tech/dalvik/instruction-formats.jd

page.title=Dalvik VM Instruction Formats
@jd:body

<!--
    Copyright 2013 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.
-->
<h2>Introduction and Overview</h2>

<p>This document lists the instruction formats used by Dalvik bytecode
and is meant to be used in conjunction with the
<a href="dalvik-bytecode.html">bytecode reference document</a>.</p>

<h3>Bitwise descriptions</h3>

<p>The first column in the format table lists the bitwise layout of
the format. It consists of one or more space-separated "words" each of
which describes a 16-bit code unit. Each character in a word
represents four bits, read from high bits to low, with vertical bars
("<code>|</code>") interspersed to aid in reading. Uppercase letters
in sequence from "<code>A</code>" are used to indicate fields within
the format (which then get defined further by the syntax column). The term
"<code>op</code>" is used to indicate the position of an eight-bit
opcode within the format. A slashed zero
("<code>&Oslash;</code>") is used to indicate that all bits must be
zero in the indicated position.</p>

<p>For the most part, lettering proceeds from earlier code units to
later code units, and low-order to high-order within a code unit.
However, there are a few exceptions to this general rule, which are
done in order to make the naming of similar-meaning parts be the same
across different instruction formats. These cases are noted explicitly
in the format descriptions.</p>

<p>For example, the format "<code>B|A|<i>op</i> CCCC</code>" indicates
that the format consists of two 16-bit code units. The first word
consists of the opcode in the low eight bits and a pair of four-bit
values in the high eight bits; and the second word consists of a single
16-bit value.</p>

<h3>Format IDs</h3>

<p>The second column in the format table indicates the short identifier
for the format, which is used in other documents and in code to identify
the format.</p>

<p>Most format IDs consist of three characters, two digits followed by a
letter. The first digit indicates the number of 16-bit code units in the
format. The second digit indicates the maximum number of registers that the
format contains (maximum, since some formats can accomodate a variable
number of registers), with the special designation "<code>r</code>" indicating
that a range of registers is encoded. The final letter semi-mnemonically
indicates the type of any extra data encoded by the format. For example,
format "<code>21t</code>" is of length two, contains one register reference,
and additionally contains a branch target.</p>

<p>Suggested static linking formats have an additional
"<code>s</code>" suffix, making them four characters total. Similarly,
suggested "inline" linking formats have an additional "<code>i</code>"
suffix. (In this context, inline linking is like static linking,
except with more direct ties into a virtual machine's implementation.)
Finally, a couple oddball suggested formats (e.g.,
"<code>20bc</code>") include two pieces of data which are both
represented in its format ID.</p>

<p>The full list of typecode letters are as follows. Note that some
forms have different sizes, depending on the format:</p>

<table class="letters">
<thead>
<tr>
  <th>Mnemonic</th>
  <th>Bit Sizes</th>
  <th>Meaning</th>
</tr>
</thead>
<tbody>
<tr>
  <td>b</td>
  <td>8</td>
  <td>immediate signed <b>b</b>yte</td>
</tr>
<tr>
  <td>c</td>
  <td>16, 32</td>
  <td><b>c</b>onstant pool index</td>
</tr>
<tr>
  <td>f</td>
  <td>16</td>
  <td>inter<b>f</b>ace constants (only used in statically linked formats)
  </td>
</tr>
<tr>
  <td>h</td>
  <td>16</td>
  <td>immediate signed <b>h</b>at (high-order bits of a 32- or 64-bit
    value; low-order bits are all <code>0</code>)
  </td>
</tr>
<tr>
  <td>i</td>
  <td>32</td>
  <td>immediate signed <b>i</b>nt, or 32-bit float</td>
</tr>
<tr>
  <td>l</td>
  <td>64</td>
  <td>immediate signed <b>l</b>ong, or 64-bit double</td>
</tr>
<tr>
  <td>m</td>
  <td>16</td>
  <td><b>m</b>ethod constants (only used in statically linked formats)</td>
</tr>
<tr>
  <td>n</td>
  <td>4</td>
  <td>immediate signed <b>n</b>ibble</td>
</tr>
<tr>
  <td>s</td>
  <td>16</td>
  <td>immediate signed <b>s</b>hort</td>
</tr>
<tr>
  <td>t</td>
  <td>8, 16, 32</td>
  <td>branch <b>t</b>arget</td>
</tr>
<tr>
  <td>x</td>
  <td>0</td>
  <td>no additional data</td>
</tr>
</tbody>
</table>

