docs/java-constraints.html - platform/dalvik - Gitiles

 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">

 <html>
   <head>
     <title>Java bytecode constraints</title>
     <link rel=stylesheet href="java-constraints.css">
   </head>

   <body>
     <h1>
       Bytecode constraints
     </h1>

     <p>
       From the point of view of a piece of code written in the Java
       programming language or targeted in the same way to <code>.class</code>
       files, the Dalvik VM aims to behave in a way
       that is fully consistent with the language's definition.
       That is, the code running in Dalvik will behave the same as it
       would have running in any other virtual machine. This includes
       verification failures.
       The Dx/Dalvik system will check roughly the same
       constraints that any other VM would, except as noted in the file
       <a href="verifier.html">verifier.html</a>. The following table briefly
       lists all Dx/Dalvik verification constraints together their analogs
       from the book <i>The Java<super>TM</super> Language Specification</i>,
       second edition. In the numbering scheme, the first three
       elements refer to the specification chapter, the fourth one to the
       bullet inside that chapter. The failure mode specifies whether the
       constraint will fail during the Dx conversion or during verification in
       the VM itself.
     </p>

     <h2>
       Static constraints
     </h2>

     <p>
     Static constraints are constraints on individual elements of the bytecode.
     They usually can be checked without employing control or data-flow analysis
     techniques.
     </p>

     <table>
       <tr>
         <th>
           Identifier
         </th>

         <th>
           Description
         </th>

         <th>
           Spec equivalent
         </th>

         <th>
           Failure mode
         </th>
       </tr>

       <tr>
         <td>
           A1
         </td>

         <td>
           The <code>code</code> array must not be empty.
         </td>

         <td>
           4.8.1.1
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A2
         </td>

         <td>
           The <code>code</code> array must not be larger than 65535 bytes.
         </td>

         <td>
           4.8.1.2
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A3
         </td>

         <td>
           The first opcode in <code>code</code> array must have index
           <code>0</code>.
         </td>

         <td>
           4.8.1.3
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A4
         </td>

         <td>
           The <code>code</code> array must only contain valid opcodes.
         </td>

         <td>
           4.8.1.4
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A5
         </td>

         <td>
           The index of instruction <code>n+1</code> must equal the index of
           instruction <code>n</code> plus the length of instruction
           <code>n</code>, taking into account a possible <code>wide</code>
           instruction. Opcodes modified by a <code>wide</code> instruction must
           not be directly reachable.
         </td>

         <td>
           4.8.1.5
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A6
         </td>

         <td>
           The last instruction in <code>code</code> array must end at index
           <code>code_length-1</code>.
         </td>

         <td>
           4.8.1.6
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A7
         </td>

         <td>
           All jump and branch targets must be opcodes within the same method.
           Opcodes modified by a <code>wide</code> instruction must not be
           directly reachable via a jump or branch instruction.
         </td>

         <td>
           4.8.1.7
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A8
         </td>

         <td>
           All targets of a <code>tableswitch</code> instruction must be opcodes
           within the same method. Upper and lower bounds must be consistent.
           Opcodes modified by a <code>wide</code> instruction must not be
           directly reachable via a <code>tableswitch</code> instruction.
         </td>

         <td>
           4.8.1.8
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A9
         </td>

         <td>
           All targets of a <code>lookupswitch</code> instruction must be opcodes
           within the same method. Its table must be consistent and sorted
           low-to-high. Opcodes modified by a <code>wide</code> instruction must
           not be directly reachable via a <code>lookupswitch</code> instruction.
         </td>

         <td>
           4.8.1.9
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A10
         </td>

         <td>
           The operands of <code>ldc</code> and <code>ldc_w</code> instructions
           must be valid indices into the constant pool. The respective entries
           must be of type <code>CONSTANT_Integer</code>,
           <code>CONSTANT_Float</code>, or <code>CONSTANT_String</code>.
         </td>

         <td>
           4.8.1.10
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A11
         </td>

