auto import from //depot/cupcake/@135843
diff --git a/docs/dalvik-bytecode.css b/docs/dalvik-bytecode.css
new file mode 100644
index 0000000..e4a5caa
--- /dev/null
+++ b/docs/dalvik-bytecode.css
@@ -0,0 +1,165 @@
+h1 {
+ font-family: serif;
+ color: #222266;
+}
+
+h2 {
+ font-family: serif;
+ border-top-style: solid;
+ border-top-width: 2px;
+ border-color: #ccccdd;
+ padding-top: 12px;
+ margin-top: 48px;
+ margin-bottom: 2px;
+ color: #222266;
+}
+
+@media print {
+ table {
+ font-size: 8pt;
+ }
+}
+
+@media screen {
+ table {
+ font-size: 10pt;
+ }
+}
+
+
+/* general for all tables */
+
+table {
+ border-collapse: collapse;
+ margin-top: 12px;
+}
+
+table th {
+ font-family: sans-serif;
+ background: #aabbff;
+}
+
+table td {
+ font-family: sans-serif;
+ border-top-style: solid;
+ border-bottom-style: solid;
+ border-width: 1px;
+ border-color: #aaaaff;
+ padding-top: 4px;
+ padding-bottom: 4px;
+ padding-left: 4px;
+ padding-right: 6px;
+ background: #eeeeff;
+}
+
+table td p {
+ margin-top: 4pt;
+ margin-bottom: 0pt;
+}
+
+
+
+/* opcodes table */
+
+table.instruc {
+ margin-top: 24px;
+ margin-bottom: 24px;
+ margin-left: 48px;
+ margin-right: 48px;
+}
+
+table.instruc td {
+ font-family: sans-serif;
+ border-top-style: solid;
+ border-bottom-style: solid;
+ border-width: 1px;
+ padding-top: 4px;
+ padding-bottom: 4px;
+ padding-left: 2px;
+ padding-right: 2px;
+}
+
+table.instruc td:first-child {
+ font-family: monospace;
+ font-size: 90%;
+ vertical-align: top;
+ width: 12%;
+}
+
+table.instruc td:first-child + td {
+ font-family: monospace;
+ font-size: 90%;
+ vertical-align: top;
+ width: 23%;
+}
+
+table.instruc td:first-child + td i {
+ font-family: sans-serif;
+ font-size: 90%;
+}
+
+table.instruc td:first-child + td + td {
+ vertical-align: top;
+ width: 28%;
+}
+
+table.instruc td:first-child + td + td + td {
+ vertical-align: top;
+ width: 37%;
+}
+
+
+/* supplemental opcode format table */
+
+table.supplement {
+ margin-top: 24px;
+ margin-bottom: 24px;
+ margin-left: 48px;
+ margin-right: 48px;
+}
+
+table.supplement td:first-child {
+ font-family: monospace;
+ vertical-align: top;
+ width: 20%;
+}
+
+table.supplement td:first-child + td {
+ font-family: monospace;
+ vertical-align: top;
+ width: 20%;
+}
+
+table.supplement td:first-child + td + td {
+ font-family: sans-serif;
+ vertical-align: top;
+ width: 60%;
+}
+
+
+/* math details table */
+
+table.math {
+ margin-top: 24px;
+ margin-bottom: 24px;
+ margin-left: 48px;
+ margin-right: 48px;
+}
+
+table.math td:first-child {
+ font-family: monospace;
+ vertical-align: top;
+ width: 10%;
+}
+
+table.math td:first-child + td {
+ font-family: monospace;
+ vertical-align: top;
+ width: 30%;
+}
+
+table.math td:first-child + td + td {
+ font-family: sans-serif;
+ vertical-align: top;
+ width: 60%;
+}
diff --git a/docs/dalvik-bytecode.html b/docs/dalvik-bytecode.html
new file mode 100644
index 0000000..4945d60
--- /dev/null
+++ b/docs/dalvik-bytecode.html
@@ -0,0 +1,1500 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>Bytecode for the Dalvik VM</title>
+<link rel=stylesheet href="dalvik-bytecode.css">
+</head>
+
+<body>
+
+<h1>Bytecode for the Dalvik VM</h1>
+<p>Copyright © 2007 The Android Open Source Project
+
+<h2>General Design</h2>
+
+<ul>
+<li>The machine model and calling conventions are meant to approximately
+ imitate common real architectures and C-style calling conventions:
+ <ul>
+ <li>The VM is register-based, and frames are fixed in size upon creation.
+ Each frame consists of a particular number of registers (specified by
+ the method) as well as any adjunct data needed to execute the method,
+ such as (but not limited to) the program counter and a reference to the
+ <code>.dex</code> file that contains the method.
+ </li>
+ <li>Registers are 32 bits wide. Adjacent register pairs are used for 64-bit
+ values.
+ </li>
+ <li>In terms of bitwise representation, <code>(Object) null == (int)
+ 0</code>.
+ </li>
+ <li>The <i>N</i> arguments to a method land in the last <i>N</i> registers
+ of the method's invocation frame, in order. Wide arguments consume
+ two registers. Instance methods are passed a <code>this</code> reference
+ as their first argument.
+ </li>
+ </ul>
+<li>The storage unit in the instruction stream is a 16-bit unsigned quantity.
+ Some bits in some instructions are ignored / must-be-zero.
+</li>
+<li>Instructions aren't gratuitously limited to a particular type. For
+ example, instructions that move 32-bit register values without interpretation
+ don't have to specify whether they are moving ints or floats.
+</li>
+<li>There are separately enumerated and indexed constant pools for
+ references to strings, types, fields, and methods.
+</li>
+<li>Bitwise literal data is represented in-line in the instruction stream.</li>
+<li>Because, in practice, it is uncommon for a method to need more than
+ 16 registers, and because needing more than eight registers <i>is</i>
+ reasonably common, many instructions may only address the first 16
+ registers. When reasonably possible, instructions allow references to
+ up to the first 256 registers. In cases where an instruction variant isn't
+ available to address a desired register, it is expected that the register
+ contents get moved from the original register to a low register (before the
+ operation) and/or moved from a low result register to a high register
+ (after the operation).
+</li>
+<li>There are several "pseudo-instructions" that are used to hold
+ variable-length data referred to by regular instructions (for example,
+ <code>fill-array-data</code>). Such instructions must never be
+ encountered during the normal flow of execution. In addition, the
+ instructions must be located on even-numbered bytecode offsets (that is,
+ 4-byte aligned). In order to meet this requirement, dex generation tools
+ should emit an extra <code>nop</code> instruction as a spacer if such an
+ instruction would otherwise be unaligned. Finally, though not required,
+ it is expected that most tools will choose to emit these instructions at
+ the ends of methods, since otherwise it would likely be the case that
+ additional instructions would be needed to branch around them.
+</li>
+<li>When installed on a running system, some instructions may be altered,
+ changing their format, as an install-time static linking optimization.
+ This is to allow for faster execution once linkage is known.
+ See the associated
+ <a href="instruction-formats.html">instruction formats document</a>
+ for the suggested variants. The word "suggested" is used advisedly;
+ it is not mandatory to implement these.
+</li>
+<li>Human-syntax and mnemonics:
+ <ul>
+ <li>Dest-then-source ordering for arguments.</li>
+ <li>Some opcodes have a disambiguating suffix with respect to the type(s)
+ they operate on: Type-general 64-bit opcodes
+ are suffixed with <code>-wide</code>.
+ Type-specific opcodes are suffixed with their type (or a
+ straightforward abbreviation), one of: <code>-boolean</code>
+ <code>-byte</code> <code>-char</code> <code>-short</code>
+ <code>-int</code> <code>-long</code> <code>-float</code>
+ <code>-double</code> <code>-object</code> <code>-string</code>
+ <code>-class</code> <code>-void</code>. Type-general 32-bit opcodes
+ are unmarked.
+ </li>
+ <li>Some opcodes have a disambiguating suffix to distinguish
+ otherwise-identical operations that have different instruction layouts
+ or options. These suffixes are separated from the main names with a slash
+ ("<code>/</code>") and mainly exist at all to make there be a one-to-one
+ mapping with static constants in the code that generates and interprets
+ executables (that is, to reduce ambiguity for humans).
+ </li>
+ </ul>
+</li>
+<li>See the <a href="instruction-formats.html">instruction formats
+ document</a> for more details about the various instruction formats
+ (listed under "Op & Format") as well as details about the opcode
+ syntax.
+</li>
+</ul>
+
+<h2>Summary of Instruction Set</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>00 10x</td>
+ <td>nop</td>
+ <td> </td>
+ <td>Waste cycles.</td>
+</tr>
+<tr>
+ <td>01 12x</td>
+ <td>move vA, vB</td>
+ <td><code>A:</code> destination register (4 bits)<br/>
+ <code>B:</code> source register (4 bits)</td>
+ <td>Move the contents of one non-object register to another.</td>
+</tr>
+<tr>
+ <td>02 22x</td>
+ <td>move/from16 vAA, vBBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> source register (16 bits)</td>
+ <td>Move the contents of one non-object register to another.</td>
+</tr>
+<tr>
+ <td>03 32x</td>
+ <td>move/16 vAAAA, vBBBB</td>
+ <td><code>A:</code> destination register (16 bits)<br/>
+ <code>B:</code> source register (16 bits)</td>
+ <td>Move the contents of one non-object register to another.</td>
+</tr>
+<tr>
+ <td>04 12x</td>
+ <td>move-wide vA, vB</td>
+ <td><code>A:</code> destination register pair (4 bits)<br/>
+ <code>B:</code> source register pair (4 bits)</td>
+ <td>Move the contents of one register-pair to another.
+ <p><b>Note:</b>
+ It is legal to move from <code>v<i>N</i></code> to either
+ <code>v<i>N-1</i></code> or <code>v<i>N+1</i></code>, so implementations
+ must arrange for both halves of a register pair to be read before
+ anything is written.</p>
+ </td>
+</tr>
+<tr>
+ <td>05 22x</td>
+ <td>move-wide/from16 vAA, vBBBB</td>
+ <td><code>A:</code> destination register pair (8 bits)<br/>
+ <code>B:</code> source register pair (16 bits)</td>
+ <td>Move the contents of one register-pair to another.
+ <p><b>Note:</b>
+ Implementation considerations are the same as <code>move-wide</code>,
+ above.</p>
+ </td>
+</tr>
+<tr>
+ <td>06 32x</td>
+ <td>move-wide/16 vAAAA, vBBBB</td>
+ <td><code>A:</code> destination register pair (16 bits)<br/>
+ <code>B:</code> source register pair (16 bits)</td>
+ <td>Move the contents of one register-pair to another.
+ <p><b>Note:</b>
+ Implementation considerations are the same as <code>move-wide</code>,
+ above.</p>
+ </td>
+</tr>
+<tr>
+ <td>07 12x</td>
+ <td>move-object vA, vB</td>
+ <td><code>A:</code> destination register (4 bits)<br/>
+ <code>B:</code> source register (4 bits)</td>
+ <td>Move the contents of one object-bearing register to another.</td>
+</tr>
+<tr>
+ <td>08 22x</td>
+ <td>move-object/from16 vAA, vBBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> source register (16 bits)</td>
+ <td>Move the contents of one object-bearing register to another.</td>
+</tr>
+<tr>
+ <td>09 32x</td>
+ <td>move-object/16 vAAAA, vBBBB</td>
+ <td><code>A:</code> destination register (16 bits)<br/>
+ <code>B:</code> source register (16 bits)</td>
+ <td>Move the contents of one object-bearing register to another.</td>
+</tr>
+<tr>
+ <td>0a 11x</td>
+ <td>move-result vAA</td>
+ <td><code>A:</code> destination register (8 bits)</td>
+ <td>Move the single-word non-object result of the most recent
+ <code>invoke-<i>kind</i></code> into the indicated register.
+ This must be done as the instruction immediately after an
+ <code>invoke-<i>kind</i></code> whose (single-word, non-object) result
+ is not to be ignored; anywhere else is invalid.</td>
+</tr>
+<tr>
+ <td>0b 11x</td>
+ <td>move-result-wide vAA</td>
+ <td><code>A:</code> destination register pair (8 bits)</td>
+ <td>Move the double-word result of the most recent
+ <code>invoke-<i>kind</i></code> into the indicated register pair.
+ This must be done as the instruction immediately after an
+ <code>invoke-<i>kind</i></code> whose (double-word) result
+ is not to be ignored; anywhere else is invalid.</td>
+</tr>
+<tr>
+ <td>0c 11x</td>
+ <td>move-result-object vAA</td>
+ <td><code>A:</code> destination register (8 bits)</td>
+ <td>Move the object result of the most recent <code>invoke-<i>kind</i></code>
+ into the indicated register. This must be done as the instruction
+ immediately after an <code>invoke-<i>kind</i></code> or
+ <code>filled-new-array</code>
+ whose (object) result is not to be ignored; anywhere else is invalid.</td>
+</tr>
+<tr>
+ <td>0d 11x</td>
+ <td>move-exception vAA</td>
+ <td><code>A:</code> destination register (8 bits)</td>
+ <td>Save a just-caught exception into the given register. This should
+ be the first instruction of any exception handler whose caught
+ exception is not to be ignored, and this instruction may <i>only</i>
+ ever occur as the first instruction of an exception handler; anywhere
+ else is invalid.</td>
+</tr>
+<tr>
+ <td>0e 10x</td>
+ <td>return-void</td>
+ <td> </td>
+ <td>Return from a <code>void</code> method.</td>
+</tr>
+<tr>
+ <td>0f 11x</td>
+ <td>return vAA</td>
+ <td><code>A:</code> return value register (8 bits)</td>
+ <td>Return from a single-width (32-bit) non-object value-returning
+ method.
+ </td>
+</tr>
+<tr>
+ <td>10 11x</td>
+ <td>return-wide vAA</td>
+ <td><code>A:</code> return value register-pair (8 bits)</td>
+ <td>Return from a double-width (64-bit) value-returning method.</td>
+</tr>
+<tr>
+ <td>11 11x</td>
+ <td>return-object vAA</td>
+ <td><code>A:</code> return value register (8 bits)</td>
+ <td>Return from an object-returning method.</td>
+</tr>
+<tr>
+ <td>12 11n</td>
+ <td>const/4 vA, #+B</td>
+ <td><code>A:</code> destination register (4 bits)<br/>
+ <code>B:</code> signed int (4 bits)</td>
+ <td>Move the given literal value (sign-extended to 32 bits) into
+ the specified register.</td>
+</tr>
+<tr>
+ <td>13 21s</td>
+ <td>const/16 vAA, #+BBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> signed int (16 bits)</td>
+ <td>Move the given literal value (sign-extended to 32 bits) into
+ the specified register.</td>
+</tr>
+<tr>
+ <td>14 31i</td>
+ <td>const vAA, #+BBBBBBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> arbitrary 32-bit constant</td>
+ <td>Move the given literal value into the specified register.</td>
+</tr>
+<tr>
+ <td>15 21h</td>
+ <td>const/high16 vAA, #+BBBB0000</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> signed int (16 bits)</td>
+ <td>Move the given literal value (right-zero-extended to 32 bits) into
+ the specified register.</td>
+</tr>
+<tr>
+ <td>16 21s</td>
+ <td>const-wide/16 vAA, #+BBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> signed int (16 bits)</td>
+ <td>Move the given literal value (sign-extended to 64 bits) into
+ the specified register-pair.</td>
+</tr>
+<tr>
+ <td>17 31i</td>
+ <td>const-wide/32 vAA, #+BBBBBBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> signed int (32 bits)</td>
+ <td>Move the given literal value (sign-extended to 64 bits) into
+ the specified register-pair.</td>
+</tr>
+<tr>
+ <td>18 51l</td>
+ <td>const-wide vAA, #+BBBBBBBBBBBBBBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> arbitrary double-width (64-bit) constant</td>
+ <td>Move the given literal value into
+ the specified register-pair.</td>
+</tr>
+<tr>
+ <td>19 21h</td>
+ <td>const-wide/high16 vAA, #+BBBB000000000000</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> signed int (16 bits)</td>
+ <td>Move the given literal value (right-zero-extended to 64 bits) into
+ the specified register-pair.</td>
+</tr>
+<tr>
+ <td>1a 21c</td>
+ <td>const-string vAA, string@BBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> string index</td>
+ <td>Move a reference to the string specified by the given index into the
+ specified register.</td>
+</tr>
+<tr>
+ <td>1b 31c</td>
+ <td>const-string/jumbo vAA, string@BBBBBBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> string index</td>
+ <td>Move a reference to the string specified by the given index into the
+ specified register.</td>
+</tr>
+<tr>
+ <td>1c 21c</td>
+ <td>const-class vAA, type@BBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> type index</td>
+ <td>Move a reference to the class specified by the given index into the
+ specified register. In the case where the indicated type is primitive,
+ this will store a reference to the primitive type's degenerate
+ class.</td>
+</tr>
+<tr>
+ <td>1d 11x</td>
+ <td>monitor-enter vAA</td>
+ <td><code>A:</code> reference-bearing register (8 bits)</td>
+ <td>Acquire the monitor for the indicated object.</td>
+</tr>
+<tr>
+ <td>1e 11x</td>
+ <td>monitor-exit vAA</td>
+ <td><code>A:</code> reference-bearing register (8 bits)</td>
+ <td>Release the monitor for the indicated object.
+ <p><b>Note:</b>
+ If this instruction needs to throw an exception, it must do
+ so as if the pc has already advanced past the instruction.
+ It may be useful to think of this as the instruction successfully
+ executing (in a sense), and the exception getting thrown <i>after</i>
+ the instruction but <i>before</i> the next one gets a chance to
+ run. This definition makes it possible for a method to use
+ a monitor cleanup catch-all (e.g., <code>finally</code>) block as
+ the monitor cleanup for that block itself, as a way to handle the
+ arbitrary exceptions that might get thrown due to the historical
+ implementation of <code>Thread.stop()</code>, while still managing
+ to have proper monitor hygiene.</p>
+ </td>
+</tr>
+<tr>
+ <td>1f 21c</td>
+ <td>check-cast vAA, type@BBBB</td>
+ <td><code>A:</code> reference-bearing register (8 bits)<br/>
+ <code>B:</code> type index (16 bits)</td>
+ <td>Throw a <code>ClassCastException</code> if the reference in the
+ given register cannot be cast to the indicated type.
+ <p><b>Note:</b> Since <code>A</code> must always be a reference
+ (and not a primitive value), this will necessarily fail at runtime
+ (that is, it will throw an exception) if <code>B</code> refers to a
+ primitive type.</p>
+ </td>
+</tr>
+<tr>
+ <td>20 22c</td>
+ <td>instance-of vA, vB, type@CCCC</td>
+ <td><code>A:</code> destination register (4 bits)<br/>
+ <code>B:</code> reference-bearing register (4 bits)<br/>
+ <code>C:</code> type index (16 bits)</td>
+ <td>Store in the given destination register <code>1</code>
+ if the indicated reference is an instance of the given type,
+ or <code>0</code> if not.
+ <p><b>Note:</b> Since <code>B</code> must always be a reference
+ (and not a primitive value), this will always result
+ in <code>0</code> being stored if <code>C</code> refers to a primitive
+ type.</td>
+</tr>
+<tr>
+ <td>21 12x</td>
+ <td>array-length vA, vB</td>
+ <td><code>A:</code> destination register (4 bits)<br/>
+ <code>B:</code> array reference-bearing register (4 bits)</td>
+ <td>Store in the given destination register the length of the indicated
+ array, in entries</td>
+</tr>
+<tr>
+ <td>22 21c</td>
+ <td>new-instance vAA, type@BBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> type index</td>
+ <td>Construct a new instance of the indicated type, storing a
+ reference to it in the destination. The type must refer to a
+ non-array class.</td>
+</tr>
+<tr>
+ <td>23 22c</td>
+ <td>new-array vA, vB, type@CCCC</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> size register<br/>
+ <code>C:</code> type index</td>
+ <td>Construct a new array of the indicated type and size. The type
+ must be an array type.</td>
+</tr>
+<tr>
+ <td>24 35c</td>
+ <td>filled-new-array {vD, vE, vF, vG, vA}, type@CCCC</td>
+ <td><code>B:</code> array size and argument word count (4 bits)<br/>
+ <code>C:</code> type index (16 bits)<br/>
+ <code>D..G, A:</code> argument registers (4 bits each)</td>
+ <td>Construct an array of the given type and size, filling it with the
+ supplied contents. The type must be an array type. The array's
+ contents must be single-word (that is,
+ no arrays of <code>long</code> or <code>double</code>). The constructed
+ instance is stored as a "result" in the same way that the method invocation
+ instructions store their results, so the constructed instance must
+ be moved to a register with a subsequent
+ <code>move-result-object</code> instruction (if it is to be used).</td>
+</tr>
+<tr>
+ <td>25 3rc</td>
+ <td>filled-new-array/range {vCCCC .. vNNNN}, type@BBBB</td>
+ <td><code>A:</code> array size and argument word count (8 bits)<br/>
+ <code>B:</code> type index (16 bits)<br/>
+ <code>C:</code> first argument register (16 bits)<br/>
+ <code>N = A + C - 1</code></td>
+ <td>Construct an array of the given type and size, filling it with
+ the supplied contents. Clarifications and restrictions are the same
+ as <code>filled-new-array</code>, described above.</td>
+</tr>
+<tr>
+ <td>26 31t</td>
+ <td>fill-array-data vAA, +BBBBBBBB <i>(with supplemental data as specified
+ below in "<code>fill-array-data</code> Format")</i></td>
+ <td><code>A:</code> array reference (8 bits)<br/>
+ <code>B:</code> signed "branch" offset to table data pseudo-instruction
+ (32 bits)
+ </td>
+ <td>Fill the given array with the indicated data. The reference must be
+ to an array of primitives, and the data table must match it in type and
+ must contain no more elements than will fit in the array. That is,
+ the array may be larger than the table, and if so, only the initial
+ elements of the array are set, leaving the remainder alone.
+ </td>
+</tr>
+<tr>
+ <td>27 11x</td>
+ <td>throw vAA</td>
+ <td><code>A:</code> exception-bearing register (8 bits)<br/></td>
+ <td>Throw the indicated exception.</td>
+</tr>
+<tr>
+ <td>28 10t</td>
+ <td>goto +AA</td>
+ <td><code>A:</code> signed branch offset (8 bits)</td>
+ <td>Unconditionally jump to the indicated instruction.
+ <p><b>Note:</b>
+ The branch offset may not be <code>0</code>. (A spin
+ loop may be legally constructed either with <code>goto/32</code> or
+ by including a <code>nop</code> as a target before the branch.)</p>
+ </td>
+</tr>
+<tr>
+ <td>29 20t</td>
+ <td>goto/16 +AAAA</td>
+ <td><code>A:</code> signed branch offset (16 bits)<br/></td>
+ <td>Unconditionally jump to the indicated instruction.
