain instructions and data types complicate the data-flow analyzer. We now examine each of these in more detail.

4.10.2.3. Values of Types long and double

Values of the long and double types are treated specially by the verification process.

Whenever a value of type long or double is moved into a local variable at index n, index n+1 is specially marked to indicate that it has been reserved by the value at index n and must not be used as a local variable index. Any value previously at index n+1 becomes unusable.

Whenever a value is moved to a local variable at index n, the index n-1 is examined to see if it is the index of a value of type long or double. If so, the local variable at index n-1 is changed to indicate that it now contains an unusable value. Since the local variable at index n has been overwritten, the local variable at index n-1 cannot represent a value of type long or double.

Dealing with values of types long or double on the operand stack is simpler; the verifier treats them as single values on the stack. For example, the verification code for the dadd opcode (add two double values) checks that the top two items on the stack are both of type double. When calculating operand stack length, values of type long and double have length two.

Untyped instructions that manipulate the operand stack must treat values of type long and double as atomic (indivisible). For example, the verifier reports a failure if the top value on the stack is a double and it encounters an instruction such as pop or dup. The instructions pop2 or dup2 must be used instead.

4.10.2.4. Instance Initialization Methods and Newly Created Objects

Creating a new class instance is a multistep process. The statement:

...
new myClass(i, j, k);
...

can be implemented by the following:

...
new #1 // Allocate uninitialized space for myClass
dup // Duplicate object on the operand stack
iload_1 // Push i
iload_2 // Push j
iload_3 // Push k
invokespecial #5 // Invoke myClass.<init>
...

This instruction sequence leaves the newly created and initialized object on top of the operand stack. (Additional examples of compilation to the instruction set of the Java Virtual Machine are given in §3 (Compiling for the Java Virtual Machine).)

The instance initialization method (§2.9.1) for class myClass sees the new uninitialized object as its this argument in local variable 0. Before that method invokes another instance initialization method of myClass or its direct superclass on this, the only operation the method can perform on this is assigning fields declared within myClass.

When doing dataflow analysis on instance methods, the verifier initializes local variable 0 to contain an object of the current class, or, for instance initialization methods, local variable 0 contains a special type indicating an uninitialized object. After an appropriate instance initialization method is invoked (from the current class or its direct superclass) on this object, all occurrences of this special type on the verifier's model of the operand stack and in the local variable array are replaced by the current class type. The verifier rejects code that uses the new object before it has been initialized or that initializes the object more than once. In addition, it ensures that every normal return of the method has invoked an instance initialization method either in the class of this method or in the direct superclass.

Similarly, a special type is created and pushed on the verifier's model of the operand stack as the result of the Java Virtual Machine instruction new. The special type indicates the instruction by which the class instance was created and the type of the uninitialized class instance created. When an instance initialization method declared in the class of the uninitialized class instance is invoked on that class instance, all occurrences of the special type are replaced by the intended type of the class instance. This change in type may propagate to subsequent instructions as the dataflow analysis proceeds.

The instruction number needs to be stored as part of the special type, as there may be multiple not-yet-initialized instances of a class in existence on the operand stack at one time. For example, the Java Virtual Machine instruction sequence that implements:

new InputStream(new Foo(), new InputStream("foo"))

may have two uninitialized instances of InputStream on the operand stack at once. When an instance initialization method is invoked on a class instance, only those occurrences of the special type on the operand stack or in the local variable array that are the same object as the class instance are replaced.

4.10.2.5. Exceptions and finally

To implement the try-finally construct, a compiler for the Java programming language that generates class files with version number 50.0 or below may use the exception-handling facilities together with two special instructions: jsr ("jump to subroutine") and ret ("return from subroutine"). The finally clause is compiled as a subroutine within the Java Virtual Machine code for its method, much like the code for an exception handler. When a jsr instruction that invokes the subroutine is executed, it pushes its return address, the address of the instruction after the jsr that is being executed, onto the operand stack as a value of type returnAddress. The code for the subroutine stores the return address in a local variable. At the end of the subroutine, a ret instruction fetches the return address from the local variable and transfers control to the instruction at the return address.

Control can be transferred to the finally clause (the finally subroutine can be invoked) in several different ways. If the try clause completes normally, the finally subroutine is invoked via a jsr instruction before evaluating the next expression. A break or continue inside the try clause that transfers control outside the try clause executes a jsr to the code for the finally clause first. If the try clause executes a return, the compiled code does the following:

  1. Saves the return value (if any) in a local variable.

  2. Executes a jsr to the code for the finally clause.

  3. Upon return from the finally clause, returns the value saved in the local variable.

The compiler sets up a special exception handler, which catches any exception thrown by the try clause. If an exception is thrown in the try clause, this exception handler does the following:

  1. Saves the exception in a local variable.

  2. Executes a jsr to the finally clause.

  3. Upon return from the finally clause, rethrows the exception.

For more information about the implementation of the try-finally construct, see §3.13.