<h3>Syntax</h3>

<p>The third column of the format table indicates the human-oriented
syntax for instructions which use the indicated format. Each instruction
starts with the named opcode and is optionally followed by one or
more arguments, themselves separated with commas.</p>

<p>Wherever an argument refers to a field from the first column, the
letter for that field is indicated in the syntax, repeated once for
each four bits of the field. For example, an eight-bit field labeled
"<code>BB</code>" in the first column would also be labeled
"<code>BB</code>" in the syntax column.</p>

<p>Arguments which name a register have the form "<code>v<i>X</i></code>".
The prefix "<code>v</code>" was chosen instead of the more common
"<code>r</code>" exactly to avoid conflicting with (non-virtual) architectures
on which a Dalvik virtual machine might be implemented which themselves
use the prefix "<code>r</code>" for their registers. (That is, this
decision makes it possible to talk about both virtual and real registers
together without the need for circumlocution.)</p>

<p>Arguments which indicate a literal value have the form
"<code>#+<i>X</i></code>". Some formats indicate literals that only
have non-zero bits in their high-order bits; for these, the zeroes
are represented explicitly in the syntax, even though they do not
appear in the bitwise representation.</p>

<p>Arguments which indicate a relative instruction address offset have the
form "<code>+<i>X</i></code>".</p>

<p>Arguments which indicate a literal constant pool index have the form
"<code><i>kind</i>@<i>X</i></code>", where "<code><i>kind</i></code>"
indicates which constant pool is being referred to. Each opcode that
uses such a format explicitly allows only one kind of constant; see
the opcode reference to figure out the correspondence. The four
kinds of constant pool are "<code>string</code>" (string pool index),
"<code>type</code>" (type pool index), "<code>field</code>" (field
pool index), and "<code>meth</code>" (method pool index).</p>

<p>Similar to the representation of constant pool indices, there are
also suggested (optional) forms that indicate prelinked offsets or
indices. There are two types of suggested prelinked value: vtable offsets
(indicated as "<code>vtaboff</code>") and field offsets (indicated as
"<code>fieldoff</code>").</p>

<p>In the cases where a format value isn't explictly part of the syntax
but instead picks a variant, each variant is listed with the prefix
"<code>[<i>X</i>=<i>N</i>]</code>" (e.g., "<code>[A=2]</code>") to indicate
the correspondence.</p>