+ <p><b>Note:</b>
+ The branch offset may not be <code>0</code>. (A spin
+ loop may be legally constructed either with <code>goto/32</code> or
+ by including a <code>nop</code> as a target before the branch.)</p>
+ </td>
+</tr>
+<tr>
+ <td>2a 30t</td>
+ <td>goto/32 +AAAAAAAA</td>
+ <td><code>A:</code> signed branch offset (32 bits)<br/></td>
+ <td>Unconditionally jump to the indicated instruction.</td>
+</tr>
+<tr>
+ <td>2b 31t</td>
+ <td>packed-switch vAA, +BBBBBBBB <i>(with supplemental data as
+ specified below in "<code>packed-switch</code> Format")</i></td>
+ <td><code>A:</code> register to test<br/>
+ <code>B:</code> signed "branch" offset to table data pseudo-instruction
+ (32 bits)
+ </td>
+ <td>Jump to a new instruction based on the value in the
+ given register, using a table of offsets corresponding to each value
+ in a particular integral range, or fall through to the next
+ instruction if there is no match.
+ </td>
+</tr>
+<tr>
+ <td>2c 31t</td>
+ <td>sparse-switch vAA, +BBBBBBBB <i>(with supplemental data as
+ specified below in "<code>sparse-switch</code> Format")</i></td>
+ <td><code>A:</code> register to test<br/>
+ <code>B:</code> signed "branch" offset to table data pseudo-instruction
+ (32 bits)
+ </td>
+ <td>Jump to a new instruction based on the value in the given
+ register, using an ordered table of value-offset pairs, or fall
+ through to the next instruction if there is no match.
+ </td>
+</tr>
+<tr>
+ <td>2d..31 23x</td>
+ <td>cmp<i>kind</i> vAA, vBB, vCC<br/>
+ 2d: cmpl-float <i>(lt bias)</i><br/>
+ 2e: cmpg-float <i>(gt bias)</i><br/>
+ 2f: cmpl-double <i>(lt bias)</i><br/>
+ 30: cmpg-double <i>(gt bias)</i><br/>
+ 31: cmp-long
+ </td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> first source register or pair<br/>
+ <code>C:</code> second source register or pair</td>
+ <td>Perform the indicated floating point or <code>long</code> comparison,
+ storing <code>0</code> if the two arguments are equal, <code>1</code>
+ if the second argument is larger, or <code>-1</code> if the first
+ argument is larger. The "bias" listed for the floating point operations
+ indicates how <code>NaN</code> comparisons are treated: "Gt bias"
+ instructions return <code>1</code> for <code>NaN</code> comparisons,
+ and "lt bias" instructions return
+ <code>-1</code>.
+ <p>For example, to check to see if floating point
+ <code>a < b</code>, then it is advisable to use
+ <code>cmpg-float</code>; a result of <code>-1</code> indicates that
+ the test was true, and the other values indicate it was false either
+ due to a valid comparison or because one or the other values was
+ <code>NaN</code>.</p>
+ </td>
+</tr>
+<tr>
+ <td>32..37 22t</td>
+ <td>if-<i>test</i> vA, vB, +CCCC<br/>
+ 32: if-eq<br/>
+ 33: if-ne<br/>
+ 34: if-lt<br/>
+ 35: if-ge<br/>
+ 36: if-gt<br/>
+ 37: if-le<br/>
+ </td>
+ <td><code>A:</code> first register to test (4 bits)<br/>
+ <code>B:</code> second register to test (4 bits)<br/>
+ <code>C:</code> signed branch offset (16 bits)</td>
+ <td>Branch to the given destination if the given two registers' values
+ compare as specified.
+ <p><b>Note:</b>
+ The branch offset may not be <code>0</code>. (A spin
+ loop may be legally constructed either by branching around a
+ backward <code>goto</code> or by including a <code>nop</code> as
+ a target before the branch.)</p>
+ </td>
+</tr>
+<tr>
+ <td>38..3d 21t</td>
+ <td>if-<i>test</i>z vAA, +BBBB<br/>
+ 38: if-eqz<br/>
+ 39: if-nez<br/>
+ 3a: if-ltz<br/>
+ 3b: if-gez<br/>
+ 3c: if-gtz<br/>
+ 3d: if-lez<br/>
+ </td>
+ <td><code>A:</code> register to test (8 bits)<br/>
+ <code>B:</code> signed branch offset (16 bits)</td>
+ <td>Branch to the given destination if the given register's value compares
+ with 0 as specified.
+ <p><b>Note:</b>
+ The branch offset may not be <code>0</code>. (A spin
+ loop may be legally constructed either by branching around a
+ backward <code>goto</code> or by including a <code>nop</code> as
+ a target before the branch.)</p>
+ </td>
+</tr>
+<tr>
+ <td>3e..43 10x</td>
+ <td><i>(unused)</i></td>
+ <td> </td>
+ <td><i>(unused)</i></td>
+</tr>
+<tr>
+ <td>44..51 23x</td>
+ <td><i>arrayop</i> vAA, vBB, vCC<br/>
+ 44: aget<br/>
+ 45: aget-wide<br/>
+ 46: aget-object<br/>
+ 47: aget-boolean<br/>
+ 48: aget-byte<br/>
+ 49: aget-char<br/>
+ 4a: aget-short<br/>
+ 4b: aput<br/>
+ 4c: aput-wide<br/>
+ 4d: aput-object<br/>
+ 4e: aput-boolean<br/>
+ 4f: aput-byte<br/>
+ 50: aput-char<br/>
+ 51: aput-short
+ </td>
+ <td><code>A:</code> value register or pair; may be source or dest
+ (8 bits)<br/>
+ <code>B:</code> array register (8 bits)<br/>
+ <code>C:</code> index register (8 bits)</td>
+ <td>Perform the identified array operation at the identified index of
+ the given array, loading or storing into the value register.</td>
+</tr>
+<tr>
+ <td>52..5f 22c</td>
+ <td>i<i>instanceop</i> vA, vB, field@CCCC<br/>
+ 52: iget<br/>
+ 53: iget-wide<br/>
+ 54: iget-object<br/>
+ 55: iget-boolean<br/>
+ 56: iget-byte<br/>
+ 57: iget-char<br/>
+ 58: iget-short<br/>
+ 59: iput<br/>
+ 5a: iput-wide<br/>
+ 5b: iput-object<br/>
+ 5c: iput-boolean<br/>
+ 5d: iput-byte<br/>
+ 5e: iput-char<br/>
+ 5f: iput-short
+ </td>
+ <td><code>A:</code> value register or pair; may be source or dest
+ (4 bits)<br/>
+ <code>B:</code> object register (4 bits)<br/>
+ <code>C:</code> instance field reference index (16 bits)</td>
+ <td>Perform the identified object instance field operation with
+ the identified field, loading or storing into the value register.
+ <p><b>Note:</b> These opcodes are reasonable candidates for static linking,
+ altering the field argument to be a more direct offset.</p>
+ </td>
+</tr>
+<tr>
+ <td>60..6d 21c</td>
+ <td>s<i>staticop</i> vAA, field@BBBB<br/>
+ 60: sget<br/>
+ 61: sget-wide<br/>
+ 62: sget-object<br/>
+ 63: sget-boolean<br/>
+ 64: sget-byte<br/>
+ 65: sget-char<br/>
+ 66: sget-short<br/>
+ 67: sput<br/>
+ 68: sput-wide<br/>
+ 69: sput-object<br/>
+ 6a: sput-boolean<br/>
+ 6b: sput-byte<br/>
+ 6c: sput-char<br/>
+ 6d: sput-short
+ </td>
+ <td><code>A:</code> value register or pair; may be source or dest
+ (8 bits)<br/>
+ <code>B:</code> static field reference index (16 bits)</td>
+ <td>Perform the identified object static field operation with the identified
+ static field, loading or storing into the value register.
+ <p><b>Note:</b> These opcodes are reasonable candidates for static linking,
+ altering the field argument to be a more direct offset.</p>
+ </td>
+</tr>
+<tr>
+ <td>6e..72 35c</td>
+ <td>invoke-<i>kind</i> {vD, vE, vF, vG, vA}, meth@CCCC<br/>
+ 6e: invoke-virtual<br/>
+ 6f: invoke-super<br/>
+ 70: invoke-direct<br/>
+ 71: invoke-static<br/>
+ 72: invoke-interface
+ </td>
+ <td><code>B:</code> argument word count (4 bits)<br/>
+ <code>C:</code> method index (16 bits)<br/>
+ <code>D..G, A:</code> argument registers (4 bits each)</td>
+ <td>Call the indicated method. The result (if any) may be stored
+ with an appropriate <code>move-result*</code> variant as the immediately
+ subsequent instruction.
+ <p><code>invoke-virtual</code> is used to invoke a normal virtual
+ method (a method that is not <code>static</code> or <code>final</code>,
+ and is not a constructor).</p>
+ <p><code>invoke-super</code> is used to invoke the closest superclass's
+ virtual method (as opposed to the one with the same <code>method_id</code>
+ in the calling class).</p>
+ <p><code>invoke-direct</code> is used to invoke a non-<code>static</code>
+ direct method (that is, an instance method that is by its nature
+ non-overridable, namely either a <code>private</code> instance method
+ or a constructor).</p>
+ <p><code>invoke-static</code> is used to invoke a <code>static</code>
+ method (which is always considered a direct method).</p>
+ <p><code>invoke-interface</code> is used to invoke an
+ <code>interface</code> method, that is, on an object whose concrete
+ class isn't known, using a <code>method_id</code> that refers to
+ an <code>interface</code>.</p>
+ <p><b>Note:</b> These opcodes are reasonable candidates for static linking,
+ altering the method argument to be a more direct offset
+ (or pair thereof).</p>
+ </td>
+</tr>
+<tr>
+ <td>73 10x</td>
+ <td><i>(unused)</i></td>
+ <td> </td>
+ <td><i>(unused)</i></td>
+</tr>
+<tr>
+ <td>74..78 3rc</td>
+ <td>invoke-<i>kind</i>/range {vCCCC .. vNNNN}, meth@BBBB<br/>
+ 74: invoke-virtual/range<br/>
+ 75: invoke-super/range<br/>
+ 76: invoke-direct/range<br/>
+ 77: invoke-static/range<br/>
+ 78: invoke-interface/range
+ </td>
+ <td><code>A:</code> argument word count (8 bits)<br/>
+ <code>B:</code> method index (16 bits)<br/>
+ <code>C:</code> first argument register (16 bits)<br/>
+ <code>N = A + C - 1</code></td>
+ <td>Call the indicated method. See first <code>invoke-<i>kind</i></code>
+ description above for details, caveats, and suggestions.
+ </td>
+</tr>
+<tr>
+ <td>79..7a 10x</td>
+ <td><i>(unused)</i></td>
+ <td> </td>
+ <td><i>(unused)</i></td>
+</tr>
+<tr>
+ <td>7b..8f 12x</td>
+ <td><i>unop</i> vA, vB<br/>
+ 7b: neg-int<br/>
+ 7c: not-int<br/>
+ 7d: neg-long<br/>
+ 7e: not-long<br/>
+ 7f: neg-float<br/>
+ 80: neg-double<br/>
+ 81: int-to-long<br/>
+ 82: int-to-float<br/>
+ 83: int-to-double<br/>
+ 84: long-to-int<br/>
+ 85: long-to-float<br/>
+ 86: long-to-double<br/>
+ 87: float-to-int<br/>
+ 88: float-to-long<br/>
+ 89: float-to-double<br/>
+ 8a: double-to-int<br/>
+ 8b: double-to-long<br/>
+ 8c: double-to-float<br/>
+ 8d: int-to-byte<br/>
+ 8e: int-to-char<br/>
+ 8f: int-to-short
+ </td>
+ <td><code>A:</code> destination register or pair (4 bits)<br/>
+ <code>B:</code> source register or pair (4 bits)</td>
+ <td>Perform the identified unary operation on the source register,
+ storing the result in the destination register.</td>
+</tr>
+
+<tr>
+ <td>90..af 23x</td>
+ <td><i>binop</i> vAA, vBB, vCC<br/>
+ 90: add-int<br/>
+ 91: sub-int<br/>
+ 92: mul-int<br/>
+ 93: div-int<br/>
+ 94: rem-int<br/>
+ 95: and-int<br/>
+ 96: or-int<br/>
+ 97: xor-int<br/>
+ 98: shl-int<br/>
+ 99: shr-int<br/>
+ 9a: ushr-int<br/>
+ 9b: add-long<br/>
+ 9c: sub-long<br/>
+ 9d: mul-long<br/>
+ 9e: div-long<br/>
+ 9f: rem-long<br/>
+ a0: and-long<br/>
+ a1: or-long<br/>
+ a2: xor-long<br/>
+ a3: shl-long<br/>
+ a4: shr-long<br/>
+ a5: ushr-long<br/>
+ a6: add-float<br/>
+ a7: sub-float<br/>
+ a8: mul-float<br/>
+ a9: div-float<br/>
+ aa: rem-float<br/>
+ ab: add-double<br/>
+ ac: sub-double<br/>
+ ad: mul-double<br/>
+ ae: div-double<br/>
+ af: rem-double
+ </td>
+ <td><code>A:</code> destination register or pair (8 bits)<br/>
+ <code>B:</code> first source register or pair (8 bits)<br/>
+ <code>C:</code> second source register or pair (8 bits)</td>
+ <td>Perform the identified binary operation on the two source registers,
+ storing the result in the first source register.</td>
+</tr>
+<tr>
+ <td>b0..cf 12x</td>
+ <td><i>binop</i>/2addr vA, vB<br/>
+ b0: add-int/2addr<br/>
+ b1: sub-int/2addr<br/>
+ b2: mul-int/2addr<br/>
+ b3: div-int/2addr<br/>
+ b4: rem-int/2addr<br/>
+ b5: and-int/2addr<br/>
+ b6: or-int/2addr<br/>
+ b7: xor-int/2addr<br/>
+ b8: shl-int/2addr<br/>
+ b9: shr-int/2addr<br/>
+ ba: ushr-int/2addr<br/>
+ bb: add-long/2addr<br/>
+ bc: sub-long/2addr<br/>
+ bd: mul-long/2addr<br/>
+ be: div-long/2addr<br/>
+ bf: rem-long/2addr<br/>
+ c0: and-long/2addr<br/>
+ c1: or-long/2addr<br/>
+ c2: xor-long/2addr<br/>
+ c3: shl-long/2addr<br/>
+ c4: shr-long/2addr<br/>
+ c5: ushr-long/2addr<br/>
+ c6: add-float/2addr<br/>
+ c7: sub-float/2addr<br/>
+ c8: mul-float/2addr<br/>
+ c9: div-float/2addr<br/>
+ ca: rem-float/2addr<br/>
+ cb: add-double/2addr<br/>
+ cc: sub-double/2addr<br/>
+ cd: mul-double/2addr<br/>
+ ce: div-double/2addr<br/>
+ cf: rem-double/2addr
+ </td>
+ <td><code>A:</code> destination and first source register or pair
+ (4 bits)<br/>
+ <code>B:</code> second source register or pair (4 bits)</td>
+ <td>Perform the identified binary operation on the two source registers,
+ storing the result in the first source register.</td>
+</tr>
+<tr>
+ <td>d0..d7 22s</td>
+ <td><i>binop</i>/lit16 vA, vB, #+CCCC<br/>
+ d0: add-int/lit16<br/>
+ d1: rsub-int (reverse subtract)<br/>
+ d2: mul-int/lit16<br/>
+ d3: div-int/lit16<br/>
+ d4: rem-int/lit16<br/>
+ d5: and-int/lit16<br/>
+ d6: or-int/lit16<br/>
+ d7: xor-int/lit16
+ </td>
+ <td><code>A:</code> destination register (4 bits)<br/>
+ <code>B:</code> source register (4 bits)<br/>
+ <code>C:</code> signed int constant (16 bits)</td>
+ <td>Perform the indicated binary op on the indicated register (first
+ argument) and literal value (second argument), storing the result in
+ the destination register.
+ <p><b>Note:</b>
+ <code>rsub-int</code> does not have a suffix since this version is the
+ main opcode of its family. Also, see below for details on its semantics.
+ </p>
+ </td>
+</tr>
+<tr>
+ <td>d8..e2 22b</td>
+ <td><i>binop</i>/lit8 vAA, vBB, #+CC<br/>
+ d8: add-int/lit8<br/>
+ d9: rsub-int/lit8<br/>
+ da: mul-int/lit8<br/>
+ db: div-int/lit8<br/>
+ dc: rem-int/lit8<br/>
+ dd: and-int/lit8<br/>
+ de: or-int/lit8<br/>
+ df: xor-int/lit8<br/>
+ e0: shl-int/lit8<br/>
+ e1: shr-int/lit8<br/>
+ e2: ushr-int/lit8
+ </td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> source register (8 bits)<br/>
+ <code>C:</code> signed int constant (8 bits)</td>
+ <td>Perform the indicated binary op on the indicated register (first
+ argument) and literal value (second argument), storing the result
+ in the destination register.
+ <p><b>Note:</b> See below for details on the semantics of
+ <code>rsub-int</code>.</p>
+ </td>
+</tr>
+<tr>
+ <td>e3..ff 10x</td>
+ <td><i>(unused)</i></td>
+ <td> </td>
+ <td><i>(unused)</i></td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>packed-switch</code> Format</h2>
+
+<table class="supplement">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>ident</td>
+ <td>ushort = 0x0100</td>
+ <td>identifying pseudo-opcode</td>
+</tr>
+<tr>
+ <td>size</td>
+ <td>ushort</td>
+ <td>number of entries in the table</td>
+</tr>
+<tr>
+ <td>first_key</td>
+ <td>int</td>
+ <td>first (and lowest) switch case value</td>
+</tr>
+<tr>
+ <td>targets</td>
+ <td>int[]</td>
+ <td>list of <code>size</code> relative branch targets. The targets are
+ relative to the address of the switch opcode, not of this table.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<p><b>Note:</b> The total number of code units for an instance of this
+table is <code>(size * 2) + 4</code>.</p>
+
+<h2><code>sparse-switch</code> Format</h2>
+
+<table class="supplement">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>ident</td>
+ <td>ushort = 0x0200</td>
+ <td>identifying pseudo-opcode</td>
+</tr>
+<tr>
+ <td>size</td>
+ <td>ushort</td>
+ <td>number of entries in the table</td>
+</tr>
+<tr>
+ <td>keys</td>
+ <td>int[]</td>
+ <td>list of <code>size</code> key values, sorted low-to-high</td>
+</tr>
+<tr>
+ <td>targets</td>
+ <td>int[]</td>
+ <td>list of <code>size</code> relative branch targets, each corresponding
+ to the key value at the same index. The targets are
+ relative to the address of the switch opcode, not of this table.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<p><b>Note:</b> The total number of code units for an instance of this
+table is <code>(size * 4) + 2</code>.</p>
+
+<h2><code>fill-array-data</code> Format</h2>
+
+<table class="supplement">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>ident</td>
+ <td>ushort = 0x0300</td>
+ <td>identifying pseudo-opcode</td>
+</tr>
+<tr>
+ <td>element_width</td>
+ <td>ushort</td>
+ <td>number of bytes in each element</td>
+</tr>
+<tr>
+ <td>size</td>
+ <td>uint</td>
+ <td>number of elements in the table</td>
+</tr>
+<tr>
+ <td>data</td>
+ <td>ubyte[]</td>
+ <td>data values</td>
+</tr>
+</tbody>
+</table>
+
+<p><b>Note:</b> The total number of code units for an instance of this
+table is <code>(size * element_width + 1) / 2 + 4</code>.</p>
+
+
+<h2>Mathematical Operation Details</h2>
+
+<p><b>Note:</b> Floating point operations must follow IEEE 754 rules, using
+round-to-nearest and gradual underflow, except where stated otherwise.</p>
+
+<table class="math">
+<thead>
+<tr>
+ <th>Opcode</th>
+ <th>C Semantics</th>
+ <th>Notes</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>neg-int</td>
+ <td>int32 a;<br/>
+ int32 result = -a;
+ </td>
+ <td>Unary twos-complement.</td>
+</tr>
+<tr>
+ <td>not-int</td>
+ <td>int32 a;<br/>
+ int32 result = ~a;
+ </td>
+ <td>Unary ones-complement.</td>
+</tr>
+<tr>
+ <td>neg-long</td>
+ <td>int64 a;<br/>
+ int64 result = -a;
+ </td>
+ <td>Unary twos-complement.</td>
+</tr>
+<tr>
+ <td>not-long</td>
+ <td>int64 a;<br/>
+ int64 result = ~a;
+ </td>
+ <td>Unary ones-complement.</td>
+</tr>
+<tr>
+ <td>neg-float</td>
+ <td>float a;<br/>
+ float result = -a;
+ </td>
+ <td>Floating point negation.</td>
+</tr>
+<tr>
+ <td>neg-double</td>
+ <td>double a;<br/>
+ double result = -a;
+ </td>
+ <td>Floating point negation.</td>
+</tr>
+<tr>
+ <td>int-to-long</td>
+ <td>int32 a;<br/>
+ int64 result = (int64) a;
+ </td>
+ <td>Sign extension of <code>int32</code> into <code>int64</code>.</td>
+</tr>
+<tr>
+ <td>int-to-float</td>
+ <td>int32 a;<br/>
+ float result = (float) a;
+ </td>
+ <td>Conversion of <code>int32</code> to <code>float</code>, using
+ round-to-nearest. This loses precision for some values.
+ </td>
+</tr>
+<tr>
+ <td>int-to-double</td>
+ <td>int32 a;<br/>
+ double result = (double) a;
+ </td>
+ <td>Conversion of <code>int32</code> to <code>double</code>.</td>
+</tr>
+<tr>
+ <td>long-to-int</td>
+ <td>int64 a;<br/>
+ int32 result = (int32) a;
+ </td>
+ <td>Truncation of <code>int64</code> into <code>int32</code>.</td>
+</tr>
+<tr>
+ <td>long-to-float</td>
+ <td>int64 a;<br/>
+ float result = (float) a;
+ </td>
+ <td>Conversion of <code>int64</code> to <code>float</code>, using
+ round-to-nearest. This loses precision for some values.
+ </td>
+</tr>
+<tr>
+ <td>long-to-double</td>
+ <td>int64 a;<br/>
+ double result = (double) a;
+ </td>
+ <td>Conversion of <code>int64</code> to <code>double</code>, using
+ round-to-nearest. This loses precision for some values.