         <td>
           The operands of <code>ldc2_w</code> instructions must be valid indices
           into the constant pool. The respective entries must be of type
           <code>CONSTANT_Long</code> or <code>CONSTANT_Double</code>. The
           subsequent constant pool entry must be valid and remain unused.
         </td>

         <td>
           4.8.1.11
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A12
         </td>

         <td>
           The Operands of <code>get&lt;kind&gt;</code> and
           <code>put&lt;kind&gt;</code> instructions must be valid indices into
           constant pool. The respective entries must be of type
           <code>CONSTANT_Fieldref</code>.
         </td>

         <td>
           4.8.1.12
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A13
         </td>

         <td>
           The first two operands of <code>invokevirtual</code>,
           <code>invokespecial</code>, and <code>invokestatic</code> must form a
           valid 16-bit index into the constant pool. The respective entries must
           be of type <code>CONSTANT_Methodref</code>.
         </td>

         <td>
           4.8.1.13
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A14
         </td>

         <td>
           Methods whose names start with '<' must only be invoked implicitly by
           the VM, not by class file code. The only exception is the instance
           initializer, which may be invoked by <code>invokespecial</code>.
         </td>

         <td>
           4.8.1.14
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A15
         </td>

         <td>
           The first two operands of <code>invokeinterface</code> must form a
           valid 16-bit index into the constant pool. The entry must be of type
           <code>CONSTANT_Interface_Methodref</code>. The third operand must
           specify number of local variables and the fourth operand must always
           be zero.
         </td>

         <td>
           4.8.1.15
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A16
         </td>

         <td>
           The operands of <code>instanceof</code>, <code>checkcast</code>,
           <code>new</code>, and <code>anewarray</code> instructions must
           be a valid index into the constant pool. The first two operands of
           <code>multianewarray</code> instruction must form a valid 16-bit index
           into the constant pool. All respective entries must be of type
           <code>CONSTANT_Class</code>.
         </td>

         <td>
           4.8.1.16
         </td>

         <td>
           DX
         </td>
       </tr>

      <tr>
         <td>
           A17
         </td>

         <td>
           The dimensions of an array created by <code>anewarray</code>
           instructions must be less than <code>256</code>.
         </td>

         <td>
           4.8.1.17
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A18
         </td>

         <td>
           The <code>new</code> instruction must not reference array classes,
           interfaces, or abstract classes.
         </td>

         <td>
           4.8.1.18
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A19
         </td>

         <td>
           The type referenced by a <code>multinewarray</code> instruction must
           have at least as many dimensions as specified in the instruction. The
           dimensions operand must not be <code>0</code>
         </td>

         <td>
           4.8.1.19
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A20
         </td>

         <td>
           The type referenced by a <code>newarray</code> instruction must be a
           valid, non-reference type.
         </td>

         <td>
           4.8.1.20
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A21
         </td>

         <td>
           The index operand of instructions explicitly referencing single-width
           local variables must be non-negative and smaller than
           <code>max_locals</code>.
         </td>

         <td>
           4.8.1.21
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A22
         </td>

         <td>
           The index operand of instructions implicitly referencing single-width
           local variables must be non-negative and smaller than
           <code>max_locals</code>.
         </td>

         <td>
           4.8.1.22
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A23
         </td>

         <td>
           The index operand of instructions explicitly referencing double-width
           local variables must be non-negative and smaller than
           <code>max_locals-1</code>.
         </td>

         <td>
           4.8.1.23
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A24
         </td>

         <td>
           The index operand of instructions implicitly referencing double-width
           local variables must be non-negative and smaller than
           <code>max_locals-1</code>.
         </td>

         <td>
           4.8.1.24
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A25
         </td>

         <td>
           The index operand of <code>wide</code> instructions explicitly
           referencing single-width local variables must be non-negative and
           smaller than <code>max_locals</code>.
         </td>

         <td>
           4.8.1.25
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           A26
         </td>