<h2>The Formats</h2>

<table class="format">
<thead>
<tr>
  <th>Format</th>
  <th>ID</th>
  <th>Syntax</th>
  <th>Notable Opcodes Covered</th>
</tr>
</thead>
<tbody>
<tr>
  <td><i>N/A</i></td>
  <td>00x</td>
  <td><i><code>N/A</code></i></td>
  <td><i>pseudo-format used for unused opcodes; suggested for use as the
    nominal format for a breakpoint opcode</i></td>
</tr>
<tr>
  <td>&Oslash;&Oslash;|<i>op</i></td>
  <td>10x</td>
  <td><i><code>op</code></i></td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td rowspan="2">B|A|<i>op</i></td>
  <td>12x</td>
  <td><i><code>op</code></i> vA, vB</td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td>11n</td>
  <td><i><code>op</code></i> vA, #+B</td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td rowspan="2">AA|<i>op</i></td>
  <td>11x</td>
  <td><i><code>op</code></i> vAA</td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td>10t</td>
  <td><i><code>op</code></i> +AA</td>
  <td>goto</td>
</tr>
<tr>
  <td>&Oslash;&Oslash;|<i>op</i> AAAA</td></td>
  <td>20t</td>
  <td><i><code>op</code></i> +AAAA</td>
  <td>goto/16</td>
</tr>
<tr>
  <td>AA|<i>op</i> BBBB</td></td>
  <td>20bc</td>
  <td><i><code>op</code></i> AA, kind@BBBB</td>
  <td><i>suggested format for statically determined verification errors;
    A is the type of error and B is an index into a type-appropriate
    table (e.g. method references for a no-such-method error)</i></td>
</tr>
<tr>
  <td rowspan="5">AA|<i>op</i> BBBB</td>
  <td>22x</td>
  <td><i><code>op</code></i> vAA, vBBBB</td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td>21t</td>
  <td><i><code>op</code></i> vAA, +BBBB</td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td>21s</td>
  <td><i><code>op</code></i> vAA, #+BBBB</td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td>21h</td>
  <td><i><code>op</code></i> vAA, #+BBBB0000<br/>
    <i><code>op</code></i> vAA, #+BBBB000000000000
  </td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td>21c</td>
  <td><i><code>op</code></i> vAA, type@BBBB<br/>
    <i><code>op</code></i> vAA, field@BBBB<br/>
    <i><code>op</code></i> vAA, string@BBBB
  </td>
  <td>check-cast<br/>
    const-class<br/>
    const-string
  </td>
</tr>
<tr>
  <td rowspan="2">AA|<i>op</i> CC|BB</td>
  <td>23x</td>
  <td><i><code>op</code></i> vAA, vBB, vCC</td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td>22b</td>
  <td><i><code>op</code></i> vAA, vBB, #+CC</td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td rowspan="4">B|A|<i>op</i> CCCC</td>
  <td>22t</td>
  <td><i><code>op</code></i> vA, vB, +CCCC</td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td>22s</td>
  <td><i><code>op</code></i> vA, vB, #+CCCC</td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td>22c</td>
  <td><i><code>op</code></i> vA, vB, type@CCCC<br/>
    <i><code>op</code></i> vA, vB, field@CCCC
  </td>
  <td>instance-of</td>
</tr>
<tr>
  <td>22cs</td>
  <td><i><code>op</code></i> vA, vB, fieldoff@CCCC</td>
  <td><i>suggested format for statically linked field access instructions of
    format 22c</i>
  </td>
</tr>
<tr>
  <td>&Oslash;&Oslash;|<i>op</i> AAAA<sub>lo</sub> AAAA<sub>hi</sub></td></td>
  <td>30t</td>
  <td><i><code>op</code></i> +AAAAAAAA</td>
  <td>goto/32</td>
</tr>
<tr>
  <td>&Oslash;&Oslash;|<i>op</i> AAAA BBBB</td>
  <td>32x</td>
  <td><i><code>op</code></i> vAAAA, vBBBB</td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td rowspan="3">AA|<i>op</i> BBBB<sub>lo</sub> BBBB<sub>hi</sub></td>
  <td>31i</td>
  <td><i><code>op</code></i> vAA, #+BBBBBBBB</td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td>31t</td>
  <td><i><code>op</code></i> vAA, +BBBBBBBB</td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td>31c</td>
  <td><i><code>op</code></i> vAA, string@BBBBBBBB</td>
  <td>const-string/jumbo</td>
</tr>
<tr>
  <td rowspan="3">A|G|<i>op</i> BBBB F|E|D|C</td>