+ </td>
+</tr>
+<tr>
+ <td>float-to-int</td>
+ <td>float a;<br/>
+ int32 result = (int32) a;
+ </td>
+ <td>Conversion of <code>float</code> to <code>int32</code>, using
+ round-toward-zero. <code>NaN</code> and <code>-0.0</code> (negative zero)
+ convert to the integer <code>0</code>. Infinities and values with
+ too large a magnitude to be represented get converted to either
+ <code>0x7fffffff</code> or <code>-0x80000000</code> depending on sign.
+ </td>
+</tr>
+<tr>
+ <td>float-to-long</td>
+ <td>float a;<br/>
+ int64 result = (int64) a;
+ </td>
+ <td>Conversion of <code>float</code> to <code>int64</code>, using
+ round-toward-zero. The same special case rules as for
+ <code>float-to-int</code> apply here, except that out-of-range values
+ get converted to either <code>0x7fffffffffffffff</code> or
+ <code>-0x8000000000000000</code> depending on sign.
+ </td>
+</tr>
+<tr>
+ <td>float-to-double</td>
+ <td>float a;<br/>
+ double result = (double) a;
+ </td>
+ <td>Conversion of <code>float</code> to <code>double</code>, preserving
+ the value exactly.
+ </td>
+</tr>
+<tr>
+ <td>double-to-int</td>
+ <td>double a;<br/>
+ int32 result = (int32) a;
+ </td>
+ <td>Conversion of <code>double</code> to <code>int32</code>, using
+ round-toward-zero. The same special case rules as for
+ <code>float-to-int</code> apply here.
+ </td>
+</tr>
+<tr>
+ <td>double-to-long</td>
+ <td>double a;<br/>
+ int64 result = (int64) a;
+ </td>
+ <td>Conversion of <code>double</code> to <code>int64</code>, using
+ round-toward-zero. The same special case rules as for
+ <code>float-to-long</code> apply here.
+ </td>
+</tr>
+<tr>
+ <td>double-to-float</td>
+ <td>double a;<br/>
+ float result = (float) a;
+ </td>
+ <td>Conversion of <code>double</code> to <code>float</code>, using
+ round-to-nearest. This loses precision for some values.
+ </td>
+</tr>
+<tr>
+ <td>int-to-byte</td>
+ <td>int32 a;<br/>
+ int32 result = (a << 24) >> 24;
+ </td>
+ <td>Truncation of <code>int32</code> to <code>int8</code>, sign
+ extending the result.
+ </td>
+</tr>
+<tr>
+ <td>int-to-char</td>
+ <td>int32 a;<br/>
+ int32 result = a & 0xffff;
+ </td>
+ <td>Truncation of <code>int32</code> to <code>uint16</code>, without
+ sign extension.
+ </td>
+</tr>
+<tr>
+ <td>int-to-short</td>
+ <td>int32 a;<br/>
+ int32 result = (a << 16) >> 16;
+ </td>
+ <td>Truncation of <code>int32</code> to <code>int16</code>, sign
+ extending the result.
+ </td>
+</tr>
+<tr>
+ <td>add-int</td>
+ <td>int32 a, b;<br/>
+ int32 result = a + b;
+ </td>
+ <td>Twos-complement addition.</td>
+</tr>
+<tr>
+ <td>sub-int</td>
+ <td>int32 a, b;<br/>
+ int32 result = a - b;
+ </td>
+ <td>Twos-complement subtraction.</td>
+</tr>
+<tr>
+ <td>rsub-int</td>
+ <td>int32 a, b;<br/>
+ int32 result = b - a;
+ </td>
+ <td>Twos-complement reverse subtraction.</td>
+</tr>
+<tr>
+ <td>mul-int</td>
+ <td>int32 a, b;<br/>
+ int32 result = a * b;
+ </td>
+ <td>Twos-complement multiplication.</td>
+</tr>
+<tr>
+ <td>div-int</td>
+ <td>int32 a, b;<br/>
+ int32 result = a / b;
+ </td>
+ <td>Twos-complement division, rounded towards zero (that is, truncated to
+ integer). This throws <code>ArithmeticException</code> if
+ <code>b == 0</code>.
+ </td>
+</tr>
+<tr>
+ <td>rem-int</td>
+ <td>int32 a, b;<br/>
+ int32 result = a % b;
+ </td>
+ <td>Twos-complement remainder after division. The sign of the result
+ is the same as that of <code>a</code>, and it is more precisely
+ defined as <code>result == a - (a / b) * b</code>. This throws
+ <code>ArithmeticException</code> if <code>b == 0</code>.
+ </td>
+</tr>
+<tr>
+ <td>and-int</td>
+ <td>int32 a, b;<br/>
+ int32 result = a & b;
+ </td>
+ <td>Bitwise AND.</td>
+</tr>
+<tr>
+ <td>or-int</td>
+ <td>int32 a, b;<br/>
+ int32 result = a | b;
+ </td>
+ <td>Bitwise OR.</td>
+</tr>
+<tr>
+ <td>xor-int</td>
+ <td>int32 a, b;<br/>
+ int32 result = a ^ b;
+ </td>
+ <td>Bitwise XOR.</td>
+</tr>
+<tr>
+ <td>shl-int</td>
+ <td>int32 a, b;<br/>
+ int32 result = a << (b & 0x1f);
+ </td>
+ <td>Bitwise shift left (with masked argument).</td>
+</tr>
+<tr>
+ <td>shr-int</td>
+ <td>int32 a, b;<br/>
+ int32 result = a >> (b & 0x1f);
+ </td>
+ <td>Bitwise signed shift right (with masked argument).</td>
+</tr>
+<tr>
+ <td>ushr-int</td>
+ <td>uint32 a, b;<br/>
+ int32 result = a >> (b & 0x1f);
+ </td>
+ <td>Bitwise unsigned shift right (with masked argument).</td>
+</tr>
+<tr>
+ <td>add-long</td>
+ <td>int64 a, b;<br/>
+ int64 result = a + b;
+ </td>
+ <td>Twos-complement addition.</td>
+</tr>
+<tr>
+ <td>sub-long</td>
+ <td>int64 a, b;<br/>
+ int64 result = a - b;
+ </td>
+ <td>Twos-complement subtraction.</td>
+</tr>
+<tr>
+ <td>mul-long</td>
+ <td>int64 a, b;<br/>
+ int64 result = a * b;
+ </td>
+ <td>Twos-complement multiplication.</td>
+</tr>
+<tr>
+ <td>div-long</td>
+ <td>int64 a, b;<br/>
+ int64 result = a / b;
+ </td>
+ <td>Twos-complement division, rounded towards zero (that is, truncated to
+ integer). This throws <code>ArithmeticException</code> if
+ <code>b == 0</code>.
+ </td>
+</tr>
+<tr>
+ <td>rem-long</td>
+ <td>int64 a, b;<br/>
+ int64 result = a % b;
+ </td>
+ <td>Twos-complement remainder after division. The sign of the result
+ is the same as that of <code>a</code>, and it is more precisely
+ defined as <code>result == a - (a / b) * b</code>. This throws
+ <code>ArithmeticException</code> if <code>b == 0</code>.
+ </td>
+</tr>
+<tr>
+ <td>and-long</td>
+ <td>int64 a, b;<br/>
+ int64 result = a & b;
+ </td>
+ <td>Bitwise AND.</td>
+</tr>
+<tr>
+ <td>or-long</td>
+ <td>int64 a, b;<br/>
+ int64 result = a | b;
+ </td>
+ <td>Bitwise OR.</td>
+</tr>
+<tr>
+ <td>xor-long</td>
+ <td>int64 a, b;<br/>
+ int64 result = a ^ b;
+ </td>
+ <td>Bitwise XOR.</td>
+</tr>
+<tr>
+ <td>shl-long</td>
+ <td>int64 a, b;<br/>
+ int64 result = a << (b & 0x3f);
+ </td>
+ <td>Bitwise shift left (with masked argument).</td>
+</tr>
+<tr>
+ <td>shr-long</td>
+ <td>int64 a, b;<br/>
+ int64 result = a >> (b & 0x3f);
+ </td>
+ <td>Bitwise signed shift right (with masked argument).</td>
+</tr>
+<tr>
+ <td>ushr-long</td>
+ <td>uint64 a, b;<br/>
+ int64 result = a >> (b & 0x3f);
+ </td>
+ <td>Bitwise unsigned shift right (with masked argument).</td>
+</tr>
+<tr>
+ <td>add-float</td>
+ <td>float a, b;<br/>
+ float result = a + b;
+ </td>
+ <td>Floating point addition.</td>
+</tr>
+<tr>
+ <td>sub-float</td>
+ <td>float a, b;<br/>
+ float result = a - b;
+ </td>
+ <td>Floating point subtraction.</td>
+</tr>
+<tr>
+ <td>mul-float</td>
+ <td>float a, b;<br/>
+ float result = a * b;
+ </td>
+ <td>Floating point multiplication.</td>
+</tr>
+<tr>
+ <td>div-float</td>
+ <td>float a, b;<br/>
+ float result = a / b;
+ </td>
+ <td>Floating point division.</td>
+</tr>
+<tr>
+ <td>rem-float</td>
+ <td>float a, b;<br/>
+ float result = a % b;
+ </td>
+ <td>Floating point remainder after division. This function is different
+ than IEEE 754 remainder and is defined as
+ <code>result == a - roundTowardZero(a / b) * b</code>.
+ </td>
+</tr>
+<tr>
+ <td>add-double</td>
+ <td>double a, b;<br/>
+ double result = a + b;
+ </td>
+ <td>Floating point addition.</td>
+</tr>
+<tr>
+ <td>sub-double</td>
+ <td>double a, b;<br/>
+ double result = a - b;
+ </td>
+ <td>Floating point subtraction.</td>
+</tr>
+<tr>
+ <td>mul-double</td>
+ <td>double a, b;<br/>
+ double result = a * b;
+ </td>
+ <td>Floating point multiplication.</td>
+</tr>
+<tr>
+ <td>div-double</td>
+ <td>double a, b;<br/>
+ double result = a / b;
+ </td>
+ <td>Floating point division.</td>
+</tr>
+<tr>
+ <td>rem-double</td>
+ <td>double a, b;<br/>
+ double result = a % b;
+ </td>
+ <td>Floating point remainder after division. This function is different
+ than IEEE 754 remainder and is defined as
+ <code>result == a - roundTowardZero(a / b) * b</code>.
+ </td>
+</tr>
+</tbody>
+</table>
+
+</body>
+</html>
diff --git a/docs/dalvik-constraints.css b/docs/dalvik-constraints.css
new file mode 100644
index 0000000..a315a73
--- /dev/null
+++ b/docs/dalvik-constraints.css
@@ -0,0 +1,59 @@
+h1 {
+ font-family: serif;
+ color: #222266;
+}
+
+h2 {
+ font-family: serif;
+ border-top-style: solid;
+ border-top-width: 2px;
+ border-color: #ccccdd;
+ padding-top: 12px;
+ margin-top: 48px;
+ margin-bottom: 2px;
+ color: #222266;
+}
+
+@media print {
+ table {
+ font-size: 8pt;
+ }
+}
+
+@media screen {
+ table {
+ font-size: 10pt;
+ }
+}
+
+
+/* general for all tables */
+
+table {
+ border-collapse: collapse;
+ margin-top: 24px;
+ margin-bottom: 24px;
+ margin-left: 48px;
+ margin-right: 48px;
+}
+
+table th {
+ font-family: sans-serif;
+ background: #aabbff;
+ text-align: left;
+}
+
+table td {
+ font-family: sans-serif;
+ border-top-style: solid;
+ border-bottom-style: solid;
+ border-width: 1px;
+ border-color: #aaaaff;
+ padding-top: 4px;
+ padding-bottom: 4px;
+ padding-left: 4px;
+ padding-right: 6px;
+ background: #eeeeff;
+ margin-top: 4pt;
+ margin-bottom: 0pt;
+}
diff --git a/docs/dalvik-constraints.html b/docs/dalvik-constraints.html
new file mode 100644
index 0000000..105225a
--- /dev/null
+++ b/docs/dalvik-constraints.html
@@ -0,0 +1,897 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+ <head>
+ <title>Dalvik bytecode constraints</title>
+ <link rel=stylesheet href="dalvik-constraints.css">
+ </head>
+
+ <body>
+
+ <h1>Dalvik bytecode constraints</h1>
+
+<!--
+ <h1>General integrity constraints</h1>
+
+ <table>
+ <tr>
+ <th>
+ Identifier
+ </th>
+
+ <th>
+ Description
+ </th>
+ </tr>
+
+ <tr>
+ <td>
+ A1
+ </td>
+
+ <td>
+ The magic number of the DEX file must be "dex\n035\0".
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ A1
+ </td>
+
+ <td>
+ The checksum must be an Adler-32 checksum of the whole file contents
+ except magic and checksum field.
+ </td>
+ </tr>
+
+
+The signature must be a SHA-1 hash of the whole file contents except magic,
+checksum, and signature.
+
+The file_size must match the actual file size in bytes.
+
+The header_size must have the value 0x70.
+
+The endian_tag must have either the value ENDIAN_CONSTANT or
+REVERSE_ENDIAN_CONSTANT.
+
+For each of the link, string_ids, type_ids, proto_ids, field_ids, method_ids, class_defs
+and data sections, the offset and size fields must be either both zero or both
+non-zero. In the latter case, the offset must be four-byte-aligned.
+
+All offset fields in the header except map_off must be four-byte-aligned.
+
+The map_off field must be either zero or point into the data section. In the
+latter case, the data section must exist.
+
+None of the link, string_ids, type_ids, proto_ids, field_ids, method_ids, class_defs
+and data sections must overlap each other or the header.
+
+If a map exists, then each map entry must have a valid type. Each type may
+appear at most once.
+
+If a map exists, then each map entry must have a nonzero offset and size. The
+offset must point into the corresponding section of the file (i.e. a
+string_id_item must point into the string_ids section) and the explicit or
+implicit size of the item must match the actual contents and size of the
+section.
+
+If a map exists, then the offset of map entry n+1 must be greater or equal to
+the offset of map entry n plus then size of map entry n. This implies
+non-overlapping entries and low-to-high ordering.
+
+The following types of entries must have an offset that is
+four-byte-aligned: string_id_item, type_id_item, proto_id_item, field_id_item,
+method_id_item, class_def_item, type_list, code_item,
+annotations_directory_item.
+
+For each string_id_item, the string_data_off field must contain a valid
+reference into the data section. For the referenced string_data_item, the data
+field must contain a valid MUTF-8 string, and the utf16_size must match the
+decoded length of the string.
+
+For each type_id_item, the desciptor_idx field must contain a valid reference
+into the string_ids list. The referenced string must be a valid type descriptor.
+
+For each proto_id_item, the shorty_idx field must contain a valid reference
+into the string_ids list. The referenced string must be a valid shorty descriptor.
+Also, the return_type_idx field must be a valid index into the type_ids section,
+and the parameters_off field must be either zero or a valid offset pointing
+into the data section. If nonzero, the parameter list must not contain any void
+entries.
+
+For each field_id_item, both the class_idx and type_idx fields must be a valid
+ indices into the
+type_ids list. The entry referenced by class_idx must be a non-array reference type.
+In addition, the name_idx field must be a valid reference into the string_ids
+section, and the contents of the referenced entry must conform to the MemberName
+specification.
+
+For each method_id_item, the class_idx field must be a valid index into the
+type_ids section, and the
+referenced entry must be a non-array reference type. The proto_id field must
+be a valid reference into the proto_ids list. The name_idx field must be a
+valid reference into the string_ids
+section, and the contents of the referenced entry must conform to the MemberName
+specification.
+
+For each class_def_item, ...
+
+For each field_id_item, the class_idx field must be a valid index into the
+type_ids list. The referenced entry must be a non-array reference type.
+
+...
+
+-->
+
+ <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>
+ </tr>
+
+ <tr>
+ <td>
+ A1
+ </td>
+
+ <td>
+ The <code>insns</code> array must not be empty.
+ </td>
+
+ <td>
+ 4.8.1.1
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ A2
+ </td>
+
+ <td>
+ The first opcode in the <code>insns</code> array must have index zero.
+ </td>
+
+ <td>
+ 4.8.1.3
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ A3
+ </td>
+
+ <td>
+ The <code>insns</code> array must only contain valid Dalvik opcodes.
+ </td>
+
+ <td>
+ 4.8.1.4
+ </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>
+
+ <td>
+ 4.8.1.5
+ </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>
+
+ <td>
+ 4.8.1.6
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ A6
+ </td>
+
+ <td>
+ All <code>goto</code> and <code>if-<kind></code> targets must
+ be opcodes within in the same method.
+ </td>
+
+ <td>
+ 4.8.1.7
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ A7
+ </td>
+
+ <td>
+ All targets of a <code>packed-switch</code> instruction must be
+ opcodes within in the same method. The size and the list of targets
+ must be consistent.
+ </td>
+
+ <td>
+ 4.8.1.8
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ A8
+ </td>
+
+ <td>
+ All targets of a <code>sparse-switch</code> instruction must be
+ opcodes within in the same method. The corresponding table must be
+ consistent and sorted low-to-high.
+ </td>
+
+ <td>
+ 4.8.1.9
+ </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>
+
+ <td>
+ 4.8.1.10
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ A10
+ </td>
+
+ <td>
+ The <code>C</code> operand of the <code>iget<kind></code> and
+ <code>iput<kind></code> instructions must be a valid index into
+ the field constant pool. The referenced entry must represent an
+ instance field.
+ </td>
+
+ <td>
+ 4.8.1.12
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ A11
+ </td>
+
+ <td>
+ The <code>C</code> operand of the <code>sget<kind></code> and
+ <code>sput<kind></code> instructions must be a valid index into
+ the field constant pool. The referenced entry must represent a static
+ field.
+ </td>
+
+ <td>
+ 4.8.1.12
+ </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. In all cases, the referenced
+ <code>method_id</code> must belong to a class (not an interface).
+ </td>
+
+ <td>
+ 4.8.1.13
+ </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. In all cases, the referenced
+ <code>method_id</code> must belong to a class (not an interface).
+ </td>
+
+ <td>
+ 4.8.1.13
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ A14
+ </td>
+
+ <td>
+ A method the name of which starts with a '<' must only be invoked
+ implicitly by the VM, not by code originating from a Dex file. The
+ only exception is the instance initializer, which may be invoked by
+ <code>invoke-direct</code>.
+ </td>
+
+ <td>
+ 4.8.1.14
+ </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>
+
+ <td>
+ 4.8.1.15
+ </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>
+
+ <td>
+ 4.8.1.15
+ </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>
+
+ <td>
+ 4.8.1.16
+ </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>
+
+ <td>
+ 4.8.1.16
+ </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>
+
+ <td>
+ 4.8.1.17
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ A20
+ </td>
+
+ <td>
+ The <code>new</code> instruction must not refer to array classes,
+ interfaces, or abstract classes.
+ </td>
+
+ <td>
+ 4.8.1.18
+ </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>
+
+ <td>
+ 4.8.1.20
+ </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>
+
+ <td>
+ 4.8.1.21
+ </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>
+
+ <td>
+ 4.8.1.23
+ </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>
+ </tr>
+
+ <tr>
+ <td>
+ B1
+ </td>
+
+ <td>
+ The number and types of arguments (registers and immediate values)
+ must always match the instruction.
+ </td>
+
+ <td>
+ 4.8.2.1
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ B2
+ </td>
+
+ <td>
+ Register pairs must never be broken up.
+ </td>
+
+ <td>
+ 4.8.2.3
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ B3
+ </td>
+
+ <td>
+ A register (or pair) has to be assigned first before it can be
+ read.
+ </td>
+
+ <td>
+ 4.8.2.4
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ B4
+ </td>
+
+ <td>
+ An <code>invoke-direct</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>
+ </tr>
+
+ <tr>
+ <td>
+ B5
+ </td>
+
+ <td>
+ An instance initializer must only be invoked on an uninitialized
+ instance.
+ </td>
+
+ <td>
+ 4.8.2.8
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ B6
+ </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>
+ </tr>
+
+ <tr>
+ <td>
+ B7
+ </td>
+
+ <td>
+ A register which holds the result of a <code>new-instance</code>code>
+ instruction must not be used if the same
+ <code>new-instance</code>code> instruction is again executed before
+ the instance is initialized.
+ </td>
+
+ <td>
+ 4.8.2.10
+ </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>
+
+ <td>
+ 4.8.2.11
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ B9
+ </td>
+
+ <td>
+ All actual method arguments must be assignment-compatible with their
+ respective formal arguments.
+ </td>
+
+ <td>
+ 4.8.2.12
+ </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>
+
+ <td>
+ 4.8.2.13
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ B11
+ </td>
+
+ <td>
+ A <code>return<kind></code> instruction must match its
+ method's return type.
+ </td>
+
+ <td>
+ 4.8.2.14
+ </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>
+
+ <td>
+ 4.8.2.15
+ </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>
+
+ <td>
+ 4.8.2.16
+ </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>
+
+ <td>
+ 4.8.2.17
+ </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>
+
+ <td>
+ 4.8.2.18
+ </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>
+
+ <td>
+ 4.8.2.19
+ </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>
+
+ <td>
+ 4.8.2.20
+ </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>
+
+ <td>
+ 4.8.2.3, 4.8.2.4
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ B19
+ </td>
+
+ <td>
+ A <code>move-result<kind></code> instruction must be immediately
+ preceded (in the <code>insns</code> array) by an
+ <code><invoke-kind></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>
+
+ <td>
+ -
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ B20
+ </td>
+
+ <td>
+ A <code>move-result<kind></code> instruction must be immediately
+ preceded (in actual control flow) by a matching
+ <code>return-<kind></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>
+
+ <td>
+ -
+ </td>
+ </tr>
+
+ <tr>
+ <td>
+ B21
+ </td>
+
+ <td>
+ A <code>move-exception</code> instruction must only appear as the
+ first instruction in an exception handler.
+ </td>
+
+ <td>
+ -
+ </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>
+
+ <td>
+ -
+ </td>
+ </tr>
+ </table>
+
+ </body>
+</html>
diff --git a/docs/debugger.html b/docs/debugger.html
new file mode 100644
index 0000000..6e23f0d
--- /dev/null
+++ b/docs/debugger.html
@@ -0,0 +1,211 @@
+<html>
+<head>
+<title>Dalvik Debugger Support</title>
+</head>
+
+<body>
+<h1>Dalvik Debugger Support</h1>
+
+<p>
+The Dalvik virtual machine supports source-level debugging with many popular
+development environments. Any tool that allows remote debugging over JDWP
+(the
+<a href="http://java.sun.com/javase/6/docs/technotes/guides/jpda/jdwp-spec.html">
+Java Debug Wire Protocol</a>) is expected work. Supported debuggers
+include jdb, Eclipse, IntelliJ, and JSwat.