         <td>
           The index operand of <code>wide</code> instructions explicitly
           referencing double-width local variables must be non-negative and
           smaller than <code>max_locals-1</code>.
         </td>

         <td>
           4.8.1.25
         </td>

         <td>
           DX
         </td>
       </tr>
     </table>

     <h2>
       Structural constraints
     </h2>

     <p>
     Structural constraints are constraints on relationships between several
     elements of the bytecode. They usually can't be checked without employing
     control or data-flow analysis techniques.
     </p>

     <table>
       <tr>
         <th>
           Identifier
         </th>

         <th>
           Description
         </th>

         <th>
           Spec equivalent
         </th>

         <th>
           Failure mode
         </th>
       </tr>

       <tr>
         <td>
           B1
         </td>

         <td>
           The number and types of arguments (operands and local variables) must
           always match the instruction.
         </td>

         <td>
           4.8.2.1
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           B2
         </td>

         <td>
           The operand stack must have the same depth for all executions paths
           leading to an instruction.
         </td>

         <td>
           4.8.2.2
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           B3
         </td>

         <td>
           Local variable pairs must never be broken up.
         </td>

         <td>
           4.8.2.3
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           B4
         </td>

         <td>
           A local variable (or pair) has to be assigned first before it can be
           read.
         </td>

         <td>
           4.8.2.4
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           B5
         </td>

         <td>
           The operand stack must never grow beyond <code>max_stack</code>.
         </td>

         <td>
           4.8.2.5
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           B6
         </td>

         <td>
           The operand stack must never underflow.
         </td>

         <td>
           4.8.2.6
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           B7
         </td>

         <td>
           An <code>invokespecial</code> instruction must only invoke an instance
           initializer or a method in the current class or one of its
           superclasses.
         </td>

         <td>
           4.8.2.7
         </td>

         <td>
           VM
         </td>
       </tr>

       <tr>
         <td>
           B8
         </td>

         <td>
           An instance initializer must only be invoked on an uninitialized
           instance residing on the operand stack.
         </td>

         <td>
           4.8.2.8
         </td>

         <td>
           VM
         </td>
       </tr>

       <tr>
         <td>
           B9
         </td>

         <td>
           Instance methods may only be invoked on and instance fields may only
           be accessed on already initialized instances.
         </td>

         <td>
           4.8.2.9
         </td>

         <td>
           VM
         </td>
       </tr>

       <tr>
         <td>
           B10
         </td>

         <td>
            The must be no backwards branches with uninitialized instances on the
            operand stack or in local variables. There must be no code protected
            by an exception handler that contains local variables with
            uninitialized instances.
         </td>

         <td>
           4.8.2.10
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           B11
         </td>

         <td>
            An instance initializer must call another instance initializer (same
            class or superclass) before any instance members can be accessed.
            Exceptions are non-inherited instance fields, which can be assigned
            before calling another initializer, and the <code>Object</code> class
            in general.
         </td>

         <td>
           4.8.2.11
         </td>

         <td>
           VM
         </td>
       </tr>

       <tr>
         <td>
           B12
         </td>

         <td>
            All actual method arguments must be assignment-compatible with formal
            arguments.
         </td>

         <td>
           4.8.2.12
         </td>

         <td>
           VM
         </td>
       </tr>

       <tr>
         <td>
           B13
         </td>

         <td>
            For each instance method invocation, the actual instance must be
            assignment-compatible with the class or interface specified in the
            instruction.
         </td>

         <td>
           4.8.2.13
         </td>

         <td>
           VM
         </td>
       </tr>

       <tr>
         <td>
           B14
         </td>

         <td>
            A returns instruction must match its method's return type.
         </td>

         <td>
           4.8.2.14
         </td>

         <td>
           VM
         </td>
       </tr>

       <tr>
         <td>
           B15
         </td>

         <td>
            When accessing protected members of a superclass, the actual type of
            the instance being accessed must be either the current class or one
            of its subclasses.
         </td>

