page.title=Constraints
@jd:body

<!--
    Copyright 2015 The Android Open Source Project

    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.
    You may obtain a copy of the License at

        http://www.apache.org/licenses/LICENSE-2.0

    Unless required by applicable law or agreed to in writing, software
    distributed under the License is distributed on an "AS IS" BASIS,
    WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    See the License for the specific language governing permissions and
    limitations under the License.
-->


<div id="qv-wrapper">
<div id="qv">
  <h2 id="Contents">In this document</h2>
  <ol id="auto-toc">
  </ol>
</div>
</div>

<p>A <code>.dex</code> file is the transport format for Dalvik Bytecode. There are certain
syntactical and semantical constraints for a file to be a valid <code>.dex</code> file, and
a runtime is required to support only valid .dex files.</p>

    <h2 id="gen-constraints">General .dex integrity constraints</h2>

    <p>General integrity constraints are concerned with the larger structure of a
    <code>.dex</code> file, as described in detail in <a href="dex-format.html"><code>.dex</code>
    format</a>.</p>

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

        <th>
          Description
        </th>
      </tr>

      <tr>
        <td>
          G1
        </td>

        <td>
          The <code>magic</code> number of the <code>.dex</code> file must be
          <code>dex\n035\0</code> or <code>dex\n037\0</code>.
        </td>
      </tr>

      <tr>
        <td>
          G2
        </td>

        <td>
          The checksum must be an Adler-32 checksum of the whole file contents
          except <code>magic</code> and <code>checksum</code> field.
        </td>
      </tr>

      <tr>
        <td>
          G3
        </td>

        <td>
          The signature must be a SHA-1 hash of the whole file contents except <code>magic</code>,
          <code>checksum</code>, and <code>signature</code>.
        </td>
      </tr>

      <tr>
        <td>
          G4
        </td>

        <td>
          The <code>file_size</code> must match the actual file size in bytes.
        </td>
      </tr>


      <tr>
        <td>
          G5
        </td>

        <td>
          The <code>header_size</code> must have the value: <code>0x70</code>
        </td>
      </tr>

      <tr>
        <td>
          G6
        </td>

        <td>
          The <code>endian_tag</code> must have either the value:
          <code>ENDIAN_CONSTANT</code> or <code>REVERSE_ENDIAN_CONSTANT</code>
        </td>
      </tr>

      <tr>
        <td>
          G7
        </td>

        <td>
          For each of the <code>link</code>, <code>string_ids</code>,
          <code>type_ids</code>, <code>proto_ids</code>,
          <code>field_ids</code>, <code>method_ids</code>, <code>class_defs</code>, and
          <code>data</code> sections, the <code>offset</code> and <code>size</code> fields must be
          either both zero or both non-zero. In the latter case, the offset must be
          four-byte-aligned.
        </td>
      </tr>

      <tr>
        <td>
          G8
        </td>

        <td>
          All offset fields in the header except <code>map_off</code> must be four-byte-aligned.
        </td>
      </tr>

      <tr>
        <td>
          G9
        </td>

        <td>
          The <code>map_off</code> field must be either zero or point into the
          data section. In the latter case, the <code>data</code> section must exist.
        </td>
      </tr>

      <tr>
        <td>
          G10
        </td>

        <td>
          None of the <code>link</code>, <code>string_ids</code>,
          <code>type_ids</code>, <code>proto_ids</code>, <code>field_ids</code>,
          <code>method_ids</code>, <code>class_defs</code> and <code>data</code> sections
          must overlap each other or the header.
        </td>
      </tr>

      <tr>
        <td>
          G11
        </td>

        <td>
          If a map exists, then each map entry must have a valid type. Each type may appear at most once.
        </td>
      </tr>

      <tr>
        <td>
          G12
        </td>

        <td>
          If a map exists, then each map entry must have a non-zero offset and
          size. The offset must point into the corresponding section of the file (i.e. a
          <code>string_id_item</code> must point into the <code>string_ids</code> section) and the explicit or
          implicit size of the item must match the actual contents and size of the
          section.
        </td>
      </tr>