+</p><p>
+The VM does not support tools based on JVMTI (Java Virtual
+Machine Tool Interface). This is a relatively intrusive approach that
+relies on bytecode insertion, something the Dalvik VM does not currently
+support.
+</p><p>
+Dalvik's implementation of JDWP also includes hooks for supporting
+DDM (Dalvik Debug Monitor) features, notably as implemented by DDMS
+(Dalvik Debug Monitor Server) and the Eclipse ADT plugin. The protocol
+and VM interaction is described in some detail
+<a href="debugmon.html">here</a>.
+</p><p>
+All of the debugger support in the VM lives in the <code>dalvik/vm/jdwp</code>
+directory, and is almost entirely isolated from the rest of the VM sources.
+<code>dalvik/vm/Debugger.c</code> bridges the gap. The goal in doing so
+was to make it easier to re-use the JDWP code in other projects.
+</p><p>
+
+
+<h2>Implementation</h2>
+
+<p>
+Every VM that has debugging enabled starts a "JDWP" thread. The thread
+typically sits idle until DDMS or a debugger connects. The thread is
+only responsible for handling requests from the debugger; VM-initated
+communication, such as notifying the debugger when the VM has stopped at
+a breakpoint, are sent from the affected thread.
+</p><p>
+When the VM is embedded in the Android framework,
+debugging is enabled in the VM unless the system property
+<code>ro.secure</code> is set to </code>1</code>. On these
+"secure" devices, debugging is only enabled in app processes whose
+manifest contains <code>android:debuggable="true"</code> in the
+<code><application></code> element.
+
+</p><p>
+The VM recognizes the difference between a connection from DDMS and a
+connection from a debugger (either directly or in concert with DDMS).
+A connection from DDMS alone doesn't result in a change in VM behavior,
+but when the VM sees debugger packets it allocates additional data
+structures and may switch to a different implementation of the interpreter.
+</p><p>
+Because Dalvik maps bytecode into memory read-only, some common
+techniques are difficult to implement without allocating additional memory.
+For example, suppose the debugger sets a breakpoint in a method. The
+quick way to handle this is to insert a breakpoint instruction directly
+into the code. When the instruction is reached, the breakpoint handler
+engages. Without this, it's necessary to perform an "is there a breakpoint
+here" scan. Even with some optimizations, the debug-enabled interpreter
+is much slower than the regular interpreter (perhaps 5x).
+</p><p>
+The JDWP protocol is stateless, so the VM handles individual debugger
+requests as they arrive, and posts events to the debugger as they happen.
+</p><p>
+
+
+<h2>Debug Data</h2>
+<p> Source code debug data, which includes mappings of source code to
+bytecode and lists describing which registers are used to hold method
+arguments and local variables, are optionally emitted by the Java compiler.
+When <code>dx</code> converts Java bytecode to Dalvik bytecode, it must
+also convert this debug data.
+</p><p>
+<code>dx</code> must also ensure that it doesn't perform operations
+that confuse the debugger. For example, re-using registers that hold
+method arguments and the "<code>this</code>" pointer is allowed in
+Dalvik bytecode if the values are never used or no longer needed.
+This can be very confusing for the debugger (and the programmer)
+since the values have method scope and aren't expected to disappear. For
+this reason, <code>dx</code> generates sub-optimal code in some situations
+when debugging support is enabled.
+</p><p>
+Some of the debug data is used for other purposes; in particular, having
+filename and line number data is necessary for generating useful exception
+stack traces. This data can be omitted by <code>dx</code> to make the DEX
+file smaller.
+</p><p>
+
+
+<h2>Usage</h2>
+
+<p>
+The Dalvik VM supports many of the same command-line flags that other popular
+desktop VMs do. To start a VM with debugging enabled, you add a command-line
+flag with some basic options. The basic incantation looks something
+like this:
+
+<pre>-Xrunjdwp:transport=dt_socket,address=8000,server=y,suspend=y</pre>
+or
+<pre>-agentlib:jdwp=transport=dt_socket,address=8000,server=y,suspend=y</pre>
+
+</p><p>
+After the initial prefix, options are provided as name=value pairs. The
+options currently supported by the Dalvik VM are:
+<dl>
+ <dt>transport (no default)</dt>
+ <dd>Communication transport mechanism to use. Dalvik supports
+ TCP/IP sockets (<code>dt_socket</code>) and connection over USB
+ through ADB (<code>dt_android_adb</code>).
+ </dd>
+
+ <dt>server (default='n')</dt>
+ <dd>Determines whether the VM acts as a client or a server. When
+ acting as a server, the VM waits for a debugger to connect to it.
+ When acting as a client, the VM attempts to connect to a waiting
+ debugger.
+ </dd>
+
+ <dt>suspend (default='n')</dt>
+ <dd>If set to 'y', the VM will wait for a debugger connection
+ before executing application code. When the debugger connects (or
+ when the VM finishes connecting to the debugger), the VM tells the
+ debugger that it has suspended, and will not proceed until told
+ to resume. If set to 'n', the VM just plows ahead.
+ </dd>
+
+ <dt>address (default="")</dt>
+ <dd>This must be <code>hostname:port</code> when <code>server=n</code>,
+ but can be just <code>port</code> when <code>server=y</code>. This
+ specifies the IP address and port number to connect or listen to.
+ <br>
+ Listening on port 0 has a special meaning: try to
+ listen on port 8000; if that fails, try 8001, 8002, and so on. (This
+ behavior is non-standard and may be removed from a future release.)
+ <br>This option has no meaning for <code>transport=dt_android_adb</code>.
+ </dd>
+
+ <dt>help (no arguments)</dt>
+ <dd>If this is the only option, a brief usage message is displayed.
+ </dd>
+
+ <dt>launch, onthrow, oncaught, timeout</dt>
+ <dd>These options are accepted but ignored.
+ </dd>
+</dl>
+
+</p><p>
+To debug a program on an Android device using DDMS over USB, you could
+use a command like this:
+<pre>% dalvikvm -agentlib:jdwp=transport=dt_android_adb,suspend=y,server=y -cp /data/foo.jar Foo</pre>
+
+This tells the Dalvik VM to run the program with debugging enabled, listening
+for a connection from DDMS, and waiting for a debugger. The program will show
+up with an app name of "?" in the process list, because it wasn't started
+from the Android application framework. From here you would connect your
+debugger to the appropriate DDMS listen port (e.g.
+<code>jdb -attach localhost:8700</code> after selecting it in the app list).
+
+</p><p>
+To debug a program on an Android device using TCP/IP bridged across ADB,
+you would first need to set up forwarding:
+<pre>% adb forward tcp:8000 tcp:8000
+% adb shell dalvikvm -agentlib:jdwp=transport=dt_socket,address=8000,suspend=y,server=y -cp /data/foo.jar Foo</pre>
+and then <code>jdb -attach localhost:8000</code>.
+</p><p>
+(In the above examples, the VM will be suspended when you attach. In jdb,
+type <code>cont</code> to continue.)
+</p><p>
+The DDMS integration makes the <code>dt_android_adb</code> transport much
+more convenient when debugging on an Android device, but when working with
+Dalvik on the desktop it makes sense to use the TCP/IP transport.
+</p><p>
+
+
+<h2>Known Issues and Limitations</h2>
+
+</p><p>
+Most of the optional features JDWP allows are not implemented. These
+include field access watchpoints and better tracking of monitors.
+</p><p>
+Not all JDWP requests are implemented. In particular, anything that
+never gets emitted by the debuggers we've used is not supported and will
+result in error messages being logged. Support will be added when a
+use case is uncovered.
+</p><p>
+
+</p><p>
+The debugger and garbage collector are somewhat loosely
+integrated at present. The VM currently guarantees that any object the
+debugger is aware of will not be garbage collected until after the
+debugger disconnects. This can result in a build-up over time while the
+debugger is connected.
+</p><p>
+The situation is exacerbated by a flaw in the exception processing code,
+which results in nearly all exceptions being added to the "do not discard"
+list, even if the debugger never sees them. Having a debugger attached
+to a program that throws lots of exceptions can result in out-of-memory
+errors. This will be fixed in a future release.
+</p><p>
+
+
+<address>Copyright © 2009 The Android Open Source Project</address>
+</p>
+
+</body>
+</html>
diff --git a/docs/debugmon.html b/docs/debugmon.html
new file mode 100644
index 0000000..cf56ef5
--- /dev/null
+++ b/docs/debugmon.html
@@ -0,0 +1,736 @@
+<HTML>
+
+
+<head>
+ <title>Dalvik VM Debug Monitor</title>
+ <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
+ <link href="http://www.google.com/favicon.ico" type="image/x-icon"
+ rel="shortcut icon">
+ <link href="../android.css" type="text/css" rel="stylesheet">
+ <script language="JavaScript1.2" type="text/javascript">
+function highlight(name) {
+ if (document.getElementById) {
+ tags = [ 'span', 'div', 'tr', 'td' ];
+ for (i in tags) {
+ elements = document.getElementsByTagName(tags[i]);
+ if (elements) {
+ for (j = 0; j < elements.length; j++) {
+ elementName = elements[j].getAttribute("id");
+ if (elementName == name) {
+ elements[j].style.backgroundColor = "#C0F0C0";
+ } else if (elementName && elementName.indexOf("rev") == 0) {
+ elements[j].style.backgroundColor = "#FFFFFF";
+ }
+ }
+ }
+ }
+ }
+}
+ </script>
+</head>
+<body onload="prettyPrint()">
+
+<h1><a name="My_Project_"></a>Dalvik VM<br>Debug Monitor</h1>
+
+<!-- Status is one of: Draft, Current, Needs Update, Obsolete -->
+<p style="text-align:center"><strong>Status:</strong><em>Draft</em>
+<small>(as of March 6, 2007)</small></p>
+<address>
+[authors]
+<address>
+
+<!-- last modified date can be different to the "Status date." It automatically
+updates
+whenever the file is modified. -->
+<i>Modified:</i>
+ <!-- this script automatically sets the modified date,you don't need to modify
+it -->
+ <script type=text/javascript>
+ <!--
+ var lm = new Date(document.lastModified);
+ document.write(lm.toDateString());
+ //-->
+ </script>
+</address>
+
+<p><br>
+<HR>
+
+<h2>Introduction</h2>
+
+<p>It's extremely useful to be able to monitor the live state of the
+VM. For Android, we need to monitor multiple VMs running on a device
+connected through USB or a wireless network connection. This document
+describes a debug monitor server that interacts with multiple VMs, and
+an API that VMs and applications can use to provide information
+to the monitor.
+
+<p>Some things we can monitor with the Dalvik Debug Monitor ("DDM"):
+<ul>
+ <li> Thread states. Track thread creation/exit, busy/idle status.
+ <li> Overall heap status, useful for a heap bitmap display or
+ fragmentation analysis.
+</ul>
+
+<p>It is possible for something other than a VM to act as a DDM client, but
+that is a secondary goal. Examples include "logcat" log extraction
+and system monitors for virtual memory usage and load average.
+
+<p>It's also possible for the DDM server to be run on the device, with
+the information presented through the device UI. However, the initial goal
+is to provide a display tool that takes advantage of desktop tools and
+screen real estate.
+
+<p>This work is necessary because we are unable to use standard JVMTI-based
+tools with Dalvik. JVMTI relies on bytecode insertion, which is not
+currently possible because Dalvik doesn't support Java bytecode.
+
+<p>The DDM server is written in the Java programming language
+for portability. It uses a desktop
+UI toolkit (SWT) for its interface.
+
+
+<h2>Protocol</h2>
+
+<p>To take advantage of existing infrastructure we are piggy-backing the
+DDM protocol on top of JDWP (the Java Debug Wire Protocol, normally spoken
+between a VM and a debugger). To a
+non-DDM client, the DDM server just looks like a debugger.
+
+<p>The JDWP protocol is very close to what we want to use. In particular:
+<ul>
+ <li>It explicitly allows for vendor-defined packets, so there is no
+ need to "bend" the JDWP spec.
+ <li>Events may be posted from the VM at arbitrary points. Such
+ events do not elicit a response from the debugger, meaning the client
+ can post data and immediately resume work without worrying about the
+ eventual response.
+ <li>The basic protocol is stateless and asynchronous. Request packets
+ from the debugger side include a serial number, which the VM includes
+ in the response packet. This allows multiple simultaneous
+ conversations, which means the DDM traffic can be interleaved with
+ debugger traffic.
+</ul>
+
+<p>There are a few issues with using JDWP for our purposes:
+<ul>
+ <li>The VM only expects one connection from a debugger, so you couldn't
+ attach the monitor and a debugger at the same time. This will be
+ worked around by connecting the debugger to the monitor and passing the
+ traffic through. (We're already doing the pass-through with "jdwpspy";
+ requires some management of our request IDs though.) This should
+ be more convenient than the current "guess the port
+ number" system when we're attached to a device.
+ <li>The VM behaves differently when a debugger is attached. It will
+ run more slowly, and any objects passed to the monitor or debugger are
+ immune to GC. We can work around this by not enabling the slow path
+ until non-DDM traffic is observed. We also want to have a "debugger
+ has connected/disconnected" message that allows the VM to release
+ debugger-related resources without dropping the net connection.
+ <li>Non-DDM VMs should not freak out when DDM connects. There are
+ no guarantees here for 3rd-party VMs (e.g. a certain mainstream VM,
+ which crashes instantly), but our older JamVM can be
+ configured to reject the "hello" packet.
+</ul>
+
+
+<h3>Connection Establishment</h3>
+
+<p>There are two basic approaches: have the server contact the VMs, and
+have the VMs contact the server. The former is less "precise" than the
+latter, because you have to scan for the clients, but it has some
+advantages.
+
+<p>There are three interesting scenarios:
+<ol>
+ <li>The DDM server is started, then the USB-attached device is booted
+ or the simulator is launched.
+ <li>The device or simulator is already running when the DDM server
+ is started.
+ <li>The DDM server is running when an already-started device is
+ attached to USB.
+</ol>
+<p>If we have the VMs connect to the DDM server on startup, we only handle
+case #1. If the DDM server scans for VMs when it starts, we only handle
+case #2. Neither handles case #3, which is probably the most important
+of the bunch as the device matures.
+<p>The plan is to have a drop-down menu with two entries,
+"scan workstation" and "scan device".
+The former causes the DDM server to search for VMs on "localhost", the
+latter causes it to search for VMs on the other side of an ADB connection.
+The DDM server will scan for VMs every few seconds, either checking a
+range of known VM ports (e.g. 8000-8040) or interacting with some sort
+of process database on the device. Changing modes causes all existing
+connections to be dropped.
+<p>When the DDM server first starts, it will try to execute "adb usb"
+to ensure that the ADB server is running. (Note it will be necessary
+to launch the DDM server from a shell with "adb" in the path.) If this
+fails, talking to the device will still be possible so long as the ADB
+daemon is already running.
+
+<h4>Connecting a Debugger</h4>
+
+<p>With the DDM server sitting on the JDWP port of all VMs, it will be
+necessary to connect the debugger through the DDM server. Each VM being
+debugged will have a separate port being listened to by the DDM server,
+allowing you to connect a debugger to one or more VMs simultaneously.
+
+<p>In the common case, however, the developer will only want to debug
+a single VM. One port (say 8700) will be listened to by the DDM server,
+and anything connecting to it will be connected to the "current VM"
+(selected in the UI). This should allow developers to focus on a
+single application, which may otherwise shift around in the ordering, without
+having to adjust their IDE settings to a different port every time they
+restart the device.
+
+
+<h3>Packet Format</h3>
+
+<p>Information is sent in chunks. Each chunk starts with:
+<pre>
+u4 type
+u4 length
+</pre>
+and contains a variable amount of type-specific data.
+Unrecognized types cause an empty response from the client and
+are quietly ignored by the server. [Should probably return an error;
+need an "error" chunk type and a handler on the server side.]
+
+<p>The same chunk type may have different meanings when sent in different
+directions. For example, the same type may be used for both a query and
+a response to the query. For sanity the type must always be used in
+related transactions.
+
+<p>This is somewhat redundant with the JDWP framing, which includes a
+4-byte length and a two-byte type code ("command set" and "command"; a
+range of command set values is designated for "vendor-defined commands
+and extensions"). Using the chunk format allows us to remain independent
+of the underlying transport, avoids intrusive integration
+with JDWP client code, and provides a way to send multiple chunks in a
+single transmission unit. [I'm taking the multi-chunk packets into
+account in the design, but do not plan to implement them unless the need
+arises.]
+
+<p>Because we may be sending data over a slow USB link, the chunks may be
+compressed. Compressed chunks are written as a chunk type that
+indicates the compression, followed by the compressed length, followed
+by the original chunk type and the uncompressed length. For zlib's deflate
+algorithm, the chunk type is "ZLIB".
+
+<p>Following the JDWP model, packets sent from the server to the client
+are always acknowledged, but packets sent from client to server never are.
+The JDWP error code field is always set to "no error"; failure responses
+from specific requests must be encoded into the DDM messages.
+
+<p>In what follows "u4" is an unsigned 32-bit value and "u1" is an
+unsigned 8-bit value. Values are written in big-endian order to match
+JDWP.
+
+
+<h3>Initial Handshake</h3>
+
+<p>After the JDWP handshake, the server sends a HELO chunk to the client.
+If the client's JDWP layer rejects it, the server assumes that the client
+is not a DDM-aware VM, and does not send it any further DDM queries.
+<p>On the client side, upon seeing a HELO it can know that a DDM server
+is attached and prepare accordingly. The VM should not assume that a
+debugger is attached until a non-DDM packet arrives.
+
+<h4>Chunk HELO (server --> client)</h4>
+<p>Basic "hello" message.
+<pre>
+u4 DDM server protocol version
+</pre>
+
+
+<h4>Chunk HELO (client --> server, reply only)</h4>
+Information about the client. Must be sent in response to the HELO message.
+<pre>
+u4 DDM client protocol version
+u4 pid
+u4 VM ident string len (in 16-bit units)
+u4 application name len (in 16-bit units)
+var VM ident string (UTF-16)
+var application name (UTF-16)
+</pre>
+
+<p>If the client does not wish to speak to the DDM server, it should respond
+with a JDWP error packet. This is the same behavior you'd get from a VM
+that doesn't support DDM.
+
+
+<h3>Debugger Management</h3>
+<p>VMs usually prepare for debugging when a JDWP connection is established,
+and release debugger-related resources when the connection drops. We want
+to open the JDWP connection early and hold it open after the debugger
+disconnects.
+<p>The VM can tell when a debugger attaches, because it will start seeing
+non-DDM JDWP traffic, but it can't identify the disconnect. For this reason,
+we need to send a packet to the client when the debugger disconnects.
+<p>If the DDM server is talking to a non-DDM-aware client, it will be
+necessary to drop and re-establish the connection when the debugger goes away.
+(This also works with DDM-aware clients; this packet is an optimization.)
+
+<h4>Chunk DBGD (server --> client)</h4>
+<p>Debugger has disconnected. The client responds with a DBGD to acknowledge
+receipt. No data in request, no response required.
+
+
+<h3>VM Info</h3>
+<p>Update the server's info about the client.
+
+<h4>Chunk APNM (client --> server)</h4>
+
+<p>If a VM's application name changes -- possible in our environment because
+of the "pre-initialized" app processes -- it must send up one of these.
+<pre>
+u4 application name len (in 16-bit chars)
+var application name (UTF-16)
+</pre>
+
+<h4>Chunk WAIT (client --> server)</h4>
+
+<p>This tells DDMS that one or more threads are waiting on an external
+event. The simplest use is to tell DDMS that the VM is waiting for a
+debugger to attach.
+<pre>
+u1 reason (0 = wait for debugger)
+</pre>
+If DDMS is attached, the client VM sends this up when waitForDebugger()
+is called. If waitForDebugger() is called before DDMS attaches, the WAIT
+chunk will be sent up at about the same time as the HELO response.
+
+
+<h3>Thread Status</h3>
+
+<p>The client can send updates when their status changes, or periodically
+send thread state info, e.g. 2x per
+second to allow a "blinkenlights" display of thread activity.
+
+<h4>Chunk THEN (server --> client)</h4>
+
+<p>Enable thread creation/death notification.
+<pre>
+u1 boolean (true=enable, false=disable)
+</pre>
+<p>The response is empty. The client generates THCR packets for all
+known threads. (Note the THCR packets may arrive before the THEN
+response.)
+
+<h4>Chunk THCR (client --> server)</h4>
+<p>Thread Creation notification.
+<pre>
+u4 VM-local thread ID (usually a small int)
+u4 thread name len (in 16-bit chars)
+var thread name (UTF-16)
+</pre>
+
+<h4>Chunk THDE (client --> server)</h4>
+<p>Thread Death notification.
+<pre>
+u4 VM-local thread ID
+</pre>
+
+<h4>Chunk THST (server --> client)</h4>
+
+<p>Enable periodic thread activity updates.
+Threads in THCR messages are assumed to be in the "initializing" state. A
+THST message should follow closely on the heels of THCR.
+<pre>
+u4 interval, in msec
+</pre>
+<p>An interval of 0 disables the updates. This is done periodically,
+rather than every time the thread state changes, to reduce the amount
+of data that must be sent for an actively running VM.
+
+<h4>Chunk THST (client --> server)</h4>
+<p>Thread Status, describing the state of one or more threads. This is
+most useful when creation/death notifications are enabled first. The
+overall layout is:
+<pre>
+u4 count
+var thread data
+</pre>
+Then, for every thread:
+<pre>
+u4 VM-local thread ID
+u1 thread state
+u1 suspended
+</pre>
+<p>"thread state" must be one of:
+<ul> <!-- don't use ol, we may need (-1) or sparse -->
+ <li> 1 - running (now executing or ready to do so)
+ <li> 2 - sleeping (in Thread.sleep())
+ <li> 3 - monitor (blocked on a monitor lock)
+ <li> 4 - waiting (in Object.wait())
+ <li> 5 - initializing
+ <li> 6 - starting
+ <li> 7 - native (executing native code)
+ <li> 8 - vmwait (waiting on a VM resource)
+</ul>
+<p>"suspended" will be 0 if the thread is running, 1 if not.
+<p>[Any reason not to make "suspended" be the high bit of "thread state"?
+Do we need to differentiate suspend-by-GC from suspend-by-debugger?]
+<p>[We might be able to send the currently-executing method. This is a
+little risky in a running VM, and increases the size of the messages
+considerably, but might be handy.]