         <td>
           4.8.2.15
         </td>

         <td>
           VM
         </td>
       </tr>

      <tr>
         <td>
           B16
         </td>

         <td>
            The type of a value stored into a static field must be
            assignment-compatible with or convertible to the field's type.
         </td>

         <td>
           4.8.2.16
         </td>

         <td>
           VM
         </td>
       </tr>

       <tr>
         <td>
           B17
         </td>

         <td>
            The type of a value stored into a field must be assignment-compatible
            with or convertible to the field's type.
         </td>

         <td>
           4.8.2.17
         </td>

         <td>
           VM
         </td>
       </tr>

       <tr>
         <td>
           B18
         </td>

         <td>
            The type of every value stored into an array must be
            assignment-compatible with the array's component type.
         </td>

         <td>
           4.8.2.18
         </td>

         <td>
           VM
         </td>
       </tr>

       <tr>
         <td>
           B19
         </td>

         <td>
            The operand of an <code>athrow</code> instruction must be
            assignment-compatible with <code>java.lang.Throwable</code>.
         </td>

         <td>
           4.8.2.19
         </td>

         <td>
           VM
         </td>
       </tr>

       <tr>
         <td>
           B20
         </td>

         <td>
            The last reachable instruction of a method must either be a backwards
            jump or branch, a return, or an <code>athrow</code> instruction. It
            must not be possible to leave the <code>code</code> array at the
            bottom.
         </td>

         <td>
           4.8.2.20
         </td>

         <td>
           VM
         </td>
       </tr>

       <tr>
         <td>
           B21
         </td>

         <td>
            Local variable values must not be used as return addresses.
         </td>

         <td>
           4.8.2.21
         </td>

         <td>
           VM
         </td>
       </tr>

       <tr>
         <td>
           B22
         </td>

         <td>
           There must be a single, uniquely determined return instruction per
           subroutine call.
         </td>

         <td>
           4.8.2.22
         </td>

         <td>
           VM
         </td>
       </tr>

       <tr>
         <td>
           B23
         </td>

         <td>
           Subroutine calls must not be directly or indirectly self-recursive.
         </td>

         <td>
           4.8.2.23
         </td>

         <td>
           DX
         </td>
       </tr>

       <tr>
         <td>
           B24
         </td>

         <td>
            <code>ReturnAddress</code> instances must not be reused. If a
            subroutine returns to a <code>ReturnAddress</code> further up the
            stack than where its original call instruction is located, then all
            <code>ReturnAddress</code> instances further down the stack must
            never be used.
         </td>

         <td>
           4.8.2.24
         </td>

         <td>
           DX
         </td>
       </tr>

     </table>
   </body>
 </html>
	<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">

	<html>
	<head>
	<title>Java bytecode constraints</title>
	<link rel=stylesheet href="java-constraints.css">
	</head>

	<body>
	<h1>
	Bytecode constraints
	</h1>

	<p>
	From the point of view of a piece of code written in the Java
	programming language or targeted in the same way to <code>.class</code>
	files, the Dalvik VM aims to behave in a way
	that is fully consistent with the language's definition.
	That is, the code running in Dalvik will behave the same as it
	would have running in any other virtual machine. This includes
	verification failures.
	The Dx/Dalvik system will check roughly the same
	constraints that any other VM would, except as noted in the file
	<a href="verifier.html">verifier.html</a>. The following table briefly
	lists all Dx/Dalvik verification constraints together their analogs
	from the book <i>The Java<super>TM</super> Language Specification</i>,
	second edition. In the numbering scheme, the first three
	elements refer to the specification chapter, the fourth one to the
	bullet inside that chapter. The failure mode specifies whether the
	constraint will fail during the Dx conversion or during verification in
	the VM itself.
	</p>

	<h2>
	Static constraints
	</h2>

	<p>
	Static constraints are constraints on individual elements of the bytecode.
	They usually can be checked without employing control or data-flow analysis
	techniques.
	</p>