The code for the finally clause presents a special problem to the verifier. Usually, if a particular instruction can be reached via multiple paths and a particular local variable contains incompatible values through those multiple paths, then the local variable becomes unusable. However, a finally clause might be called from several different places, yielding several different circumstances:

  • The invocation from the exception handler may have a certain local variable that contains an exception.

  • The invocation to implement return may have some local variable that contains the return value.

  • The invocation from the bottom of the try clause may have an indeterminate value in that same local variable.

The code for the finally clause itself might pass verification, but after completing the updating all the successors of the ret instruction, the verifier would note that the local variable that the exception handler expects to hold an exception, or that the return code expects to hold a return value, now contains an indeterminate value.

Verifying code that contains a finally clause is complicated. The basic idea is the following:

  • Each instruction keeps track of the list of jsr targets needed to reach that instruction. For most code, this list is empty. For instructions inside code for the finally clause, it is of length one. For multiply nested finally code (extremely rare!), it may be longer than one.

  • For each instruction and each jsr needed to reach that instruction, a bit vector is maintained of all local variables accessed or modified since the execution of the jsr instruction.

  • When executing the ret instruction, which implements a return from a subroutine, there must be only one possible subroutine from which the instruction can be returning. Two different subroutines cannot "merge" their execution to a single ret instruction.

  • To perform the data-flow analysis on a ret instruction, a special procedure is used. Since the verifier knows the subroutine from which the instruction must be returning, it can find all the jsrinstructions that call the subroutine and merge the state of the operand stack and local variable array at the time of the ret instruction into the operand stack and local variable array of the instructions following the jsr. Merging uses a special set of values for local variables:

    • For any local variable that the bit vector (constructed above) indicates has been accessed or modified by the subroutine, use the type of the local variable at the time of the ret.

    • For other local variables, use the type of the local variable before the jsr instruction.

4.11. Limitations of the Java Virtual Machine

The following limitations of the Java Virtual Machine are implicit in the class file format:

  • The per-class or per-interface constant pool is limited to 65535 entries by the 16-bit constant_pool_count field of the ClassFile structure (§4.1). This acts as an internal limit on the total complexity of a single class or interface.

  • The number of fields that may be declared by a class or interface is limited to 65535 by the size of the fields_count item of the ClassFile structure (§4.1).

    Note that the value of the fields_count item of the ClassFile structure does not include fields that are inherited from superclasses or superinterfaces.

  • The number of methods that may be declared by a class or interface is limited to 65535 by the size of the methods_count item of the ClassFile structure (§4.1).

    Note that the value of the methods_count item of the ClassFile structure does not include methods that are inherited from superclasses or superinterfaces.

  • The number of direct superinterfaces of a class or interface is limited to 65535 by the size of the interfaces_count item of the ClassFile structure (§4.1).

  • The greatest number of local variables in the local variables array of a frame created upon invocation of a method (§2.6) is limited to 65535 by the size of the max_locals item of the Codeattribute (§4.7.3) giving the code of the method, and by the 16-bit local variable indexing of the Java Virtual Machine instruction set.

    Note that values of type long and double are each considered to reserve two local variables and contribute two units toward the max_locals value, so use of local variables of those types further reduces this limit.

  • The size of an operand stack in a frame (§2.6) is limited to 65535 values by the max_stack field of the Code attribute (§4.7.3).

    Note that values of type long and double are each considered to contribute two units toward the max_stack value, so use of values of these types on the operand stack further reduces this limit.

  • The number of method parameters is limited to 255 by the definition of a method descriptor (§4.3.3), where the limit includes one unit for this in the case of instance or interface method invocations.

    Note that a method descriptor is defined in terms of a notion of method parameter length in which a parameter of type long or double contributes two units to the length, so parameters of these types further reduce the limit.

  • The length of field and method names, field and method descriptors, and other constant string values (including those referenced by ConstantValue (§4.7.2) attributes) is limited to 65535 characters by the 16-bit unsigned length item of the CONSTANT_Utf8_info structure (§4.4.7).

    Note that the limit is on the number of bytes in the encoding and not on the number of encoded characters. UTF-8 encodes some characters using two or three bytes. Thus, strings incorporating multibyte characters are further constrained.

  • The number of dimensions in an array is limited to 255 by the size of the dimensions opcode of the multianewarray instruction and by the constraints imposed on the multianewarrayanewarray, and newarray instructions (§4.9.1§4.9.2).

JVM Specification 9th Edition (4) Chapter 4. The class File Format的更多相关文章

  1. JVM Specification 9th Edition (3) Chapter 2. The Structure of the Java Virtual Machine

    Chapter 2. The Structure of the Java Virtual Machine 内容列表 2.1. The class File Format (class文件的格式) 2. ...