+
+
+<h3>Heap Status</h3>
+
+<p>The client sends what amounts to a color-coded bitmap to the server,
+indicating which stretches of memory are free and which are in use. For
+compactness the bitmap is run-length encoded, and based on multi-byte
+"allocation units" rather than byte counts.
+
+<p>In the future the server will be able to correlate the bitmap with more
+detailed object data, so enough information is provided to associate the
+bitmap data with virtual addresses.
+
+<p>Heaps may be broken into segments within the VM, and due to memory
+constraints it may be desirable to send the bitmap in smaller pieces,
+so the protocol allows the heap data to be sent in several chunks.
+To avoid ambiguity, the client is required
+to send explicit "start" and "end" messages during an update.
+
+<p>All messages include a "heap ID" that can be used to differentiate
+between multiple independent virtual heaps or perhaps a native heap. The
+client is allowed to send information about different heaps simultaneously,
+so all heap-specific information is tagged with a "heap ID".
+
+<h4>Chunk HPIF (server --> client)</h4>
+<p>Request heap info.
+<pre>
+u1 when to send
+</pre>
+<p>The "when" values are:
+<pre>
+0: never
+1: immediately
+2: at the next GC
+3: at every GC
+</pre>
+
+<h4>Chunk HPIF (client --> server, reply only)</h4>
+<p>Heap Info. General information about the heap, suitable for a summary
+display.
+<pre>
+u4 number of heaps
+</pre>
+For each heap:
+<pre>
+u4 heap ID
+u8 timestamp in ms since Unix epoch
+u1 capture reason (same as 'when' value from server)
+u4 max heap size in bytes (-Xmx)
+u4 current heap size in bytes
+u4 current number of bytes allocated
+u4 current number of objects allocated
+</pre>
+<p>[We can get some of this from HPSG, more from HPSO.]
+<p>[Do we need a "heap overhead" stat here, indicating how much goes to
+waste? e.g. (8 bytes per object * number of objects)]
+
+<h4>Chunk HPSG (server --> client)</h4>
+<p>Request transmission of heap segment data.
+<pre>
+u1 when to send
+u1 what to send
+</pre>
+<p>The "when" to send will be zero to disable transmission, 1 to send
+during a GC. Other values are currently undefined. (Could use to pick
+which part of the GC to send it, or cause periodic transmissions.)
+<p>The "what" field is currently 0 for HPSG and 1 for HPSO.
+<p>No reply is expected.
+
+<h4>Chunk NHSG (server --> client)</h4>
+<p>Request transmission of native heap segment data.
+<pre>
+u1 when to send
+u1 what to send
+</pre>
+<p>The "when" to send will be zero to disable transmission, 1 to send
+during a GC. Other values are currently undefined.
+<p>The "what" field is currently ignored.
+<p>No reply is expected.
+
+<h4>Chunk HPST/NHST (client --> server)</h4>
+<p>This is a Heap Start message. It tells the server to discard any
+existing notion of what the client's heap looks like, and prepare for
+new information. HPST indicates a virtual heap dump and must be followed
+by zero or more HPSG/HPSO messages and an HPEN. NHST indicates a native
+heap dump and must be followed by zero or more NHSG messages and an NHEN.
+
+<p>The only data item is:
+<pre>
+u4 heap ID
+</pre>
+
+<h4>Chunk HPEN/NHEN (client --> server)</h4>
+<p>Heap End, indicating that all information about the heap has been sent.
+A HPST will be paired with an HPEN and an NHST will be paired with an NHEN.
+
+<p>The only data item is:
+<pre>
+u4 heap ID
+</pre>
+
+<h4>Chunk HPSG (client --> server)</h4>
+<p>Heap segment data. Each chunk describes all or part of a contiguous
+stretch of heap memory.
+<pre>
+u4 heap ID
+u1 size of allocation unit, in bytes (e.g. 8 bytes)
+u4 virtual address of segment start
+u4 offset of this piece (relative to the virtual address)
+u4 length of piece, in allocation units
+var usage data
+</pre>
+<p>The "usage data" indicates the status of each allocation unit. The data
+is a stream of pairs of bytes, where the first byte indicates the state
+of the allocation unit, and the second byte indicates the number of
+consecutive allocation units with the same state.
+<p>The bits in the "state" byte have the following meaning:
+<pre>
++---------------------------------------+
+| 7 | 6 | 5 | 4 | 3 | 2 | 1 | 0 |
++---------------------------------------+
+| P | U0 | K2 | K1 | K0 | S2 | S1 | S0 |
++---------------------------------------+
+</pre>
+<ul>
+ <li>'S': solidity
+ <ul>
+ <li>0=free
+ <li>1=has hard reference
+ <li>2=has soft reference
+ <li>3=has weak reference
+ <li>4=has phantom reference
+ <li>5=pending finalization
+ <li>6=marked, about to be swept
+ </ul>
+ <li>'K': kind
+ <ul>
+ <li>0=object
+ <li>1=class object
+ <li>2=array of byte/boolean
+ <li>3=array of char/short
+ <li>4=array of Object/int/float
+ <li>5=array of long/double
+ </ul>
+ <li>'P': partial flag (not used for HPSG)
+ <li>'U': unused, must be zero
+</ul>
+
+<p>The use of the various 'S' types depends on when the information is
+sent. The current plan is to send it either immediately after a GC,
+or between the "mark" and "sweep" phases of the GC. For a fancy generational
+collector, we may just want to send it up periodically.
+
+<p>The run-length byte indicates the number of allocation units minus one, so a
+length of 255 means there are 256 consecutive units with this state. In
+some cases, e.g. arrays of bytes, the actual size of the data is rounded
+up the nearest allocation unit.
+<p>For HPSG, the runs do not end at object boundaries. It is not possible
+to tell from this bitmap whether a run contains one or several objects.
+(But see HPSO, below.)
+<p>[If we find that we have many long runs, we can overload the 'P' flag
+or dedicate the 'U' flag to indicate that we have a 16-bit length instead
+of 8-bit. We can also use a variable-width integer scheme for the length,
+encoding 1-128 in one byte, 1-16384 in two bytes, etc.]
+<p>[Alternate plan for 'K': array of byte, array of char, array of Object,
+array of miscellaneous primitive type]
+<p>To parse the data, the server runs through the usage data until either
+(a) the end of the chunk is reached, or (b) all allocation units have been
+accounted for. (If these two things don't happen at the same time, the
+chunk is rejected.)
+<p>Example: suppose a VM has a heap at 0x10000 that is 0x2000 bytes long
+(with an 8-byte allocation unit size, that's 0x0400 units long).
+The client could send one chunk (allocSize=8, virtAddr=0x10000, offset=0,
+length=0x0400) or two (allocSize=8, virtAddr=0x10000, offset=0, length=0x300;
+then allocSize=8, virtAddr=0x10000, offset=0x300, length=0x100).
+<p>The client must encode the entire heap, including all free space at
+the end, or the server will not have an accurate impression of the amount
+of memory in the heap. This refers to the current heap size, not the
+maximum heap size.
+
+<h4>Chunk HPSO (client --> server)</h4>
+<p>This is essentially identical to HPSG, but the runs are terminated at
+object boundaries. If an object is larger than 256 allocation units, the
+"partial" flag is set in all runs except the last.
+<p>The resulting unpacked bitmap is identical, but the object boundary
+information can be used to gain insights into heap layout.
+<p>[Do we want to have a separate message for this? Maybe just include
+a "variant" flag in the HPST packet. Another possible form of output
+would be one that indicates the age, in generations, of each block of
+memory. That would provide a quick visual indication of "permanent vs.
+transient residents", perhaps with a 16-level grey scale.]
+
+<h4>Chunk NHSG (client --> server)</h4>
+<p>Native heap segment data. Each chunk describes all or part of a
+contiguous stretch of native heap memory. The format is the same as
+for HPSG, except that only solidity values 0 (= free) and 1 (= hard
+reference) are used, and the kind value is always 0 for free chunks
+and 7 for allocated chunks, indicating a non-VM object.
+<pre>
+u4 heap ID
+u1 size of allocation unit, in bytes (e.g. 8 bytes)
+u4 virtual address of segment start
+u4 offset of this piece (relative to the virtual address)
+u4 length of piece, in allocation units
+var usage data
+</pre>
+
+<h3>Generic Replies</h3>
+
+The client-side chunk handlers need a common way to report simple success
+or failure. By convention, an empty reply packet indicates success.
+
+<h4>Chunk FAIL (client --> server, reply only)</h4>
+<p>The chunk includes a machine-readable error code and a
+human-readable error message. Server code can associate the failure
+with the original request by comparing the JDWP packet ID.
+<p>This allows a standard way of, for example, rejecting badly-formed
+request packets.
+<pre>
+u4 error code
+u4 error message len (in 16-bit chars)
+var error message (UTF-16)
+</pre>
+
+<h3>Miscellaneous</h3>
+
+<h4>Chunk EXIT (server --> client)</h4>
+<p>Cause the client to exit with the specified status, using System.exit().
+Useful for certain kinds of testing.
+<pre>
+u4 exit status
+</pre>
+
+<h4>Chunk DTRC (server --> client)</h4>
+<p>[TBD] start/stop dmtrace; can send the results back over the wire. For
+size reasons we probably need "sending", "data", "key", "finished" as
+4 separate chunks/packets rather than one glob.
+
+
+<h2>Client API</h2>
+
+<p>The API is written in the Java programming language
+for convenience. The code is free to call native methods if appropriate.
+
+<h3>Chunk Handler API</h3>
+
+<p>The basic idea is that arbitrary code can register handlers for
+specific chunk types. When a DDM chunk with that type arrives, the
+appropriate handler is invoked. The handler's return value provides the
+response to the server.
+
+<p>There are two packages. android.ddm lives in the "framework" library,
+and has all of the chunk handlers and registration code. It can freely
+use Android classes. org.apache.harmony.dalvik.ddmc lives in the "core"
+library, and has
+some base classes and features that interact with the VM. Nothing should
+need to modify the org.apache.harmony.dalvik.ddmc classes.
+
+<p>The DDM classes pass chunks of data around with a simple class:
+
+<pre class=prettyprint>
+class Chunk {
+ int type;
+ byte[] data;
+ int offset, length;
+};
+</pre>
+
+<p>The chunk handlers accept and return them:
+<pre class=prettyprint>
+public Chunk handleChunk(Chunk request)
+</pre>
+<p>The code is free to parse the chunk and generate a response in any
+way it chooses. Big-endian byte ordering is recommended but not mandatory.
+<p>Chunk handlers will be notified when a DDM server connects or disconnects,
+so that they can perform setup and cleanup operations:
+<pre class=prettyprint>
+public void connected()
+public void disconnected()
+</pre>
+
+<p>The method processes the request, formulates a response, and returns it.
+If the method returns null, an empty JDWP success message will be returned.
+<p>The request/response interaction is essentially asynchronous in the
+protocol. The packets are linked together with the JDWP message ID.
+<p>[We could use ByteBuffer here instead of byte[], but it doesn't gain
+us much. Wrapping a ByteBuffer around an array is easy. We don't want
+to pass the full packet in because we could have multiple chunks in one
+request packet. The DDM code needs to collect and aggregate the responses
+to all chunks into a single JDWP response packet. Parties wanting to
+write multiple chunks in response to a single chunk should send a null
+response back and use "sendChunk()" to send the data independently.]
+
+<h3>Unsolicited event API</h3>
+
+<p>If a piece of code wants to send a chunk of data to the server at some
+arbitrary time, it may do so with a method provided by
+org.apache.harmony.dalvik.DdmServer:
+
+<pre class=prettyprint>
+public static void sendChunk(Chunk chunk)
+</pre>
+
+<p>There is no response or status code. No exceptions are thrown.
+
+
+<h2>Server API</h2>
+
+<p>This is similar to the client side in many ways, but makes extensive
+use of ByteBuffer in a perhaps misguided attempt to use java.nio.channels
+and avoid excessive thread creation and unnecessary data copying.
+
+<p>Upon receipt of a packet, the server will identify it as one of:
+<ol>
+ <li>Message to be passed through to the debugger
+ <li>Response to an earlier request
+ <li>Unsolicited event packet
+</ol>
+<p>To handle (2), when messages are sent from the server to the client,
+the message must be paired with a callback method. The response might be
+delayed for a while -- or might never arrive -- so the server can't block
+waiting for responses from the client.
+<p>The chunk handlers look like this:
+<pre class=prettyprint>
+public void handleChunk(Client client, int type,
+ ByteBuffer data, boolean isReply, int msgId)
+</pre>
+<p>The arguments are:
+<dl>
+ <dt>client
+ <dd>An object representing the client VM that send us the packet.
+ <dt>type
+ <dd>The 32-bit chunk type.
+ <dt>data
+ <dd>The data. The data's length can be determined by calling data.limit().
+ <dt>isReply
+ <dd>Set to "true" if this was a reply to a message we sent earlier,
+ "false" if the client sent this unsolicited.
+ <dt>msgId
+ <dd>The JDWP message ID. Useful for connecting replies with requests.
+</dl>
+<p>If a handler doesn't like the contents of a packet, it should log an
+error message and return. If the handler doesn't recognize the packet at
+all, it can call the superclass' handleUnknownChunk() method.
+
+<p>As with the client, the server code can be notified when clients
+connect or disconnect. This allows the handler to send initialization
+code immediately after a connect, or clean up after a disconnect.
+<p>Data associated with a client can be stored in a ClientData object,
+which acts as a general per-client dumping around for VM and UI state.
+
+
+<P><BR>
+
+<HR>
+
+<address>Copyright © 2007 The Android Open Source Project</address>
+
+</body>
+</HTML>
diff --git a/docs/dex-format.css b/docs/dex-format.css
new file mode 100644
index 0000000..17e935f
--- /dev/null
+++ b/docs/dex-format.css
@@ -0,0 +1,387 @@
+h1 {
+ font-family: serif;
+ border-top-style: solid;
+ border-top-width: 5px;
+ padding-top: 9pt;
+ margin-top: 40pt;
+ color: #222266;
+}
+
+h1.title {
+ border: none;
+}
+
+h2 {
+ font-family: serif;
+ border-top-style: solid;
+ border-top-width: 2px;
+ border-color: #ccccdd;
+ padding-top: 9pt;
+ margin-top: 40pt;
+ margin-bottom: 2pt;
+ color: #222266;
+}
+
+h3 {
+ font-family: serif;
+ font-style: bold;
+ margin-top: 20pt;
+ margin-bottom: 2pt;
+ color: #222266;
+}
+
+h4 {
+ font-family: serif;
+ font-style: italic;
+ margin-top: 2pt;
+ margin-bottom: 2pt;
+ color: #666688;
+}
+
+@media print {
+ table {
+ font-size: 8pt;
+ }
+}
+
+@media screen {
+ table {
+ font-size: 10pt;
+ }
+}
+
+pre {
+ background: #eeeeff;
+ border-color: #aaaaff;
+ border-style: solid;
+ border-width: 1px;
+ margin-left: 40pt;
+ margin-right: 40pt;
+ padding: 6pt;
+}
+
+table {
+ border-collapse: collapse;
+ margin-top: 10pt;
+ margin-left: 40pt;
+ margin-right: 40pt;
+}
+
+table th {
+ font-family: sans-serif;
+ background: #aabbff;
+}
+
+table td {
+ font-family: sans-serif;
+ border-top-style: solid;
+ border-bottom-style: solid;
+ border-width: 1px;
+ border-color: #aaaaff;
+ padding-top: 3pt;
+ padding-bottom: 3pt;
+ padding-left: 3pt;
+ padding-right: 4pt;
+ background: #eeeeff;
+}
+
+table p {
+ margin-bottom: 0pt;
+}
+
+/* for the bnf syntax sections */
+
+table.bnf {
+ background: #eeeeff;
+ border-color: #aaaaff;
+ border-style: solid;
+ border-width: 1px;
+ margin-top: 3pt;
+ margin-bottom: 3pt;
+ padding-top: 2pt;
+ padding-bottom: 6pt;
+ padding-left: 6pt;
+ padding-right: 6pt;
+}
+
+table.bnf td {
+ border: none;
+ padding-left: 6pt;
+ padding-right: 6pt;
+ padding-top: 1pt;
+ padding-bottom: 1pt;
+}
+
+table.bnf td:first-child {
+ padding-right: 0pt;
+ width: 8pt;
+}
+
+table.bnf td:first-child td {
+ padding-left: 0pt;
+}
+
+table.bnf td.def {
+ padding-top: 6pt;
+}
+
+table.bnf td.bar {
+ padding-left: 15pt;
+}
+
+table.bnf code {
+ font-weight: bold;
+}
+
+
+/* for the type name guide */
+
+table.guide {
+ margin-top: 20pt;
+ margin-bottom: 20pt;
+}
+
+table.guide td:first-child {
+ font-family: monospace;
+ width: 15%;
+}
+
+table.guide td:first-child + td {
+ font-family: sans-serif;
+ width: 85%;
+}
+
+
+/* for the LEB128 example tables */
+
+table.leb128Bits {
+ margin-top: 20pt;
+ margin-bottom: 20pt;
+}
+
+table.leb128Bits td {
+ border-left: solid #aaaaff 1px;
+ border-right: solid #aaaaff 1px;
+}
+
+table.leb128Bits td.start1 {
+ border-left: none;
+}
+
+table.leb128Bits td.start2 {
+ border-left: solid #000 2px;
+}
+
+table.leb128Bits td.end2 {
+ border-right: none;
+}
+
+table.leb128 {
+ margin-top: 20pt;
+ margin-bottom: 20pt;
+}
+
+table.leb128 td:first-child {
+ font-family: monospace;
+ text-align: center;
+ width: 31%;
+}
+
+table.leb128 td:first-child + td {
+ font-family: monospace;
+ text-align: center;
+ width: 23%;
+}
+
+table.leb128 td:first-child + td + td {
+ font-family: monospace;
+ text-align: center;
+ width: 23%;
+}
+
+table.leb128 td:first-child + td + td + td {
+ font-family: monospace;
+ text-align: center;
+ width: 23%;
+}
+
+
+/* for the general format tables */
+
+table.format {
+ margin-top: 20pt;
+ margin-bottom: 20pt;
+}
+
+table.format td:first-child {
+ font-family: monospace;
+ width: 20%;
+}
+
+table.format td:first-child + td {
+ font-family: monospace;
+ width: 20%;
+}
+
+table.format td:first-child + td + td {
+ width: 60%;
+}
+
+table.format td i {
+ font-family: sans-serif;
+}
+
+
+/* for the type code table */
+
+table.typeCodes {
+ margin-top: 20pt;
+ margin-bottom: 20pt;
+}
+
+table.typeCodes td:first-child {
+ font-family: monospace;
+ width: 30%;
+}
+
+table.typeCodes td:first-child + td {
+ font-family: monospace;
+ width: 30%;
+}
+
+table.typeCodes td:first-child + td + td {
+ font-family: monospace;
+ width: 10%;
+}
+
+table.typeCodes td:first-child + td + td + td {
+ font-family: monospace;
+ width: 30%;
+}
+
+table.typeCodes td i {
+ font-family: sans-serif;
+}
+
+
+/* for the access flags table */
+
+table.accessFlags {
+ margin-top: 20pt;
+ margin-bottom: 20pt;
+}
+
+table.accessFlags td:first-child {
+ font-family: monospace;
+ width: 10%;
+}
+
+table.accessFlags td:first-child + td {
+ font-family: monospace;
+ width: 6%;
+}
+
+table.accessFlags td:first-child + td + td {
+ width: 28%;
+}
+
+table.accessFlags td:first-child + td + td + td {
+ width: 28%;
+}
+
+table.accessFlags td:first-child + td + td + td + td {
+ width: 28%;
+}
+
+table.accessFlags i {
+ font-family: sans-serif;
+}
+
+
+/* for the descriptor table */
+
+table.descriptor {
+ margin-top: 20pt;
+ margin-bottom: 20pt;
+}
+
+table.descriptor td:first-child {
+ font-family: monospace;
+ width: 25%;
+}
+
+table.descriptor td:first-child + td {
+ font-family: sans-serif;
+ width: 75%;
+}
+
+
+/* for the debug bytecode table */
+
+table.debugByteCode {
+ margin-top: 20pt;
+ margin-bottom: 20pt;
+}
+
+table.debugByteCode td:first-child {
+ font-family: monospace;
+ width: 20%;
+}
+
+table.debugByteCode td:first-child + td {
+ font-family: monospace;
+ width: 5%;
+}
+
+table.debugByteCode td:first-child + td + td{
+ font-family: monospace;
+ width: 15%;
+}
+
+table.debugByteCode td:first-child + td + td + td {
+ width: 25%;
+}
+
+table.debugByteCode td:first-child + td + td + td + td {
+ width: 35%;
+}
+
+table.debugByteCode i {
+ font-family: sans-serif;
+}
+
+
+/* for the encoded value table */
+
+table.encodedValue {
+ margin-top: 20pt;
+ margin-bottom: 20pt;
+}
+
+table.encodedValue td:first-child {
+ font-family: monospace;
+ width: 12%;
+}
+
+table.encodedValue td:first-child + td {
+ font-family: monospace;
+ width: 10%;
+}
+
+table.encodedValue td:first-child + td + td {
+ font-family: monospace;
+ width: 15%;
+}
+
+table.encodedValue td:first-child + td + td + td {
+ font-family: monospace;
+ width: 15%;
+}
+
+table.encodedValue td:first-child + td + td + td + td {
+ width: 48%;
+}
+
+table.encodedValue td i {
+ font-family: sans-serif;
+}
diff --git a/docs/dex-format.html b/docs/dex-format.html
new file mode 100644
index 0000000..88a7fb0
--- /dev/null
+++ b/docs/dex-format.html
@@ -0,0 +1,3043 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>.dex — Dalvik Executable Format</title>
+<link rel=stylesheet href="dex-format.css">
+</head>
+
+<body>
+
+<h1 class="title"><code>.dex</code> — Dalvik Executable Format</h1>
+<p>Copyright © 2007 The Android Open Source Project
+
+<p>This document describes the layout and contents of <code>.dex</code>
+files, which are used to hold a set of class definitions and their associated
+adjunct data.</p>
+
+<h1>Guide To Types</h1>
+
+<table class="guide">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>byte</td>
+ <td>8-bit signed int</td>
+</tr>
+<tr>
+ <td>ubyte</td>
+ <td>8-bit unsigned int</td>
+</tr>
+<tr>
+ <td>short</td>
+ <td>16-bit signed int, little-endian</td>
+</tr>
+<tr>
+ <td>ushort</td>
+ <td>16-bit unsigned int, little-endian</td>
+</tr>
+<tr>
+ <td>int</td>
+ <td>32-bit signed int, little-endian</td>
+</tr>
+<tr>
+ <td>uint</td>
+ <td>32-bit unsigned int, little-endian</td>
+</tr>
+<tr>
+ <td>long</td>
+ <td>64-bit signed int, little-endian</td>
+</tr>
+<tr>
+ <td>ulong</td>
+ <td>64-bit unsigned int, little-endian</td>
+</tr>
+<tr>
+ <td>sleb128</td>
+ <td>signed LEB128, variable-length (see below)</td>
+</tr>
+<tr>
+ <td>uleb128</td>
+ <td>unsigned LEB128, variable-length (see below)</td>
+</tr>
+<tr>
+ <td>uleb128p1</td>
+ <td>unsigned LEB128 plus <code>1</code>, variable-length (see below)</td>
+</tr>
+</tbody>
+</table>
+
+<h3>LEB128</h3>
+
+<p>LEB128 ("<b>L</b>ittle-<b>E</b>ndian <b>B</b>ase <b>128</b>") is a
+variable-length encoding for
+arbitrary signed or unsigned integer quantities. The format was
+borrowed from the <a href="http://dwarfstd.org/Dwarf3Std.php">DWARF3</a>
+specification. In a <code>.dex</code> file, LEB128 is only ever used to
+encode 32-bit quantities.</p>
+
+<p>Each LEB128 encoded value consists of one to five
+bytes, which together represent a single 32-bit value. Each
+byte has its most significant bit set except for the final byte in the
+sequence, which has its most significant bit clear. The remaining
+seven bits of each byte are payload, with the least significant seven
+bits of the quantity in the first byte, the next seven in the second
+byte and so on. In the case of a signed LEB128 (<code>sleb128</code>),
+the most significant payload bit of the final byte in the sequence is
+sign-extended to produce the final value. In the unsigned case
+(<code>uleb128</code>), any bits not explicitly represented are
+interpreted as <code>0</code>.