	<table>
	<tr>
	<th>
	Identifier
	</th>

	<th>
	Description
	</th>

	<th>
	Spec equivalent
	</th>

	<th>
	Failure mode
	</th>
	</tr>

	<tr>
	<td>
	A1
	</td>

	<td>
	The <code>code</code> array must not be empty.
	</td>

	<td>
	4.8.1.1
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A2
	</td>

	<td>
	The <code>code</code> array must not be larger than 65535 bytes.
	</td>

	<td>
	4.8.1.2
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A3
	</td>

	<td>
	The first opcode in <code>code</code> array must have index
	<code>0</code>.
	</td>

	<td>
	4.8.1.3
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A4
	</td>

	<td>
	The <code>code</code> array must only contain valid opcodes.
	</td>

	<td>
	4.8.1.4
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A5
	</td>

	<td>
	The index of instruction <code>n+1</code> must equal the index of
	instruction <code>n</code> plus the length of instruction
	<code>n</code>, taking into account a possible <code>wide</code>
	instruction. Opcodes modified by a <code>wide</code> instruction must
	not be directly reachable.
	</td>

	<td>
	4.8.1.5
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A6
	</td>

	<td>
	The last instruction in <code>code</code> array must end at index
	<code>code_length-1</code>.
	</td>

	<td>
	4.8.1.6
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A7
	</td>

	<td>
	All jump and branch targets must be opcodes within the same method.
	Opcodes modified by a <code>wide</code> instruction must not be
	directly reachable via a jump or branch instruction.
	</td>

	<td>
	4.8.1.7
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A8
	</td>

	<td>
	All targets of a <code>tableswitch</code> instruction must be opcodes
	within the same method. Upper and lower bounds must be consistent.
	Opcodes modified by a <code>wide</code> instruction must not be
	directly reachable via a <code>tableswitch</code> instruction.
	</td>

	<td>
	4.8.1.8
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A9
	</td>

	<td>
	All targets of a <code>lookupswitch</code> instruction must be opcodes
	within the same method. Its table must be consistent and sorted
	low-to-high. Opcodes modified by a <code>wide</code> instruction must
	not be directly reachable via a <code>lookupswitch</code> instruction.
	</td>

	<td>
	4.8.1.9
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A10
	</td>

	<td>
	The operands of <code>ldc</code> and <code>ldc_w</code> instructions
	must be valid indices into the constant pool. The respective entries
	must be of type <code>CONSTANT_Integer</code>,
	<code>CONSTANT_Float</code>, or <code>CONSTANT_String</code>.
	</td>

	<td>
	4.8.1.10
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A11
	</td>

	<td>
	The operands of <code>ldc2_w</code> instructions must be valid indices
	into the constant pool. The respective entries must be of type
	<code>CONSTANT_Long</code> or <code>CONSTANT_Double</code>. The
	subsequent constant pool entry must be valid and remain unused.
	</td>

	<td>
	4.8.1.11
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A12
	</td>

	<td>
	The Operands of <code>get<kind></code> and
	<code>put<kind></code> instructions must be valid indices into
	constant pool. The respective entries must be of type
	<code>CONSTANT_Fieldref</code>.
	</td>

	<td>
	4.8.1.12
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A13
	</td>

	<td>
	The first two operands of <code>invokevirtual</code>,
	<code>invokespecial</code>, and <code>invokestatic</code> must form a
	valid 16-bit index into the constant pool. The respective entries must
	be of type <code>CONSTANT_Methodref</code>.
	</td>

	<td>
	4.8.1.13
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A14
	</td>

	<td>
	Methods whose names start with '<' must only be invoked implicitly by
	the VM, not by class file code. The only exception is the instance
	initializer, which may be invoked by <code>invokespecial</code>.
	</td>