      <tr>
        <td>
          G13
        </td>

        <td>
          If a map exists, then the offset of map entry <code>n+1</code> must be greater or
          equal to the offset of map entry <code>n plus than size of map entry n</code>. This implies
          non-overlapping entries and low-to-high ordering.
        </td>
      </tr>

      <tr>
        <td>
          G14
        </td>

        <td>
          The following types of entries must have an offset that is
          four-byte-aligned: <code>string_id_item</code>,
          <code>type_id_item</code>, <code>proto_id_item</code>,
          <code>field_id_item</code>,
          <code>method_id_item</code>, <code>class_def_item</code>,
          <code>type_list</code>, <code>code_item</code>,
          <code>annotations_directory_item</code>.
        </td>
      </tr>

      <tr>
        <td>
          G15
        </td>

        <td>
          For each <code>string_id_item</code>, the <code>string_data_off</code> field must contain a
          valid reference into the <code>data</code> section. For the referenced <code>string_data_item</code>, the
          <code>data</code> field must contain a valid MUTF-8 string, and the <code>utf16_size</code> must match
          the decoded length of the string.
        </td>
      </tr>

      <tr>
        <td>
          G16
        </td>

        <td>
          For each <code>type_id_item</code>, the <code>descriptor_idx</code> field must contain a valid
          reference into the <code>string_ids</code> list. The referenced string must be a valid type
          descriptor.
        </td>
      </tr>

      <tr>
        <td>
          G17
        </td>

        <td>
          For each <code>proto_id_item</code>, the <code>shorty_idx</code> field must contain a valid
          reference into the <code>string_ids</code> list. The referenced string must be a valid
          shorty descriptor. Also, the <code>return_type_idx</code> field must be a valid index into
          the <code>type_ids</code> section, and the <code>parameters_off</code> field must be either zero or a
          valid offset pointing into the <code>data</code> section. If non-zero, the parameter list
          must not contain any void entries.
        </td>
      </tr>

      <tr>
        <td>
          G18
        </td>

        <td>
          For each <code>field_id_item</code>, both the <code>class_idx</code> and <code>type_idx</code> fields must
          be valid indices into the <code>type_ids</code> list. The entry referenced by <code>class_idx</code>
          must be a non-array reference type. In addition, the <code>name_idx</code> field must be a
          valid reference into the <code>string_ids</code> section, and the contents of the referenced
          entry must conform to the <code>MemberName</code> specification.
        </td>
      </tr>

      <tr>
        <td>
          G19
        </td>

        <td>
          For each <code>method_id_item</code>, the <code>class_idx</code> field must be a valid index
          into the <code>type_ids</code> section, and the referenced entry must be a non-array
          reference type. The <code>proto_id</code> field must be a valid reference into the <code>proto_ids</code>
          list. The <code>name_idx</code> field must be a valid reference into the <code>string_ids</code> section,
          and the contents of the referenced entry must conform to the <code>MemberName</code>
          specification.
        </td>
      </tr>

      <tr>
        <td>
          G20
        </td>

        <td>
          For each <code>field_id_item</code>, the <code>class_idx</code> field must be a valid index
          into the <code>type_ids</code> list. The referenced entry must be a non-array reference
          type.
        </td>
      </tr>

</table>

<h2 id="static-constraints">
      Static bytecode 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>
      </tr>

      <tr>
        <td>
          A1
        </td>

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

      <tr>
        <td>
          A2
        </td>

        <td>
          The first opcode in the <code>insns</code> array must have index zero.
        </td>
      </tr>

      <tr>
        <td>
          A3
        </td>

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

      <tr>
        <td>
          A4
        </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 possible operands.
        </td>
      </tr>

      <tr>
        <td>
          A5
        </td>

        <td>
          The last instruction in the <code>insns</code> array must end at index
          <code>insns_size-1</code>.
        </td>
      </tr>

      <tr>
        <td>
          A6
        </td>

        <td>
          All <code>goto</code> and <code>if-&lt;kind&gt;</code> targets must
          be opcodes within the same method.
        </td>
      </tr>

      <tr>
        <td>
          A7
        </td>

        <td>
          All targets of a <code>packed-switch</code> instruction must be
          opcodes within the same method. The size and the list of targets
          must be consistent.
        </td>
      </tr>