+
+<table class="leb128Bits">
+<thead>
+<tr><th colspan="16">Bitwise diagram of a two-byte LEB128 value</th></tr>
+<tr>
+ <th colspan="8">First byte</td>
+ <th colspan="8">Second byte</td>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td class="start1"><code>1</code></td>
+ <td>bit<sub>6</sub></td>
+ <td>bit<sub>5</sub></td>
+ <td>bit<sub>4</sub></td>
+ <td>bit<sub>3</sub></td>
+ <td>bit<sub>2</sub></td>
+ <td>bit<sub>1</sub></td>
+ <td>bit<sub>0</sub></td>
+ <td class="start2"><code>0</code></td>
+ <td>bit<sub>13</sub></td>
+ <td>bit<sub>12</sub></td>
+ <td>bit<sub>11</sub></td>
+ <td>bit<sub>10</sub></td>
+ <td>bit<sub>9</sub></td>
+ <td>bit<sub>8</sub></td>
+ <td class="end2">bit<sub>7</sub></td>
+</tr>
+</tbody>
+</table>
+
+<p>The variant <code>uleb128p1</code> is used to represent a signed
+value, where the representation is of the value <i>plus one</i> encoded
+as a <code>uleb128</code>. This makes the encoding of <code>-1</code>
+(alternatively thought of as the unsigned value <code>0xffffffff</code>)
+— but no other negative number — a single byte, and is
+useful in exactly those cases where the represented number must either
+be non-negative or <code>-1</code> (or <code>0xffffffff</code>),
+and where no other negative values are allowed (or where large unsigned
+values are unlikely to be needed).</p>
+
+<p>Here are some examples of the formats:</p>
+
+<table class="leb128">
+<thead>
+<tr>
+ <th>Encoded Sequence</th>
+ <th>As <code>sleb128</code></th>
+ <th>As <code>uleb128</code></th>
+ <th>As <code>uleb128p1</code></th>
+</tr>
+</thead>
+<tbody>
+ <tr><td>00</td><td>0</td><td>0</td><td>-1</td></tr>
+ <tr><td>01</td><td>1</td><td>1</td><td>0</td></tr>
+ <tr><td>7f</td><td>-1</td><td>127</td><td>126</td></tr>
+ <tr><td>80 7f</td><td>-128</td><td>16256</td><td>16255</td></tr>
+</tbody>
+</table>
+
+<h1>Overall File Layout</h1>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>header</td>
+ <td>header_item</td>
+ <td>the header</td>
+</tr>
+<tr>
+ <td>string_ids</td>
+ <td>string_id_item[]</td>
+ <td>string identifiers list. These are identifiers for all the strings
+ used by this file, either for internal naming (e.g., type descriptors)
+ or as constant objects referred to by code. This list must be sorted
+ by string contents, using UTF-16 code point values (not in a
+ locale-sensitive manner).
+ </td>
+</tr>
+<tr>
+ <td>type_ids</td>
+ <td>type_id_item[]</td>
+ <td>type identifiers list. These are identifiers for all types (classes,
+ arrays, or primitive types) referred to by this file, whether defined
+ in the file or not. This list must be sorted by <code>string_id</code>
+ index.
+ </td>
+</tr>
+<tr>
+ <td>proto_ids</td>
+ <td>proto_id_item[]</td>
+ <td>method prototype identifiers list. These are identifiers for all
+ prototypes referred to by this file. This list must be sorted in
+ return-type (by <code>type_id</code> index) major order, and then
+ by arguments (also by <code>type_id</code> index).
+ </td>
+</tr>
+<tr>
+ <td>field_ids</td>
+ <td>field_id_item[]</td>
+ <td>field identifiers list. These are identifiers for all fields
+ referred to by this file, whether defined in the file or not. This
+ list must be sorted, where the defining type (by <code>type_id</code>
+ index) is the major order, field name (by <code>string_id</code> index)
+ is the intermediate order, and type (by <code>type_id</code> index)
+ is the minor order.
+ </td>
+</tr>
+<tr>
+ <td>method_ids</td>
+ <td>method_id_item[]</td>
+ <td>method identifiers list. These are identifiers for all methods
+ referred to by this file, whether defined in the file or not. This
+ list must be sorted, where the defining type (by <code>type_id</code>
+ index) is the major order, method name (by <code>string_id</code>
+ index) is the intermediate order, and method
+ prototype (by <code>proto_id</code> index) is the minor order.
+ </td>
+</tr>
+<tr>
+ <td>class_defs</td>
+ <td>class_def_item[]</td>
+ <td>class definitions list. The classes must be ordered such that a given
+ class's superclass and implemented interfaces appear in the
+ list earlier than the referring class.
+ </td>
+</tr>
+<tr>
+ <td>data</td>
+ <td>ubyte[]</td>
+ <td>data area, containing all the support data for the tables listed above.
+ Different items have different alignment requirements, and
+ padding bytes are inserted before each item if necessary to achieve
+ proper alignment.
+ </td>
+</tr>
+<tr>
+ <td>link_data</td>
+ <td>ubyte[]</td>
+ <td>data used in statically linked files. The format of the data in
+ this section is left unspecified by this document;
+ this section is empty in unlinked files, and runtime implementations
+ may use it as they see fit.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h1>Bitfield, String, and Constant Definitions</h1>
+
+<h2><code>DEX_FILE_MAGIC</code></h2>
+<h4>embedded in <code>header_item</code></h4>
+
+<p>The constant array/string <code>DEX_FILE_MAGIC</code> is the list of
+bytes that must appear at the beginning of a <code>.dex</code> file
+in order for it to be recognized as such. The value intentionally
+contains a newline (<code>"\n"</code> or <code>0x0a</code>) and a
+null byte (<code>"\0"</code> or <code>0x00</code>) in order to help
+in the detection of certain forms of corruption. The value also
+encodes a format version number as three decimal digits, which is
+expected to increase monotonically over time as the format evolves.</p>
+
+<pre>
+ubyte[8] DEX_FILE_MAGIC = { 0x64 0x65 0x78 0x0a 0x30 0x33 0x35 0x00 }
+ = "dex\n035\0"
+</pre>
+
+<p><b>Note:</b> At least a couple earlier versions of the format have
+been used in widely-available public software releases. For example,
+version <code>009</code> was used for the M3 releases of the
+Android platform (November-December 2007),
+and version <code>013</code> was used for the M5 releases of the Android
+platform (February-March 2008). In several respects, these earlier versions
+of the format differ significantly from the version described in this
+document.</p>
+
+<h2><code>ENDIAN_CONSTANT</code> and <code>REVERSE_ENDIAN_CONSTANT</code></h2>
+<h4>embedded in <code>header_item</code></h4>
+
+<p>The constant <code>ENDIAN_CONSTANT</code> is used to indicate the
+endianness of the file in which it is found. Although the standard
+<code>.dex</code> format is little-endian, implementations may choose
+to perform byte-swapping. Should an implementation come across a
+header whose <code>endian_tag</code> is <code>REVERSE_ENDIAN_CONSTANT</code>
+instead of <code>ENDIAN_CONSTANT</code>, it would know that the file
+has been byte-swapped from the expected form.</p>
+
+<pre>
+uint ENDIAN_CONSTANT = 0x12345678;
+uint REVERSE_ENDIAN_CONSTANT = 0x78563412;
+</pre>
+
+<h2><code>NO_INDEX</code></h2>
+<h4>embedded in <code>class_def_item</code> and
+<code>debug_info_item</code></h4>
+
+<p>The constant <code>NO_INDEX</code> is used to indicate that
+an index value is absent.</p>
+
+<p><b>Note:</b> This value isn't defined to be
+<code>0</code>, because that is in fact typically a valid index.</p>
+
+<p><b>Also Note:</b> The chosen value for <code>NO_INDEX</code> is
+representable as a single byte in the <code>uleb128p1</code> encoding.</p>
+
+<pre>
+uint NO_INDEX = 0xffffffff; // == -1 if treated as a signed int
+</pre>
+
+<h2><code>access_flags</code> Definitions</h2>
+<h4>embedded in <code>class_def_item</code>,
+<code>field_item</code>, <code>method_item</code>, and
+<code>InnerClass</code></h4>
+
+<p>Bitfields of these flags are used to indicate the accessibility and
+overall properties of classes and class members.</p>
+
+<table class="accessFlags">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Value</th>
+ <th>For Classes (and <code>InnerClass</code> annotations)</th>
+ <th>For Fields</th>
+ <th>For Methods</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>ACC_PUBLIC</td>
+ <td>0x1</td>
+ <td><code>public</code>: visible everywhere</td>
+ <td><code>public</code>: visible everywhere</td>
+ <td><code>public</code>: visible everywhere</td>
+</tr>
+<tr>
+ <td>ACC_PRIVATE</td>
+ <td>0x2</td>
+ <td><super>*</super>
+ <code>private</code>: only visible to defining class
+ </td>
+ <td><code>private</code>: only visible to defining class</td>
+ <td><code>private</code>: only visible to defining class</td>
+</tr>
+<tr>
+ <td>ACC_PROTECTED</td>
+ <td>0x4</td>
+ <td><super>*</super>
+ <code>protected</code>: visible to package and subclasses
+ </td>
+ <td><code>protected</code>: visible to package and subclasses</td>
+ <td><code>protected</code>: visible to package and subclasses</td>
+</tr>
+<tr>
+ <td>ACC_STATIC</td>
+ <td>0x8</td>
+ <td><super>*</super>
+ <code>static</code>: is not constructed with an outer
+ <code>this</code> reference</td>
+ <td><code>static</code>: global to defining class</td>
+ <td><code>static</code>: does not take a <code>this</code> argument</td>
+</tr>
+<tr>
+ <td>ACC_FINAL</td>
+ <td>0x10</td>
+ <td><code>final</code>: not subclassable</td>
+ <td><code>final</code>: immutable after construction</td>
+ <td><code>final</code>: not overridable</td>
+</tr>
+<tr>
+ <td>ACC_SYNCHRONIZED</td>
+ <td>0x20</td>
+ <td> </td>
+ <td> </td>
+ <td><code>synchronized</code>: associated lock automatically acquired
+ around call to this method. <b>Note:</b> This is only valid to set when
+ <code>ACC_NATIVE</code> is also set.</td>
+</tr>
+<tr>
+ <td>ACC_VOLATILE</td>
+ <td>0x40</td>
+ <td> </td>
+ <td><code>volatile</code>: special access rules to help with thread
+ safety</td>
+ <td> </td>
+</tr>
+<tr>
+ <td>ACC_BRIDGE</td>
+ <td>0x40</td>
+ <td> </td>
+ <td> </td>
+ <td>bridge method, added automatically by compiler as a type-safe
+ bridge</td>
+</tr>
+<tr>
+ <td>ACC_TRANSIENT</td>
+ <td>0x80</td>
+ <td> </td>
+ <td><code>transient</code>: not to be saved by default serialization</td>
+ <td> </td>
+</tr>
+<tr>
+ <td>ACC_VARARGS</td>
+ <td>0x80</td>
+ <td> </td>
+ <td> </td>
+ <td>last argument should be treated as a "rest" argument by compiler</td>
+</tr>
+<tr>
+ <td>ACC_NATIVE</td>
+ <td>0x100</td>
+ <td> </td>
+ <td> </td>
+ <td><code>native</code>: implemented in native code</td>
+</tr>
+<tr>
+ <td>ACC_INTERFACE</td>
+ <td>0x200</td>
+ <td><code>interface</code>: multiply-implementable abstract class</td>
+ <td> </td>
+ <td> </td>
+</tr>
+<tr>
+ <td>ACC_ABSTRACT</td>
+ <td>0x400</td>
+ <td><code>abstract</code>: not directly instantiable</td>
+ <td> </td>
+ <td><code>abstract</code>: unimplemented by this class</td>
+</tr>
+<tr>
+ <td>ACC_STRICT</td>
+ <td>0x800</td>
+ <td> </td>
+ <td> </td>
+ <td><code>strictfp</code>: strict rules for floating-point arithmetic</td>
+</tr>
+<tr>
+ <td>ACC_SYNTHETIC</td>
+ <td>0x1000</td>
+ <td>not directly defined in source code</td>
+ <td>not directly defined in source code</td>
+ <td>not directly defined in source code</td>
+</tr>
+<tr>
+ <td>ACC_ANNOTATION</td>
+ <td>0x2000</td>
+ <td>declared as an annotation class</td>
+ <td> </td>
+ <td> </td>
+</tr>
+<tr>
+ <td>ACC_ENUM</td>
+ <td>0x4000</td>
+ <td>declared as an enumerated type</td>
+ <td>declared as an enumerated value</td>
+ <td> </td>
+</tr>
+<tr>
+ <td><i>(unused)</i></td>
+ <td>0x8000</td>
+ <td> </td>
+ <td> </td>
+ <td> </td>
+</tr>
+<tr>
+ <td>ACC_CONSTRUCTOR</td>
+ <td>0x10000</td>
+ <td> </td>
+ <td> </td>
+ <td>constructor method (class or instance initializer)</td>
+</tr>
+<tr>
+ <td>ACC_DECLARED_<br/>SYNCHRONIZED</td>
+ <td>0x20000</td>
+ <td> </td>
+ <td> </td>
+ <td>declared <code>synchronized</code>. <b>Note:</b> This has no effect on
+ execution (other than in reflection of this flag, per se).
+ </td>
+</tr>
+</tbody>
+</table>
+
+<p><super>*</super> Only allowed on for <code>InnerClass</code> annotations,
+and must not ever be on in a <code>class_def_item</code>.</p>
+
+<h2>MUTF-8 (Modified UTF-8) Encoding</h2>
+
+<p>As a concession to easier legacy support, the <code>.dex</code> format
+encodes its string data in a de facto standard modified UTF-8 form, hereafter
+referred to as MUTF-8. This form is identical to standard UTF-8, except:</p>
+
+<ul>
+ <li>Only the one-, two-, and three-byte encodings are used.</li>
+ <li>Code points in the range <code>U+10000</code> …
+ <code>U+10ffff</code> are encoded as a surrogate pair, each of
+ which is represented as a three-byte encoded value.</li>
+ <li>The code point <code>U+0000</code> is encoded in two-byte form.</li>
+ <li>A plain null byte (value <code>0</code>) indicates the end of
+ a string, as is the standard C language interpretation.</li>
+</ul>
+
+<p>The first two items above can be summarized as: MUTF-8
+is an encoding format for UTF-16, instead of being a more direct
+encoding format for Unicode characters.</p>
+
+<p>The final two items above make it simultaneously possible to include
+the code point <code>U+0000</code> in a string <i>and</i> still manipulate
+it as a C-style null-terminated string.</p>
+
+<p>However, the special encoding of <code>U+0000</code> means that, unlike
+normal UTF-8, the result of calling the standard C function
+<code>strcmp()</code> on a pair of MUTF-8 strings does not always
+indicate the properly signed result of comparison of <i>unequal</i> strings.
+When ordering (not just equality) is a concern, the most straightforward
+way to compare MUTF-8 strings is to decode them character by character,
+and compare the decoded values. (However, more clever implementations are
+also possible.)</p>
+
+<p>Please refer to <a href="http://unicode.org">The Unicode
+Standard</a> for further information about character encoding.
+MUTF-8 is actually closer to the (relatively less well-known) encoding
+<a href="http://www.unicode.org/reports/tr26/">CESU-8</a> than to UTF-8
+per se.</p>
+
+<h2><code>encoded_value</code> Encoding</h2>
+<h4>embedded in <code>annotation_element</code> and
+<code>encoded_array_item</code></h4>
+
+<p>An <code>encoded_value</code> is an encoded piece of (nearly)
+arbitrary hierarchically structured data. The encoding is meant to
+be both compact and straightforward to parse.</p>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>(value_arg << 5) | value_type</td>
+ <td>ubyte</td>
+ <td>byte indicating the type of the immediately subsequent
+ <code>value</code> along
+ with an optional clarifying argument in the high-order three bits.
+ See below for the various <code>value</code> definitions.
+ In most cases, <code>value_arg</code> encodes the length of
+ the immediately-subsequent <code>value</code> in bytes, as
+ <code>(size - 1)</code>, e.g., <code>0</code> means that
+ the value requires one byte, and <code>7</code> means it requires
+ eight bytes; however, there are exceptions as noted below.
+ </td>
+</tr>
+<tr>
+ <td>value</td>
+ <td>ubyte[]</td>
+ <td>bytes representing the value, variable in length and interpreted
+ differently for different <code>value_type</code> bytes, though
+ always little-endian. See the various value definitions below for
+ details.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h3>Value Formats</h3>
+
+<table class="encodedValue">
+<thead>
+<tr>
+ <th>Type Name</th>
+ <th><code>value_type</code></th>
+ <th><code>value_arg</code> Format</th>
+ <th><code>value</code> Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>VALUE_BYTE</td>
+ <td>0x00</td>
+ <td><i>(none; must be <code>0</code>)</i></td>
+ <td>ubyte[1]</td>
+ <td>signed one-byte integer value</td>
+</tr>
+<tr>
+ <td>VALUE_SHORT</td>
+ <td>0x02</td>
+ <td>size - 1 (0…1)</td>
+ <td>ubyte[size]</td>
+ <td>signed two-byte integer value, sign-extended</td>
+</tr>
+<tr>
+ <td>VALUE_CHAR</td>
+ <td>0x03</td>
+ <td>size - 1 (0…1)</td>
+ <td>ubyte[size]</td>
+ <td>unsigned two-byte integer value, zero-extended</td>
+</tr>
+<tr>
+ <td>VALUE_INT</td>
+ <td>0x04</td>
+ <td>size - 1 (0…3)</td>
+ <td>ubyte[size]</td>
+ <td>signed four-byte integer value, sign-extended</td>
+</tr>
+<tr>
+ <td>VALUE_LONG</td>
+ <td>0x06</td>
+ <td>size - 1 (0…7)</td>
+ <td>ubyte[size]</td>
+ <td>signed eight-byte integer value, sign-extended</td>
+</tr>
+<tr>
+ <td>VALUE_FLOAT</td>
+ <td>0x10</td>
+ <td>size - 1 (0…3)</td>
+ <td>ubyte[size]</td>
+ <td>four-byte bit pattern, zero-extended <i>to the right</i>, and
+ interpreted as an IEEE754 32-bit floating point value
+ </td>
+</tr>
+<tr>
+ <td>VALUE_DOUBLE</td>
+ <td>0x11</td>
+ <td>size - 1 (0…7)</td>
+ <td>ubyte[size]</td>
+ <td>eight-byte bit pattern, zero-extended <i>to the right</i>, and
+ interpreted as an IEEE754 64-bit floating point value
+ </td>
+</tr>
+<tr>
+ <td>VALUE_STRING</td>
+ <td>0x17</td>
+ <td>size - 1 (0…3)</td>
+ <td>ubyte[size]</td>
+ <td>unsigned (zero-extended) four-byte integer value,
+ interpreted as an index into
+ the <code>string_ids</code> section and representing a string value
+ </td>
+</tr>
+<tr>
+ <td>VALUE_TYPE</td>
+ <td>0x18</td>
+ <td>size - 1 (0…3)</td>
+ <td>ubyte[size]</td>
+ <td>unsigned (zero-extended) four-byte integer value,
+ interpreted as an index into
+ the <code>type_ids</code> section and representing a reflective
+ type/class value
+ </td>
+</tr>
+<tr>
+ <td>VALUE_FIELD</td>
+ <td>0x19</td>
+ <td>size - 1 (0…3)</td>
+ <td>ubyte[size]</td>
+ <td>unsigned (zero-extended) four-byte integer value,
+ interpreted as an index into
+ the <code>field_ids</code> section and representing a reflective
+ field value
+ </td>
+</tr>
+<tr>
+ <td>VALUE_METHOD</td>
+ <td>0x1a</td>
+ <td>size - 1 (0…3)</td>
+ <td>ubyte[size]</td>
+ <td>unsigned (zero-extended) four-byte integer value,
+ interpreted as an index into
+ the <code>method_ids</code> section and representing a reflective
+ method value
+ </td>
+</tr>
+<tr>
+ <td>VALUE_ENUM</td>
+ <td>0x1b</td>
+ <td>size - 1 (0…3)</td>
+ <td>ubyte[size]</td>
+ <td>unsigned (zero-extended) four-byte integer value,
+ interpreted as an index into
+ the <code>field_ids</code> section and representing the value of
+ an enumerated type constant
+ </td>
+</tr>
+<tr>
+ <td>VALUE_ARRAY</td>
+ <td>0x1c</td>
+ <td><i>(none; must be <code>0</code>)</i></td>
+ <td>encoded_array</td>
+ <td>an array of values, in the format specified by
+ "<code>encoded_array</code> Format" below. The size
+ of the <code>value</code> is implicit in the encoding.