  2. JVM Specification 9th Edition (2) Chapter 1. Introduction

    Chapter 1. Introduction 翻译太累了,我就这样的看英文吧. 内容列表 1.1. A Bit of History 1.2. The Java Virtual Machine 1. ...

  3. JVM Specification 9th Edition (4) Chapter 3. Compiling for the Java Virtual Machine

    Chapter 3. Compiling for the Java Virtual Machine 内容列表 3.1. Format of Examples 3.2. Use of Constants ...

  4. JVM Specification 9th Edition (1) Cover

    这个就是Java虚拟机规范第9版的网页版封面了,上面是4个大牛的名字,先来了解以下吧,万一那天有幸遇见呢. Tim Lindholm Frank Yellin Gilad Bracha Alex Bu ...

  5. Java提高篇——JVM加载class文件的原理机制

    在面试java工程师的时候,这道题经常被问到,故需特别注意. 1.JVM 简介 JVM 是我们Javaer 的最基本功底了,刚开始学Java 的时候,一般都是从“Hello World ”开始的,然后 ...

  6. java虚拟机学习-慢慢琢磨JVM(2)

    1 JVM简介 JVM是我们Javaer的最基本功底了,刚开始学Java的时候,一般都是从“Hello World”开始的,然后会写个复杂点class,然后再找一些开源框架,比如Spring,Hibe ...

  7. jvm比较详尽的内存结构

     JVM内存结构 2012-09-17 15:27:59 分类: Java 本文转自:http://www.blogjava.net/nkjava/archive/2012/03/14/371831. ...

  8. JVM介绍(二)

    1 JVM简介 JVM是我们Javaer的最基本功底了,刚开始学Java的时候,一般都是从“Hello World”开始的,然后会写个复杂点class,然后再找一些开源框架,比如Spring,Hibe ...

  9. 慢慢琢磨JVM

    1 JVM简介 JVM是一个Javaer的最基本功底了,刚开始学Java的时候,一般都是从“Hello World”开始的,然后会写个复杂点class,然后再找一些开源框架,比如Spring,Hibe ...

随机推荐

  1. thinkphp5.0 中使用第三方无命名空间的类库

    ThinkPHP5建议所有的扩展类库都使用命名空间定义,如果你的类库没有使用命名空间,则不支持自动加载,必须使用Loader::import方法先导入文件后才能使用. 首先要在文件头部使用loader ...

  2. pgsql常用命令

    连接数据库, 默认的用户和数据库是postgres psql -U user -d dbname 切换数据库,相当于mysql的use dbname \c dbname 列举数据库,相当于mysql的 ...

  3. 如何使用awk的比较操作符

    对于使用 awk 命令的用户来说,处理一行文本中的数字或者字符串时,使用比较运算符来过滤文本和字符串是十分方便的.下面的部分我们介绍"awk"的比较运算符. awk 中的比较运算符 ...

  4. idea下一次Jar包依赖问题的解决过程

    项目导入后有几个类显示没找到依赖类. 第一步,定位到该类所在的jar包:pom文件没有报错,所以这应该是一个jar包版本问题,通过import路劲大概确定了jar包,我本地这个jar包一共有两个版本, ...

  5. Notepad++的使用

    \t 制表符.  \n 新行.  . 匹配任意字符.  | 匹配表达式左边和右边的字符. 例如, "ab|bc" 匹配 "ab" 或者 "bc&quo ...

  6. wifi简单笔记

    什么是wifi: Wi-Fi是一种可以将个人电脑.手持设备(如PDA.手机)等终端以无线方式互相连接的技术.Wi-Fi是一个无线网路通信技术的品牌,由Wi-Fi联盟(Wi-Fi Alliance)所持 ...

  7. 集合系列之fail-fast 与fail-safe 区别

    一:快速失败(fail—fast) 在用迭代器遍历一个集合对象时,如果遍历过程中对集合对象的内容进行了修改(增加.删除.修改),则会抛出Concurrent Modification Exceptio ...

  8. Odoo/OpenERP 日志配置、使用及实现

    当应用处于生产环境时,日志提供了有价值的运行时调试及监控信息,并且,也是一个有用的调试工具对于处于开发阶段的应用来说.此文描述在Odoo8.0中日志的配置.使用及实现 日志配置        Odoo ...

  9. std::shared_ptr 和 std::weak_ptr的用法以及引用计数的循环引用问题

    在std::shared_ptr被引入之前,C++标准库中实现的用于管理资源的智能指针只有std::auto_ptr一个而已.std::auto_ptr的作用非常有限,因为它存在被管理资源的所有权转移 ...

  10. 基于Netty的RPC简易实现

    代码地址如下:http://www.demodashi.com/demo/13448.html 可以给你提供思路 也可以让你学到Netty相关的知识 当然,这只是一种实现方式 需求 看下图,其实这个项 ...