      <tr>
        <td>
          A8
        </td>

        <td>
          All targets of a <code>sparse-switch</code> instruction must be
          opcodes within the same method. The corresponding table must be
          consistent and sorted low-to-high.
        </td>
      </tr>

      <tr>
        <td>
          A9
        </td>

        <td>
          The <code>B</code> operand of the <code>const-string</code> and
          <code>const-string/jumbo</code> instructions must be a valid index
          into the string constant pool.
        </td>
      </tr>

      <tr>
        <td>
          A10
        </td>

        <td>
          The <code>C</code> operand of the <code>iget&lt;kind&gt;</code> and
          <code>iput&lt;kind&gt;</code> instructions must be a valid index into
          the field constant pool. The referenced entry must represent an
          instance field.
        </td>
      </tr>

      <tr>
        <td>
          A11
        </td>

        <td>
          The <code>C</code> operand of the <code>sget&lt;kind&gt;</code> and
          <code>sput&lt;kind&gt;</code> instructions must be a valid index into
          the field constant pool. The referenced entry must represent a static
          field.
        </td>
      </tr>

      <tr>
        <td>
          A12
        </td>

        <td>
          The <code>C</code> operand of the <code>invoke-virtual</code>,
          <code>invoke-super</code>, <code>invoke-direct</code> and
          <code>invoke-static</code> instructions must be a valid index into the
          method constant pool.
        </td>
      </tr>

      <tr>
        <td>
          A13
        </td>

        <td>
          The <code>B</code> operand of the <code>invoke-virtual/range</code>,
          <code>invoke-super/range</code>, <code>invoke-direct/range</code>, and
          <code>invoke-static/range</code> instructions must be a valid index
          into the method constant pool.
        </td>
      </tr>

      <tr>
        <td>
          A14
        </td>

        <td>
          A method the name of which starts with a '&lt;' must only be invoked
          implicitly by the VM, not by code originating from a <code>.dex</code> file. The
          only exception is the instance initializer, which may be invoked by
          <code>invoke-direct</code>.
        </td>
      </tr>

      <tr>
        <td>
          A15
        </td>

        <td>
          The <code>C</code> operand of the <code>invoke-interface</code>
          instruction must be a valid index into the method constant pool. The
          referenced <code>method_id</code> must belong to an interface (not a
          class).
        </td>
      </tr>

      <tr>
        <td>
          A16
        </td>

        <td>
          The <code>B</code> operand of the <code>invoke-interface/range</code>
          instruction must be a valid index into the method constant pool.
          The referenced <code>method_id</code> must belong to an interface (not
          a class).
        </td>
      </tr>

      <tr>
        <td>
          A17
        </td>

        <td>
          The <code>B</code> operand of the <code>const-class</code>,
          <code>check-cast</code>, <code>new-instance</code>, and
          <code>filled-new-array/range</code> instructions must be a valid index
          into the type constant pool.
        </td>
      </tr>

      <tr>
        <td>
          A18
        </td>

        <td>
          The <code>C</code> operand of the <code>instance-of</code>,
          <code>new-array</code>, and <code>filled-new-array</code>
          instructions must be a valid index into the type constant pool.
        </td>
      </tr>

     <tr>
        <td>
          A19
        </td>

        <td>
          The dimensions of an array created by a <code>new-array</code>
          instruction must be less than <code>256</code>.
        </td>
      </tr>

      <tr>
        <td>
          A20
        </td>

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

      <tr>
        <td>
          A21
        </td>

        <td>
          The type referred to by a <code>new-array</code> instruction must be
          a valid, non-reference type.
        </td>
      </tr>

      <tr>
        <td>
          A22
        </td>

        <td>
          All registers referred to by an instruction in a single-width
          (non-pair) fashion must be valid for the current method. That is,
          their indices must be non-negative and smaller than
          <code>registers_size</code>.
        </td>
      </tr>

      <tr>
        <td>
          A23
        </td>

        <td>
          All registers referred to by an instruction in a double-width (pair)
          fashion must be valid for the current method. That is, their indices
          must be non-negative and smaller than <code>registers_size-1</code>.
        </td>
      </tr>
      <tr>
        <td>
          A24
        </td>