+ </td>
+</tr>
+<tr>
+ <td>VALUE_ANNOTATION</td>
+ <td>0x1d</td>
+ <td><i>(none; must be <code>0</code>)</i></td>
+ <td>encoded_annotation</td>
+ <td>a sub-annotation, in the format specified by
+ "<code>encoded_annotation</code> Format" below. The size
+ of the <code>value</code> is implicit in the encoding.
+ </td>
+</tr>
+<tr>
+ <td>VALUE_NULL</td>
+ <td>0x1e</td>
+ <td><i>(none; must be <code>0</code>)</i></td>
+ <td><i>(none)</i></td>
+ <td><code>null</code> reference value</td>
+</tr>
+<tr>
+ <td>VALUE_BOOLEAN</td>
+ <td>0x1f</td>
+ <td>boolean (0…1)</td>
+ <td><i>(none)</i></td>
+ <td>one-bit value; <code>0</code> for <code>false</code> and
+ <code>1</code> for <code>true</code>. The bit is represented in the
+ <code>value_arg</code>.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h3><code>encoded_array</code> Format</h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>size</td>
+ <td>uleb128</td>
+ <td>number of elements in the array</td>
+</tr>
+<tr>
+ <td>values</td>
+ <td>encoded_value[size]</td>
+ <td>a series of <code>size</code> <code>encoded_value</code> byte
+ sequences in the format specified by this section, concatenated
+ sequentially.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h3><code>encoded_annotation</code> Format</h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>type_idx</td>
+ <td>uleb128</td>
+ <td>type of the annotation. This must be a class (not array or primitive)
+ type.
+ </td>
+</tr>
+<tr>
+ <td>size</td>
+ <td>uleb128</td>
+ <td>number of name-value mappings in this annotation</td>
+</tr>
+<tr>
+ <td>elements</td>
+ <td>annotation_element[size]</td>
+ <td>elements of the annotataion, represented directly in-line (not as
+ offsets). Elements must be sorted in increasing order by
+ <code>string_id</code> index.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h3><code>annotation_element</code> Format</h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>name_idx</td>
+ <td>uleb128</td>
+ <td>element name, represented as an index into the
+ <code>string_ids</code> section. The string must conform to the
+ syntax for <i>MemberName</i>, defined above.
+ </td>
+</tr>
+<tr>
+ <td>value</td>
+ <td>encoded_value</td>
+ <td>element value</td>
+</tr>
+</tbody>
+</table>
+
+<h2>String Syntax</h2>
+
+<p>There are several kinds of item in a <code>.dex</code> file which
+ultimately refer to a string. The following BNF-style definitions
+indicate the acceptable syntax for these strings.</p>
+
+<h3><i>SimpleName</i></h3>
+
+<p>A <i>SimpleName</i> is the basis for the syntax of the names of other
+things. The <code>.dex</code> format allows a fair amount of latitude
+here (much more than most common source languages). In brief, a simple
+name may consist of any low-ASCII alphabetic character or digit, a few
+specific low-ASCII symbols, and most non-ASCII code points that are not
+control, space, or special characters. Note that surrogate code points
+(in the range <code>U+d800</code> … <code>U+dfff</code>) are not
+considered valid name characters, per se, but Unicode supplemental
+characters <i>are</i> valid (which are represented by the final
+alternative of the rule for <i>SimpleNameChar</i>), and they should be
+represented in a file as pairs of surrogate code points in the MUTF-8
+encoding.</p>
+
+<table class="bnf">
+ <tr><td colspan="2" class="def"><i>SimpleName</i> →</td></tr>
+ <tr>
+ <td/>
+ <td><i>SimpleNameChar</i> (<i>SimpleNameChar</i>)*</td>
+ </tr>
+
+ <tr><td colspan="2" class="def"><i>SimpleNameChar</i> →</td></tr>
+ <tr>
+ <td/>
+ <td><code>'A'</code> … <code>'Z'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'a'</code> … <code>'z'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'0'</code> … <code>'9'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'$'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'-'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'_'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>U+00a1</code> … <code>U+1fff</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>U+2010</code> … <code>U+2027</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>U+2030</code> … <code>U+d7ff</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>U+e000</code> … <code>U+ffef</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>U+10000</code> … <code>U+10ffff</code></td>
+ </tr>
+</table>
+
+<h3><i>MemberName</i></h3>
+<h4>used by <code>field_id_item</code> and <code>method_id_item</code></h4>
+
+<p>A <i>MemberName</i> is the name of a member of a class, members being
+fields, methods, and inner classes.</p>
+
+<table class="bnf">
+ <tr><td colspan="2" class="def"><i>MemberName</i> →</td></tr>
+ <tr>
+ <td/>
+ <td><i>SimpleName</i></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'<'</code> <i>SimpleName</i> <code>'>'</code></td>
+ </tr>
+</table>
+
+<h3><i>FullClassName</i></h3>
+
+<p>A <i>FullClassName</i> is a fully-qualified class name, including an
+optional package specifier followed by a required name.</p>
+
+<table class="bnf">
+ <tr><td colspan="2" class="def"><i>FullClassName</i> →</td></tr>
+ <tr>
+ <td/>
+ <td><i>OptionalPackagePrefix</i> <i>SimpleName</i></td>
+ </tr>
+
+ <tr><td colspan="2" class="def"><i>OptionalPackagePrefix</i> →</td></tr>
+ <tr>
+ <td/>
+ <td>(<i>SimpleName</i> <code>'/'</code>)*</td>
+ </tr>
+</table>
+
+<h3><i>TypeDescriptor</i></h3>
+<h4>used by <code>type_id_item</code></h4>
+
+<p>A <i>TypeDescriptor</i> is the representation of any type, including
+primitives, classes, arrays, and <code>void</code>. See below for
+the meaning of the various versions.</p>
+
+<table class="bnf">
+ <tr><td colspan="2" class="def"><i>TypeDescriptor</i> →</td></tr>
+ <tr>
+ <td/>
+ <td><code>'V'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><i>FieldTypeDescriptor</i></td>
+ </tr>
+
+ <tr><td colspan="2" class="def"><i>FieldTypeDescriptor</i> →</td></tr>
+ <tr>
+ <td/>
+ <td><i>NonArrayFieldTypeDescriptor</i></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td>(<code>'['</code> * 1…255)
+ <i>NonArrayFieldTypeDescriptor</i></td>
+ </tr>
+
+ <tr>
+ <td colspan="2" class="def"><i>NonArrayFieldTypeDescriptor</i>→</td>
+ </tr>
+ <tr>
+ <td/>
+ <td><code>'Z'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'B'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'S'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'C'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'I'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'J'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'F'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'D'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'L'</code> <i>FullClassName</i> <code>';'</code></td>
+ </tr>
+</table>
+
+<h3><i>ShortyDescriptor</i></h3>
+<h4>used by <code>proto_id_item</code></h4>
+
+<p>A <i>ShortyDescriptor</i> is the short form representation of a method
+prototype, including return and parameter types, except that there is
+no distinction between various reference (class or array) types. Instead,
+all reference types are represented by a single <code>'L'</code> character.</p>
+
+<table class="bnf">
+ <tr><td colspan="2" class="def"><i>ShortyDescriptor</i> →</td></tr>
+ <tr>
+ <td/>
+ <td><i>ShortyReturnType</i> (<i>ShortyFieldType</i>)*</td>
+ </tr>
+
+ <tr><td colspan="2" class="def"><i>ShortyReturnType</i> →</td></tr>
+ <tr>
+ <td/>
+ <td><code>'V'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><i>ShortyFieldType</i></td>
+ </tr>
+
+ <tr><td colspan="2" class="def"><i>ShortyFieldType</i> →</td></tr>
+ <tr>
+ <td/>
+ <td><code>'Z'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'B'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'S'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'C'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'I'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'J'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'F'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'D'</code></td>
+ </tr>
+ <tr>
+ <td class="bar">|</td>
+ <td><code>'L'</code></td>
+ </tr>
+</table>
+
+<h2><i>TypeDescriptor</i> Semantics</h2>
+
+<p>This is the meaning of each of the variants of <i>TypeDescriptor</i>.</p>
+
+<table class="descriptor">
+<thead>
+<tr>
+ <th>Syntax</th>
+ <th>Meaning</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>V</td>
+ <td><code>void</code>; only valid for return types</td>
+</tr>
+<tr>
+ <td>Z</td>
+ <td><code>boolean</code></td>
+</tr>
+<tr>
+ <td>B</td>
+ <td><code>byte</code></td>
+</tr>
+<tr>
+ <td>S</td>
+ <td><code>short</code></td>
+</tr>
+<tr>
+ <td>C</td>
+ <td><code>char</code></td>
+</tr>
+<tr>
+ <td>I</td>
+ <td><code>int</code></td>
+</tr>
+<tr>
+ <td>J</td>
+ <td><code>long</code></td>
+</tr>
+<tr>
+ <td>F</td>
+ <td><code>float</code></td>
+</tr>
+<tr>
+ <td>D</td>
+ <td><code>double</code></td>
+</tr>
+<tr>
+ <td>L<i>fully/qualified/Name</i>;</td>
+ <td>the class <code><i>fully.qualified.Name</i></code></td>
+</tr>
+<tr>
+ <td>[<i>descriptor</i></td>
+ <td>array of <code><i>descriptor</i></code>, usable recursively for
+ arrays-of-arrays, though it is invalid to have more than 255
+ dimensions.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h1>Items and Related Structures</h1>
+
+<p>This section includes definitions for each of the top-level items that
+may appear in a <code>.dex</code> file.
+
+<h2><code>header_item</code></h2>
+<h4>appears in the <code>header</code> section</h4>
+<h4>alignment: 4 bytes</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>magic</td>
+ <td>ubyte[8] = DEX_FILE_MAGIC</td>
+ <td>magic value. See discussion above under "<code>DEX_FILE_MAGIC</code>"
+ for more details.
+ </td>
+</tr>
+<tr>
+ <td>checksum</td>
+ <td>uint</td>
+ <td>adler32 checksum of the rest of the file (everything but
+ <code>magic</code> and this field); used to detect file corruption
+ </td>
+</tr>
+<tr>
+ <td>signature</td>
+ <td>ubyte[20]</td>
+ <td>SHA-1 signature (hash) of the rest of the file (everything but
+ <code>magic</code>, <code>checksum</code>, and this field); used
+ to uniquely identify files
+ </td>
+</tr>
+<tr>
+ <td>file_size</td>
+ <td>uint</td>
+ <td>size of the entire file (including the header), in bytes
+</tr>
+<tr>
+ <td>header_size</td>
+ <td>uint = 0x70</td>
+ <td>size of the header (this entire section), in bytes. This allows for at
+ least a limited amount of backwards/forwards compatibility without
+ invalidating the format.
+ </td>
+</tr>
+<tr>
+ <td>endian_tag</td>
+ <td>uint = ENDIAN_CONSTANT</td>
+ <td>endianness tag. See discussion above under "<code>ENDIAN_CONSTANT</code>
+ and <code>REVERSE_ENDIAN_CONSTANT</code>" for more details.
+ </td>
+</tr>
+<tr>
+ <td>link_size</td>
+ <td>uint</td>
+ <td>size of the link section, or <code>0</code> if this file isn't
+ statically linked</td>
+</tr>
+<tr>
+ <td>link_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the link section, or
+ <code>0</code> if <code>link_size == 0</code>. The offset, if non-zero,
+ should be to an offset into the <code>link_data</code> section. The
+ format of the data pointed at is left unspecified by this document;
+ this header field (and the previous) are left as hooks for use by
+ runtime implementations.
+ </td>
+</tr>
+<tr>
+ <td>map_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the map item, or
+ <code>0</code> if this file has no map. The offset, if non-zero,
+ should be to an offset into the <code>data</code> section,
+ and the data should be in the format specified by "<code>map_list</code>"
+ below.
+ </td>
+</tr>
+<tr>
+ <td>string_ids_size</td>
+ <td>uint</td>
+ <td>count of strings in the string identifiers list</td>
+</tr>
+<tr>
+ <td>string_ids_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the string identifiers list, or
+ <code>0</code> if <code>string_ids_size == 0</code> (admittedly a
+ strange edge case). The offset, if non-zero,
+ should be to the start of the <code>string_ids</code> section.
+ </td>
+</tr>
+<tr>
+ <td>type_ids_size</td>
+ <td>uint</td>
+ <td>count of elements in the type identifiers list</td>
+</tr>
+<tr>
+ <td>type_ids_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the type identifiers list, or
+ <code>0</code> if <code>type_ids_size == 0</code> (admittedly a
+ strange edge case). The offset, if non-zero,
+ should be to the start of the <code>type_ids</code>
+ section.
+ </td>
+</tr>
+<tr>
+ <td>proto_ids_size</td>
+ <td>uint</td>
+ <td>count of elements in the prototype identifiers list</td>
+</tr>
+<tr>
+ <td>proto_ids_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the prototype identifiers list, or
+ <code>0</code> if <code>proto_ids_size == 0</code> (admittedly a
+ strange edge case). The offset, if non-zero,
+ should be to the start of the <code>proto_ids</code>
+ section.
+ </td>
+</tr>
+<tr>
+ <td>field_ids_size</td>
+ <td>uint</td>
+ <td>count of elements in the field identifiers list</td>
+</tr>
+<tr>
+ <td>field_ids_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the field identifiers list, or
+ <code>0</code> if <code>field_ids_size == 0</code>. The offset, if
+ non-zero, should be to the start of the <code>field_ids</code>
+ section.</td>
+</td>
+</tr>
+<tr>
+ <td>method_ids_size</td>
+ <td>uint</td>
+ <td>count of elements in the method identifiers list</td>
+</tr>
+<tr>
+ <td>method_ids_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the method identifiers list, or
+ <code>0</code> if <code>method_ids_size == 0</code>. The offset, if
+ non-zero, should be to the start of the <code>method_ids</code>
+ section.</td>
+</tr>
+<tr>
+ <td>class_defs_size</td>
+ <td>uint</td>
+ <td>count of elements in the class definitions list</td>
+</tr>
+<tr>
+ <td>class_defs_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the class definitions list, or
+ <code>0</code> if <code>class_defs_size == 0</code> (admittedly a
+ strange edge case). The offset, if non-zero,
+ should be to the start of the <code>class_defs</code> section.
+ </td>
+</tr>
+<tr>
+ <td>data_size</td>
+ <td>uint</td>
+ <td>Size of <code>data</code> section in bytes. Must be an even
+ multiple of sizeof(uint).</td>
+</tr>
+<tr>
+ <td>data_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the start of the
+ <code>data</code> section.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>map_list</code></h2>
+<h4>appears in the <code>data</code> section</h4>
+<h4>referenced from <code>header_item</code></h4>
+<h4>alignment: 4 bytes</h4>
+
+<p>This is a list of the entire contents of a file, in order. It
+contains some redundancy with respect to the <code>header_item</code>
+but is intended to be an easy form to use to iterate over an entire
+file. A given type may appear at most once in a map, but there is no
+restriction on what order types may appear in, other than the
+restrictions implied by the rest of the format (e.g., a
+<code>header</code> section must appear first, followed by a
+<code>string_ids</code> section, etc.). Additionally, the map entries must
+be ordered by initial offset and must not overlap.</p>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>size</td>
+ <td>uint</td>
+ <td>size of the list, in entries</td>
+</tr>
+<tr>
+ <td>list</td>
+ <td>map_item[size]</td>
+ <td>elements of the list</td>
+</tr>
+</tbody>
+</table>
+
+<h3><code>map_item</code> Format</h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>type</td>
+ <td>ushort</td>
+ <td>type of the items; see table below</td>
+</tr>
+<tr>
+ <td>unused</td>
+ <td>ushort</td>
+ <td><i>(unused)</i></td>
+</tr>
+<tr>
+ <td>size</td>
+ <td>uint</td>
+ <td>count of the number of items to be found at the indicated offset</td>
+</tr>
+<tr>
+ <td>offset</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the items in question</td>
+</tr>
+</tbody>
+</table>
+
+
+<h3>Type Codes</h3>
+
+<table class="typeCodes">
+<thead>
+<tr>
+ <th>Item Type</th>
+ <th>Constant</th>
+ <th>Value</th>
+ <th>Item Size In Bytes</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>header_item</td>
+ <td>TYPE_HEADER_ITEM</td>
+ <td>0x0000</td>
+ <td>0x70</td>
+</tr>
+<tr>
+ <td>string_id_item</td>
+ <td>TYPE_STRING_ID_ITEM</td>
+ <td>0x0001</td>
+ <td>0x04</td>
+</tr>
+<tr>
+ <td>type_id_item</td>
+ <td>TYPE_TYPE_ID_ITEM</td>
+ <td>0x0002</td>
+ <td>0x04</td>
+</tr>
+<tr>
+ <td>proto_id_item</td>
+ <td>TYPE_PROTO_ID_ITEM</td>
+ <td>0x0003</td>
+ <td>0x0c</td>
+</tr>
+<tr>
+ <td>field_id_item</td>
+ <td>TYPE_FIELD_ID_ITEM</td>
+ <td>0x0004</td>
+ <td>0x08</td>
+</tr>
+<tr>
+ <td>method_id_item</td>
+ <td>TYPE_METHOD_ID_ITEM</td>
+ <td>0x0005</td>
+ <td>0x08</td>
+</tr>
+<tr>
+ <td>class_def_item</td>
+ <td>TYPE_CLASS_DEF_ITEM</td>
+ <td>0x0006</td>
+ <td>0x20</td>
+</tr>
+<tr>
+ <td>map_list</td>
+ <td>TYPE_MAP_LIST</td>
+ <td>0x1000</td>
+ <td>4 + (item.size * 12)</td>
+</tr>
+<tr>
+ <td>type_list</td>
+ <td>TYPE_TYPE_LIST</td>
+ <td>0x1001</td>
+ <td>4 + (item.size * 2)</td>
+</tr>
+<tr>
+ <td>annotation_set_ref_list</td>
+ <td>TYPE_ANNOTATION_SET_REF_LIST</td>
+ <td>0x1002</td>
+ <td>4 + (item.size * 4)</td>
+</tr>
+<tr>
+ <td>annotation_set_item</td>
+ <td>TYPE_ANNOTATION_SET_ITEM</td>
+ <td>0x1003</td>
+ <td>4 + (item.size * 4)</td>
+</tr>
+<tr>
+ <td>class_data_item</td>
+ <td>TYPE_CLASS_DATA_ITEM</td>
+ <td>0x2000</td>
+ <td><i>implicit; must parse</i></td>
+</tr>
+<tr>
+ <td>code_item</td>
+ <td>TYPE_CODE_ITEM</td>
+ <td>0x2001</td>
+ <td><i>implicit; must parse</i></td>
+</tr>
+<tr>
+ <td>string_data_item</td>
+ <td>TYPE_STRING_DATA_ITEM</td>
+ <td>0x2002</td>
+ <td><i>implicit; must parse</i></td>
+</tr>
+<tr>
+ <td>debug_info_item</td>
+ <td>TYPE_DEBUG_INFO_ITEM</td>
+ <td>0x2003</td>
+ <td><i>implicit; must parse</i></td>
+</tr>
+<tr>
+ <td>annotation_item</td>
+ <td>TYPE_ANNOTATION_ITEM</td>
+ <td>0x2004</td>
+ <td><i>implicit; must parse</i></td>
+</tr>
+<tr>
+ <td>encoded_array_item</td>
+ <td>TYPE_ENCODED_ARRAY_ITEM</td>
+ <td>0x2005</td>
+ <td><i>implicit; must parse</i></td>
+</tr>
+<tr>
+ <td>annotations_directory_item</td>
+ <td>TYPE_ANNOTATIONS_DIRECTORY_ITEM</td>
+ <td>0x2006</td>
+ <td><i>implicit; must parse</i></td>
+</tr>
+</tbody>
+</table>
+
+
+<h2><code>string_id_item</code></h2>
+<h4>appears in the <code>string_ids</code> section</h4>
+<h4>alignment: 4 bytes</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>string_data_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the string data for this
+ item. The offset should be to a location
+ in the <code>data</code> section, and the data should be in the
+ format specified by "<code>string_data_item</code>" below.
+ There is no alignment requirement for the offset.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>string_data_item</code></h2>
+<h4>appears in the <code>data</code> section</h4>
+<h4>alignment: none (byte-aligned)</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>utf16_size</td>
+ <td>uleb128</td>
+ <td>size of this string, in UTF-16 code units (which is the "string
+ length" in many systems). That is, this is the decoded length of
+ the string. (The encoded length is implied by the position of
+ the <code>0</code> byte.)</td>
+</tr>
+<tr>
+ <td>data</td>
+ <td>ubyte[]</td>
+ <td>a series of MUTF-8 code units (a.k.a. octets, a.k.a. bytes)
+ followed by a byte of value <code>0</code>. See
+ "MUTF-8 (Modified UTF-8) Encoding" above for details and
+ discussion about the data format.
+ <p><b>Note:</b> It is acceptable to have a string which includes
+ (the encoded form of) UTF-16 surrogate code units (that is,
+ <code>U+d800</code> … <code>U+dfff</code>)
+ either in isolation or out-of-order with respect to the usual
+ encoding of Unicode into UTF-16. It is up to higher-level uses of
+ strings to reject such invalid encodings, if appropriate.</p>
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>type_id_item</code></h2>
+<h4>appears in the <code>type_ids</code> section</h4>
+<h4>alignment: 4 bytes</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>descriptor_idx</td>
+ <td>uint</td>
+ <td>index into the <code>string_ids</code> list for the descriptor
+ string of this type. The string must conform to the syntax for
+ <i>TypeDescriptor</i>, defined above.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>proto_id_item</code></h2>
+<h4>appears in the <code>proto_ids</code> section</h4>
+<h4>alignment: 4 bytes</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>shorty_idx</td>
+ <td>uint</td>
+ <td>index into the <code>string_ids</code> list for the short-form
+ descriptor string of this prototype. The string must conform to the
+ syntax for <i>ShortyDescriptor</i>, defined above, and must correspond
+ to the return type and parameters of this item.
+ </td>
+</tr>
+<tr>
+ <td>return_type_idx</td>
+ <td>uint</td>
+ <td>index into the <code>type_ids</code> list for the return type
+ of this prototype
+ </td>
+</tr>
+<tr>
+ <td>parameters_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the list of parameter types
+ for this prototype, or <code>0</code> if this prototype has no
+ parameters. This offset, if non-zero, should be in the
+ <code>data</code> section, and the data there should be in the
+ format specified by <code>"type_list"</code> below. Additionally, there
+ should be no reference to the type <code>void</code> in the list.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>field_id_item</code></h2>
+<h4>appears in the <code>field_ids</code> section</h4>
+<h4>alignment: 4 bytes</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>class_idx</td>
+ <td>ushort</td>
+ <td>index into the <code>type_ids</code> list for the definer of this
+ field. This must be a class type, and not an array or primitive type.