	<td>
	4.8.1.14
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A15
	</td>

	<td>
	The first two operands of <code>invokeinterface</code> must form a
	valid 16-bit index into the constant pool. The entry must be of type
	<code>CONSTANT_Interface_Methodref</code>. The third operand must
	specify number of local variables and the fourth operand must always
	be zero.
	</td>

	<td>
	4.8.1.15
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A16
	</td>

	<td>
	The operands of <code>instanceof</code>, <code>checkcast</code>,
	<code>new</code>, and <code>anewarray</code> instructions must
	be a valid index into the constant pool. The first two operands of
	<code>multianewarray</code> instruction must form a valid 16-bit index
	into the constant pool. All respective entries must be of type
	<code>CONSTANT_Class</code>.
	</td>

	<td>
	4.8.1.16
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A17
	</td>

	<td>
	The dimensions of an array created by <code>anewarray</code>
	instructions must be less than <code>256</code>.
	</td>

	<td>
	4.8.1.17
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A18
	</td>

	<td>
	The <code>new</code> instruction must not reference array classes,
	interfaces, or abstract classes.
	</td>

	<td>
	4.8.1.18
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A19
	</td>

	<td>
	The type referenced by a <code>multinewarray</code> instruction must
	have at least as many dimensions as specified in the instruction. The
	dimensions operand must not be <code>0</code>
	</td>

	<td>
	4.8.1.19
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A20
	</td>

	<td>
	The type referenced by a <code>newarray</code> instruction must be a
	valid, non-reference type.
	</td>

	<td>
	4.8.1.20
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A21
	</td>

	<td>
	The index operand of instructions explicitly referencing single-width
	local variables must be non-negative and smaller than
	<code>max_locals</code>.
	</td>

	<td>
	4.8.1.21
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A22
	</td>

	<td>
	The index operand of instructions implicitly referencing single-width
	local variables must be non-negative and smaller than
	<code>max_locals</code>.
	</td>

	<td>
	4.8.1.22
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A23
	</td>

	<td>
	The index operand of instructions explicitly referencing double-width
	local variables must be non-negative and smaller than
	<code>max_locals-1</code>.
	</td>

	<td>
	4.8.1.23
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A24
	</td>

	<td>
	The index operand of instructions implicitly referencing double-width
	local variables must be non-negative and smaller than
	<code>max_locals-1</code>.
	</td>

	<td>
	4.8.1.24
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A25
	</td>

	<td>
	The index operand of <code>wide</code> instructions explicitly
	referencing single-width local variables must be non-negative and
	smaller than <code>max_locals</code>.
	</td>

	<td>
	4.8.1.25
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	A26
	</td>

	<td>
	The index operand of <code>wide</code> instructions explicitly
	referencing double-width local variables must be non-negative and
	smaller than <code>max_locals-1</code>.
	</td>

	<td>
	4.8.1.25
	</td>

	<td>
	DX
	</td>
	</tr>
	</table>

	<h2>
	Structural constraints
	</h2>

	<p>
	Structural constraints are constraints on relationships between several
	elements of the bytecode. They usually can't be checked without employing
	control or data-flow analysis techniques.
	</p>

	<table>
	<tr>
	<th>
	Identifier
	</th>

	<th>
	Description
	</th>

	<th>
	Spec equivalent
	</th>

	<th>
	Failure mode
	</th>
	</tr>

	<tr>
	<td>
	B1
	</td>

	<td>
	The number and types of arguments (operands and local variables) must
	always match the instruction.
	</td>

	<td>
	4.8.2.1
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	B2
	</td>

	<td>
	The operand stack must have the same depth for all executions paths
	leading to an instruction.
	</td>

	<td>
	4.8.2.2
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	B3
	</td>

	<td>
	Local variable pairs must never be broken up.
	</td>

	<td>
	4.8.2.3
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	B4
	</td>

	<td>
	A local variable (or pair) has to be assigned first before it can be
	read.
	</td>