  <td>35c</td>
  <td><i>[<code>A=5</code>] <code>op</code></i> {vC, vD, vE, vF, vG},
    meth@BBBB<br/>
    <i>[<code>A=5</code>] <code>op</code></i> {vC, vD, vE, vF, vG},
    type@BBBB<br/>
    <i>[<code>A=4</code>] <code>op</code></i> {vC, vD, vE, vF},
    <i><code>kind</code></i>@BBBB<br/>
    <i>[<code>A=3</code>] <code>op</code></i> {vC, vD, vE},
    <i><code>kind</code></i>@BBBB<br/>
    <i>[<code>A=2</code>] <code>op</code></i> {vC, vD},
    <i><code>kind</code></i>@BBBB<br/>
    <i>[<code>A=1</code>] <code>op</code></i> {vC},
    <i><code>kind</code></i>@BBBB<br/>
    <i>[<code>A=0</code>] <code>op</code></i> {},
    <i><code>kind</code></i>@BBBB<br/>
    <p><i>The unusual choice in lettering here reflects a desire to make
    the count and the reference index have the same label as in format
    3rc.</i></p>
  </td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td>35ms</td>
  <td><i>[<code>A=5</code>] <code>op</code></i> {vC, vD, vE, vF, vG},
    vtaboff@BBBB<br/>
    <i>[<code>A=4</code>] <code>op</code></i> {vC, vD, vE, vF},
    vtaboff@BBBB<br/>
    <i>[<code>A=3</code>] <code>op</code></i> {vC, vD, vE},
    vtaboff@BBBB<br/>
    <i>[<code>A=2</code>] <code>op</code></i> {vC, vD},
    vtaboff@BBBB<br/>
    <i>[<code>A=1</code>] <code>op</code></i> {vC},
    vtaboff@BBBB<br/>
    <p><i>The unusual choice in lettering here reflects a desire to make
    the count and the reference index have the same label as in format
    3rms.</i></p>
  </td>
  <td><i>suggested format for statically linked <code>invoke-virtual</code>
    and <code>invoke-super</code> instructions of format 35c</i>
  </td>
</tr>
<tr>
  <td>35mi</td>
  <td><i>[<code>A=5</code>] <code>op</code></i> {vC, vD, vE, vF, vG},
    inline@BBBB<br/>
    <i>[<code>A=4</code>] <code>op</code></i> {vC, vD, vE, vF},
    inline@BBBB<br/>
    <i>[<code>A=3</code>] <code>op</code></i> {vC, vD, vE},
    inline@BBBB<br/>
    <i>[<code>A=2</code>] <code>op</code></i> {vC, vD},
    inline@BBBB<br/>
    <i>[<code>A=1</code>] <code>op</code></i> {vC},
    inline@BBBB<br/>
    <p><i>The unusual choice in lettering here reflects a desire to make
    the count and the reference index have the same label as in format
    3rmi.</i></p>
  </td>
  <td><i>suggested format for inline linked <code>invoke-static</code>
    and <code>invoke-virtual</code> instructions of format 35c</i>
  </td>
</tr>
<tr>
  <td rowspan="3">AA|<i>op</i> BBBB CCCC</td>
  <td>3rc</td>
  <td><i><code>op</code></i> {vCCCC .. vNNNN}, meth@BBBB<br/>
    <i><code>op</code></i> {vCCCC .. vNNNN}, type@BBBB<br/>
    <p><i>where <code>NNNN = CCCC+AA-1</code>, that is <code>A</code>
    determines the count <code>0..255</code>, and <code>C</code>
    determines the first register</i></p>
  </td>
  <td>&nbsp;</td>
</tr>
<tr>
  <td>3rms</td>
  <td><i><code>op</code></i> {vCCCC .. vNNNN}, vtaboff@BBBB<br/>
    <p><i>where <code>NNNN = CCCC+AA-1</code>, that is <code>A</code>
    determines the count <code>0..255</code>, and <code>C</code>
    determines the first register</i></p>
  </td>
  <td><i>suggested format for statically linked <code>invoke-virtual</code>
    and <code>invoke-super</code> instructions of format <code>3rc</code></i>
  </td>
</tr>
<tr>
  <td>3rmi</td>
  <td><i><code>op</code></i> {vCCCC .. vNNNN}, inline@BBBB<br/>
    <p><i>where <code>NNNN = CCCC+AA-1</code>, that is <code>A</code>
    determines the count <code>0..255</code>, and <code>C</code>
    determines the first register</i></p>
  </td>
  <td><i>suggested format for inline linked <code>invoke-static</code>
    and <code>invoke-virtual</code> instructions of format 3rc</i>
  </td>
</tr>
<tr>
  <td>AA|<i>op</i> BBBB<sub>lo</sub> BBBB BBBB BBBB<sub>hi</sub></td>
  <td>51l</td>
  <td><i><code>op</code></i> vAA, #+BBBBBBBBBBBBBBBB</td>
  <td>const-wide</td>
</tr>
</tbody>
</table>