        <td>
          The <code>method_id</code> operand of the <code>invoke-virtual</code>
          and <code>invoke-direct</code> instructions must belong to a class
          (not an interface). In Dex files prior to version <code>037</code>
          the same must be true of <code>invoke-super</code> and
          <code>invoke-static</code> instructions.
        </td>
      </tr>

      <tr>
        <td>
          A25
        </td>

        <td>
          The <code>method_id</code> operand of the
          <code>invoke-virtual/range</code> and
          <code>invoke-direct/range</code> instructions must belong to a class
          (not an interface). In Dex files prior to version <code>037</code>
          the same must be true of <code>invoke-super/range</code> and
          <code>invoke-static/range</code> instructions.
        </td>
      </tr>
    </table>

    <h2 id="struct-constraints">
      Structural bytecode 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>
      </tr>

      <tr>
        <td>
          B1
        </td>

        <td>
          The number and types of arguments (registers and immediate values)
          must always match the instruction.
        </td>
      </tr>

      <tr>
        <td>
          B2
        </td>

        <td>
          Register pairs must never be broken up.
        </td>
      </tr>

      <tr>
        <td>
          B3
        </td>

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

      <tr>
        <td>
          B4
        </td>

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

      <tr>
        <td>
          B5
        </td>

        <td>
          An instance initializer must be invoked only on an uninitialized
          instance.
        </td>
      </tr>

      <tr>
        <td>
          B6
        </td>

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

      <tr>
        <td>
          B7
        </td>

        <td>
          A register that holds the result of a <code>new-instance</code>
          instruction must not be used if the same
          <code>new-instance</code> instruction is again executed before
          the instance is initialized.
        </td>
      </tr>

      <tr>
        <td>
          B8
        </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>
      </tr>

      <tr>
        <td>
          B9
        </td>

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

      <tr>
        <td>
          B10
        </td>

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

      <tr>
        <td>
          B11
        </td>

        <td>
           A <code>return&lt;kind&gt;</code> instruction must match its
           method's return type.
        </td>
      </tr>

      <tr>
        <td>
          B12
        </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>
      </tr>

     <tr>
        <td>
          B13
        </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>
      </tr>

      <tr>
        <td>
          B14
        </td>

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

      <tr>
        <td>
          B15
        </td>

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

      <tr>
        <td>
          B16
        </td>

        <td>
           The <code>A</code> operand of a <code>throw</code> instruction must
           be assignment-compatible with <code>java.lang.Throwable</code>.
        </td>
      </tr>

      <tr>
        <td>
          B17
        </td>

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

      <tr>
        <td>
          B18
        </td>

        <td>
          The unassigned half of a former register pair may not be read (is
          considered invalid) until it has been re-assigned by some other
          instruction.
        </td>
      </tr>

      <tr>
        <td>
          B19
        </td>

        <td>
          A <code>move-result&lt;kind&gt;</code> instruction must be immediately
          preceded (in the <code>insns</code> array) by an
          <code>invoke-&lt;kind&gt;</code> instruction. The only exception is
          the <code>move-result-object</code> instruction, which may also be
          preceded by a <code>filled-new-array</code> instruction.
        </td>
      </tr>

      <tr>
        <td>
          B20
        </td>

        <td>
          A <code>move-result&lt;kind&gt;</code> instruction must be immediately
          preceded (in actual control flow) by a matching
          <code>return-&lt;kind&gt;</code> instruction (it must not be jumped
          to). The only exception is the <code>move-result-object</code>
          instruction, which may also be preceded by a
          <code>filled-new-array</code> instruction.
        </td>
      </tr>

      <tr>
        <td>
          B21
        </td>

        <td>
          A <code>move-exception</code> instruction must appear only as the
          first instruction in an exception handler.
        </td>
      </tr>

      <tr>
        <td>
          B22
        </td>

        <td>
          The <code>packed-switch-data</code>, <code>sparse-switch-data</code>,
          and <code>fill-array-data</code> pseudo-instructions must not be
          reachable by control flow.
        </td>
      </tr>
    </table>