	<td>
	4.8.2.4
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	B5
	</td>

	<td>
	The operand stack must never grow beyond <code>max_stack</code>.
	</td>

	<td>
	4.8.2.5
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	B6
	</td>

	<td>
	The operand stack must never underflow.
	</td>

	<td>
	4.8.2.6
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	B7
	</td>

	<td>
	An <code>invokespecial</code> instruction must only invoke an instance
	initializer or a method in the current class or one of its
	superclasses.
	</td>

	<td>
	4.8.2.7
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B8
	</td>

	<td>
	An instance initializer must only be invoked on an uninitialized
	instance residing on the operand stack.
	</td>

	<td>
	4.8.2.8
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B9
	</td>

	<td>
	Instance methods may only be invoked on and instance fields may only
	be accessed on already initialized instances.
	</td>

	<td>
	4.8.2.9
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B10
	</td>

	<td>
	The must be no backwards branches with uninitialized instances on the
	operand stack or in local variables. There must be no code protected
	by an exception handler that contains local variables with
	uninitialized instances.
	</td>

	<td>
	4.8.2.10
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	B11
	</td>

	<td>
	An instance initializer must call another instance initializer (same
	class or superclass) before any instance members can be accessed.
	Exceptions are non-inherited instance fields, which can be assigned
	before calling another initializer, and the <code>Object</code> class
	in general.
	</td>

	<td>
	4.8.2.11
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B12
	</td>

	<td>
	All actual method arguments must be assignment-compatible with formal
	arguments.
	</td>

	<td>
	4.8.2.12
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B13
	</td>

	<td>
	For each instance method invocation, the actual instance must be
	assignment-compatible with the class or interface specified in the
	instruction.
	</td>

	<td>
	4.8.2.13
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B14
	</td>

	<td>
	A returns instruction must match its method's return type.
	</td>

	<td>
	4.8.2.14
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B15
	</td>

	<td>
	When accessing protected members of a superclass, the actual type of
	the instance being accessed must be either the current class or one
	of its subclasses.
	</td>

	<td>
	4.8.2.15
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B16
	</td>

	<td>
	The type of a value stored into a static field must be
	assignment-compatible with or convertible to the field's type.
	</td>

	<td>
	4.8.2.16
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B17
	</td>

	<td>
	The type of a value stored into a field must be assignment-compatible
	with or convertible to the field's type.
	</td>

	<td>
	4.8.2.17
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B18
	</td>

	<td>
	The type of every value stored into an array must be
	assignment-compatible with the array's component type.
	</td>

	<td>
	4.8.2.18
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B19
	</td>

	<td>
	The operand of an <code>athrow</code> instruction must be
	assignment-compatible with <code>java.lang.Throwable</code>.
	</td>

	<td>
	4.8.2.19
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B20
	</td>

	<td>
	The last reachable instruction of a method must either be a backwards
	jump or branch, a return, or an <code>athrow</code> instruction. It
	must not be possible to leave the <code>code</code> array at the
	bottom.
	</td>

	<td>
	4.8.2.20
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B21
	</td>

	<td>
	Local variable values must not be used as return addresses.
	</td>

	<td>
	4.8.2.21
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B22
	</td>

	<td>
	There must be a single, uniquely determined return instruction per
	subroutine call.
	</td>

	<td>
	4.8.2.22
	</td>

	<td>
	VM
	</td>
	</tr>

	<tr>
	<td>
	B23
	</td>

	<td>
	Subroutine calls must not be directly or indirectly self-recursive.
	</td>

	<td>
	4.8.2.23
	</td>

	<td>
	DX
	</td>
	</tr>

	<tr>
	<td>
	B24
	</td>

	<td>
	<code>ReturnAddress</code> instances must not be reused. If a
	subroutine returns to a <code>ReturnAddress</code> further up the
	stack than where its original call instruction is located, then all
	<code>ReturnAddress</code> instances further down the stack must
	never be used.
	</td>

	<td>
	4.8.2.24
	</td>

	<td>
	DX
	</td>
	</tr>

	</table>
	</body>
	</html>