+ </td>
+</tr>
+<tr>
+ <td>type_idx</td>
+ <td>ushort</td>
+ <td>index into the <code>type_ids</code> list for the type of
+ this field
+ </td>
+</tr>
+<tr>
+ <td>name_idx</td>
+ <td>uint</td>
+ <td>index into the <code>string_ids</code> list for the name of this
+ field. The string must conform to the syntax for <i>MemberName</i>,
+ defined above.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>method_id_item</code></h2>
+<h4>appears in the <code>method_ids</code> section</h4>
+<h4>alignment: 4 bytes</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>class_idx</td>
+ <td>ushort</td>
+ <td>index into the <code>type_ids</code> list for the definer of this
+ method. This must be a class or array type, and not a primitive type.
+ </td>
+</tr>
+<tr>
+ <td>proto_idx</td>
+ <td>ushort</td>
+ <td>index into the <code>proto_ids</code> list for the prototype of
+ this method
+ </td>
+</tr>
+<tr>
+ <td>name_idx</td>
+ <td>uint</td>
+ <td>index into the <code>string_ids</code> list for the name of this
+ method. The string must conform to the syntax for <i>MemberName</i>,
+ defined above.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>class_def_item</code></h2>
+<h4>appears in the <code>class_defs</code> section</h4>
+<h4>alignment: 4 bytes</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>class_idx</td>
+ <td>uint</td>
+ <td>index into the <code>type_ids</code> list for this class.
+ This must be a class type, and not an array or primitive type.
+ </td>
+</tr>
+<tr>
+ <td>access_flags</td>
+ <td>uint</td>
+ <td>access flags for the class (<code>public</code>, <code>final</code>,
+ etc.). See "<code>access_flags</code> Definitions" for details.
+ </td>
+</tr>
+<tr>
+ <td>superclass_idx</td>
+ <td>uint</td>
+ <td>index into the <code>type_ids</code> list for the superclass, or
+ the constant value <code>NO_INDEX</code> if this class has no
+ superclass (i.e., it is a root class such as <code>Object</code>).
+ If present, this must be a class type, and not an array or primitive type.
+ </td>
+</tr>
+<tr>
+ <td>interfaces_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the list of interfaces, or
+ <code>0</code> if there are none. This offset
+ should be in the <code>data</code> section, and the data
+ there should be in the format specified by
+ "<code>type_list</code>" below. Each of the elements of the list
+ must be a class type (not an array or primitive type), and there
+ must not be any duplicates.
+ </td>
+</tr>
+<tr>
+ <td>source_file_idx</td>
+ <td>uint</td>
+ <td>index into the <code>string_ids</code> list for the name of the
+ file containing the original source for (at least most of) this class,
+ or the special value <code>NO_INDEX</code> to represent a lack of
+ this information. The <code>debug_info_item</code> of any given method
+ may override this source file, but the expectation is that most classes
+ will only come from one source file.
+ </td>
+</tr>
+<tr>
+ <td>annotations_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the annotations structure
+ for this class, or <code>0</code> if there are no annotations on
+ this class. This offset, if non-zero, should be in the
+ <code>data</code> section, and the data there should be in
+ the format specified by "<code>annotations_directory_item</code>" below,
+ with all items referring to this class as the definer.
+ </td>
+</tr>
+<tr>
+ <td>class_data_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the associated
+ class data for this item, or <code>0</code> if there is no class
+ data for this class. (This may be the case, for example, if this class
+ is a marker interface.) The offset, if non-zero, should be in the
+ <code>data</code> section, and the data there should be in the
+ format specified by "<code>class_data_item</code>" below, with all
+ items referring to this class as the definer.
+ </td>
+</tr>
+<tr>
+ <td>static_values_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the list of initial
+ values for <code>static</code> fields, or <code>0</code> if there
+ are none (and all <code>static</code> fields are to be initialized with
+ <code>0</code> or <code>null</code>). This offset should be in the
+ <code>data</code> section, and the data there should be in the
+ format specified by "<code>encoded_array_item</code>" below. The size
+ of the array must be no larger than the number of <code>static</code>
+ fields declared by this class, and the elements correspond to the
+ <code>static</code> fields in the same order as declared in the
+ corresponding <code>field_list</code>. The type of each array
+ element must match the declared type of its corresponding field.
+ If there are fewer elements in the array than there are
+ <code>static</code> fields, then the leftover fields are initialized
+ with a type-appropriate <code>0</code> or <code>null</code>.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>class_data_item</code></h2>
+<h4>referenced from <code>class_def_item</code></h4>
+<h4>appears in the <code>data</code> section</h4>
+<h4>alignment: none (byte-aligned)</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>static_fields_size</td>
+ <td>uleb128</td>
+ <td>the number of static fields defined in this item</td>
+</tr>
+<tr>
+ <td>instance_fields_size</td>
+ <td>uleb128</td>
+ <td>the number of instance fields defined in this item</td>
+</tr>
+<tr>
+ <td>direct_methods_size</td>
+ <td>uleb128</td>
+ <td>the number of direct methods defined in this item</td>
+</tr>
+<tr>
+ <td>virtual_methods_size</td>
+ <td>uleb128</td>
+ <td>the number of virtual methods defined in this item</td>
+</tr>
+<tr>
+ <td>static_fields</td>
+ <td>encoded_field[static_fields_size]</td>
+ <td>the defined static fields, represented as a sequence of
+ encoded elements. The fields must be sorted by
+ <code>field_idx</code> in increasing order.
+ </td>
+</tr>
+<tr>
+ <td>instance_fields</td>
+ <td>encoded_field[instance_fields_size]</td>
+ <td>the defined instance fields, represented as a sequence of
+ encoded elements. The fields must be sorted by
+ <code>field_idx</code> in increasing order.
+ </td>
+</tr>
+<tr>
+ <td>direct_methods</td>
+ <td>encoded_method[direct_methods_size]</td>
+ <td>the defined direct (any of <code>static</code>, <code>private</code>,
+ or constructor) methods, represented as a sequence of
+ encoded elements. The methods must be sorted by
+ <code>method_idx</code> in increasing order.
+ </td>
+</tr>
+<tr>
+ <td>virtual_methods</td>
+ <td>encoded_method[virtual_methods_size]</td>
+ <td>the defined virtual (none of <code>static</code>, <code>private</code>,
+ or constructor) methods, represented as a sequence of
+ encoded elements. This list should <i>not</i> include inherited
+ methods unless overridden by the class that this item represents. The
+ methods must be sorted by <code>method_idx</code> in increasing order.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<p><b>Note:</b> All elements' <code>field_id</code>s and
+<code>method_id</code>s must refer to the same defining class.</p>
+
+<h3><code>encoded_field</code> Format</h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>field_idx_diff</td>
+ <td>uleb128</td>
+ <td>index into the <code>field_ids</code> list for the identity of this
+ field (includes the name and descriptor), represented as a difference
+ from the index of previous element in the list. The index of the
+ first element in a list is represented directly.
+ </td>
+</tr>
+<tr>
+ <td>access_flags</td>
+ <td>uleb128</td>
+ <td>access flags for the field (<code>public</code>, <code>final</code>,
+ etc.). See "<code>access_flags</code> Definitions" for details.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h3><code>encoded_method</code> Format</h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>method_idx_diff</td>
+ <td>uleb128</td>
+ <td>index into the <code>method_ids</code> list for the identity of this
+ method (includes the name and descriptor), represented as a difference
+ from the index of previous element in the list. The index of the
+ first element in a list is represented directly.
+ </td>
+</tr>
+<tr>
+ <td>access_flags</td>
+ <td>uleb128</td>
+ <td>access flags for the method (<code>public</code>, <code>final</code>,
+ etc.). See "<code>access_flags</code> Definitions" for details.
+ </td>
+</tr>
+<tr>
+ <td>code_off</td>
+ <td>uleb128</td>
+ <td>offset from the start of the file to the code structure for this
+ method, or <code>0</code> if this method is either <code>abstract</code>
+ or <code>native</code>. The offset should be to a location in the
+ <code>data</code> section. The format of the data is specified by
+ "<code>code_item</code>" below.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>type_list</code></h2>
+<h4>referenced from <code>class_def_item</code> and
+<code>proto_id_item</code></h4>
+<h4>appears in the <code>data</code> section</h4>
+<h4>alignment: 4 bytes</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>size</td>
+ <td>uint</td>
+ <td>size of the list, in entries</td>
+</tr>
+<tr>
+ <td>list</td>
+ <td>type_item[size]</td>
+ <td>elements of the list</td>
+</tr>
+</tbody>
+</table>
+
+<h3><code>type_item</code> Format</h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>type_idx</td>
+ <td>ushort</td>
+ <td>index into the <code>type_ids</code> list</td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>code_item</code></h2>
+<h4>referenced from <code>method_item</code></h4>
+<h4>appears in the <code>data</code> section</h4>
+<h4>alignment: 4 bytes</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>registers_size</td>
+ <td>ushort</td>
+ <td>the number of registers used by this code</td>
+</tr>
+<tr>
+ <td>ins_size</td>
+ <td>ushort</td>
+ <td>the number of words of incoming arguments to the method that this
+ code is for</td>
+</tr>
+<tr>
+ <td>outs_size</td>
+ <td>ushort</td>
+ <td>the number of words of outgoing argument space required by this
+ code for method invocation
+ </td>
+</tr>
+<tr>
+ <td>tries_size</td>
+ <td>ushort</td>
+ <td>the number of <code>try_item</code>s for this instance. If non-zero,
+ then these appear as the <code>tries</code> array just after the
+ <code>insns</code> in this instance.
+ </td>
+</tr>
+<tr>
+ <td>debug_info_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the debug info (line numbers +
+ local variable info) sequence for this code, or <code>0</code> if
+ there simply is no information. The offset, if non-zero, should be
+ to a location in the <code>data</code> section. The format of
+ the data is specified by "<code>debug_info_item</code>" below.
+ </td>
+</tr>
+<tr>
+ <td>insns_size</td>
+ <td>uint</td>
+ <td>size of the instructions list, in 16-bit code units</td>
+</tr>
+<tr>
+ <td>insns</td>
+ <td>ushort[insns_size]</td>
+ <td>actual array of bytecode. The format of code in an <code>insns</code>
+ array is specified by the companion document
+ <a href="dalvik-bytecode.html">"Bytecode for the Dalvik VM"</a>. Note
+ that though this is defined as an array of <code>ushort</code>, there
+ are some internal structures that prefer four-byte alignment. Also,
+ if this happens to be in an endian-swapped file, then the swapping is
+ <i>only</i> done on individual <code>ushort</code>s and not on the
+ larger internal structures.
+ </td>
+</tr>
+<tr>
+ <td>padding</td>
+ <td>ushort <i>(optional)</i> = 0</td>
+ <td>two bytes of padding to make <code>tries</code> four-byte aligned.
+ This element is only present if <code>tries_size</code> is non-zero
+ and <code>insns_size</code> is odd.
+ </td>
+</tr>
+<tr>
+ <td>tries</td>
+ <td>try_item[tries_size] <i>(optional)</i></td>
+ <td>array indicating where in the code exceptions may be caught and
+ how to handle them. Elements of the array must be non-overlapping in
+ range and in order from low to high address. This element is only
+ present if <code>tries_size</code> is non-zero.
+ </td>
+</tr>
+<tr>
+ <td>handlers</td>
+ <td>encoded_catch_handler_list <i>(optional)</i></td>
+ <td>bytes representing a list of lists of catch types and associated
+ handler addresses. Each <code>try_item</code> has a byte-wise offset
+ into this structure. This element is only present if
+ <code>tries_size</code> is non-zero.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h3><code>try_item</code> Format </h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>start_addr</td>
+ <td>uint</td>
+ <td>start address of the block of code covered by this entry. The address
+ is a count of 16-bit code units to the start of the first covered
+ instruction.
+ </td>
+</tr>
+<tr>
+ <td>insn_count</td>
+ <td>ushort</td>
+ <td>number of 16-bit code units covered by this entry. The last code
+ unit covered (inclusive) is <code>start_addr + insn_count - 1</code>.
+ </td>
+</tr>
+<tr>
+ <td>handler_off</td>
+ <td>ushort</td>
+ <td>offset in bytes from the start of the associated encoded handler data
+ to the <code>catch_handler_item</code> for this entry
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h3><code>encoded_catch_handler_list</code> Format</h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>size</td>
+ <td>uleb128</td>
+ <td>size of this list, in entries</td>
+</tr>
+<tr>
+ <td>list</td>
+ <td>encoded_catch_handler[handlers_size]</td>
+ <td>actual list of handler lists, represented directly (not as offsets),
+ and concatenated sequentially</td>
+</tr>
+</tbody>
+</table>
+
+<h3><code>encoded_catch_handler</code> Format</h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>size</td>
+ <td>sleb128</td>
+ <td>number of catch types in this list. If non-positive, then this is
+ the negative of the number of catch types, and the catches are followed
+ by a catch-all handler. For example: A <code>size</code> of <code>0</code>
+ means that there is a catch-all but no explicitly typed catches.
+ A <code>size</code> of <code>2</code> means that there are two explicitly
+ typed catches and no catch-all. And a <code>size</code> of <code>-1</code>
+ means that there is one typed catch along with a catch-all.
+ </td>
+</tr>
+<tr>
+ <td>handlers</td>
+ <td>encoded_type_addr_pair[abs(size)]</td>
+ <td>stream of <code>abs(size)</code> encoded items, one for each caught
+ type, in the order that the types should be tested.
+ </td>
+</tr>
+<tr>
+ <td>catch_all_addr</td>
+ <td>uleb128 <i>(optional)</i></td>
+ <td>bytecode address of the catch-all handler. This element is only
+ present if <code>size</code> is non-positive.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h3><code>encoded_type_addr_pair</code> Format</h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>type_idx</td>
+ <td>uleb128</td>
+ <td>index into the <code>type_ids</code> list for the type of the
+ exception to catch
+ </td>
+</tr>
+<tr>
+ <td>addr</td>
+ <td>uleb128</td>
+ <td>bytecode address of the associated exception handler</td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>debug_info_item</code></h2>
+<h4>referenced from <code>code_item</code></h4>
+<h4>appears in the <code>data</code> section</h4>
+<h4>alignment: none (byte-aligned)</h4>
+
+<p>Each <code>debug_info_item</code> defines a DWARF3-inspired byte-coded
+state machine that, when interpreted, emits the positions
+table and (potentially) the local variable information for a
+<code>code_item</code>. The sequence begins with a variable-length
+header (the length of which depends on the number of method
+parameters), is followed by the state machine bytecodes, and ends
+with an <code>DBG_END_SEQUENCE</code> byte.</p>
+
+<p>The state machine consists of five registers. The
+<code>address</code> register represents the instruction offset in the
+associated <code>insns_item</code> in 16-bit code units. The
+<code>address</code> register starts at <code>0</code> at the beginning of each
+<code>debug_info</code> sequence and may only monotonically increase.
+The <code>line</code> register represents what source line number
+should be associated with the next positions table entry emitted by
+the state machine. It is initialized in the sequence header, and may
+change in positive or negative directions but must never be less than
+<code>1</code>. The <code>source_file</code> register represents the
+source file that the line number entries refer to. It is initialized to
+the value of <code>source_file_idx</code> in <code>class_def_item</code>.
+The other two variables, <code>prologue_end</code> and
+<code>epilogue_begin</code>, are boolean flags (initialized to
+<code>false</code>) that indicate whether the next position emitted
+should be considered a method prologue or epilogue. The state machine
+must also track the name and type of the last local variable live in
+each register for the <code>DBG_RESTART_LOCAL</code> code.</p>
+
+<p>The header is as follows:</p>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>line_start</td>
+ <td>uleb128</td>
+ <td>the initial value for the state machine's <code>line</code> register.
+ Does not represent an actual positions entry.
+ </td>
+</tr>
+<tr>
+ <td>parameters_size</td>
+ <td>uleb128</td>
+ <td>the number of parameter names that are encoded. There should be
+ one per method parameter, excluding an instance method's <code>this</code>,
+ if any.
+ </td>
+</tr>
+<tr>
+ <td>parameter_names</td>
+ <td>uleb128p1[parameters_size]</td>
+ <td>string index of the method parameter name. An encoded value of
+ <code>NO_INDEX</code> indicates that no name
+ is available for the associated parameter. The type descriptor
+ and signature are implied from the method descriptor and signature.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<p>The byte code values are as follows:</p>
+
+<table class="debugByteCode">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Value</th>
+ <th>Format</th>
+ <th>Arguments</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>DBG_END_SEQUENCE</td>
+ <td>0x00</td>
+ <td></td>
+ <td><i>(none)</i></td>
+ <td>terminates a debug info sequence for a <code>code_item</code></td>
+</tr>
+<tr>
+ <td>DBG_ADVANCE_PC</td>
+ <td>0x01</td>
+ <td>uleb128 addr_diff</td>
+ <td><code>addr_diff</code>: amount to add to address register</td>
+ <td>advances the address register without emitting a positions entry</td>
+</tr>
+<tr>
+ <td>DBG_ADVANCE_LINE</td>
+ <td>0x02</td>
+ <td>sleb128 line_diff</td>
+ <td><code>line_diff</code>: amount to change line register by</td>
+ <td>advances the line register without emitting a positions entry</td>
+</tr>
+<tr>
+ <td>DBG_START_LOCAL</td>
+ <td>0x03</td>
+ <td>uleb128 register_num<br/>
+ uleb128p1 name_idx<br/>
+ uleb128p1 type_idx
+ </td>
+ <td><code>register_num</code>: register that will contain local<br/>
+ <code>name_idx</code>: string index of the name<br/>
+ <code>type_idx</code>: type index of the type
+ </td>
+ <td>introduces a local variable at the current address. Either
+ <code>name_idx</code> or <code>type_idx</code> may be
+ <code>NO_INDEX</code> to indicate that that value is unknown.
+ </td>
+</tr>
+<tr>
+ <td>DBG_START_LOCAL_EXTENDED</td>
+ <td>0x04</td>
+ <td>uleb128 register_num<br/>
+ uleb128p1 name_idx<br/>
+ uleb128p1 type_idx<br/>
+ uleb128p1 sig_idx
+ </td>
+ <td><code>register_num</code>: register that will contain local<br/>
+ <code>name_idx</code>: string index of the name<br/>
+ <code>type_idx</code>: type index of the type<br/>
+ <code>sig_idx</code>: string index of the type signature
+ </td>
+ <td>introduces a local with a type signature at the current address.
+ Any of <code>name_idx</code>, <code>type_idx</code>, or
+ <code>sig_idx</code> may be <code>NO_INDEX</code>
+ to indicate that that value is unknown. (If <code>sig_idx</code> is
+ <code>-1</code>, though, the same data could be represented more
+ efficiently using the opcode <code>DBG_START_LOCAL</code>.)
+ <p><b>Note:</b> See the discussion under
+ "<code>dalvik.annotation.Signature</code>" below for caveats about
+ handling signatures.</p>
+ </td>
+</tr>
+<tr>
+ <td>DBG_END_LOCAL</td>
+ <td>0x05</td>
+ <td>uleb128 register_num</td>
+ <td><code>register_num</code>: register that contained local</td>
+ <td>marks a currently-live local variable as out of scope at the current
+ address
+ </td>
+</tr>
+<tr>
+ <td>DBG_RESTART_LOCAL</td>
+ <td>0x06</td>
+ <td>uleb128 register_num</td>
+ <td><code>register_num</code>: register to restart</td>
+ <td>re-introduces a local variable at the current address. The name
+ and type are the same as the last local that was live in the specified
+ register.
+ </td>
+</tr>
+<tr>
+ <td>DBG_SET_PROLOGUE_END</td>
+ <td>0x07</td>
+ <td></td>
+ <td><i>(none)</i></td>
+ <td>sets the <code>prologue_end</code> state machine register,
+ indicating that the next position entry that is added should be
+ considered the end of a method prologue (an appropriate place for
+ a method breakpoint). The <code>prologue_end</code> register is
+ cleared by any special (<code>>= 0x0a</code>) opcode.
+ </td>
+</tr>
+<tr>
+ <td>DBG_SET_EPILOGUE_BEGIN</td>
+ <td>0x08</td>
+ <td></td>
+ <td><i>(none)</i></td>
+ <td>sets the <code>epilogue_begin</code> state machine register,
+ indicating that the next position entry that is added should be
+ considered the beginning of a method epilogue (an appropriate place
+ to suspend execution before method exit).
+ The <code>epilogue_begin</code> register is cleared by any special
+ (<code>>= 0x0a</code>) opcode.
+ </td>
+</tr>
+<tr>
+ <td>DBG_SET_FILE</td>
+ <td>0x09</td>
+ <td>uleb128p1 name_idx</td>
+ <td><code>name_idx</code>: string index of source file name;
+ <code>NO_INDEX</code> if unknown
+ </td>
+ <td>indicates that all subsequent line number entries make reference to this
+ source file name, instead of the default name specified in
+ <code>code_item</code>
+ </td>
+</tr>
+<tr>
+ <td><i>Special Opcodes</i></td>
+ <!-- When updating the range below, make sure to search for other
+ instances of 0x0a in this section. -->
+ <td>0x0a…0xff</td>
+ <td></td>
+ <td><i>(none)</i></td>
+ <td>advances the <code>line</code> and <code>address</code> registers,
+ emits a position entry, and clears <code>prologue_end</code> and
+ <code>epilogue_begin</code>. See below for description.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h3>Special Opcodes</h3>
+
+<p>Opcodes with values between <code>0x0a</code> and <code>0xff</code>
+(inclusive) move both the <code>line</code> and <code>address</code>
+registers by a small amount and then emit a new position table entry.
+The formula for the increments are as follows:</p>
+
+<pre>
+DBG_FIRST_SPECIAL = 0x0a // the smallest special opcode
+DBG_LINE_BASE = -4 // the smallest line number increment
+DBG_LINE_RANGE = 15 // the number of line increments represented
+
+adjusted_opcode = opcode - DBG_FIRST_SPECIAL
+
+line += DBG_LINE_BASE + (adjusted_opcode % DBG_LINE_RANGE)
+address += (adjusted_opcode / DBG_LINE_RANGE)
+</pre>
+
+<h2><code>annotations_directory_item</code></h2>
+<h4>referenced from <code>class_def_item</code></h4>
+<h4>appears in the <code>data</code> section</h4>
+<h4>alignment: 4 bytes</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>class_annotations_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the annotations made directly
+ on the class, or <code>0</code> if the class has no direct annotations.
+ The offset, if non-zero, should be to a location in the
+ <code>data</code> section. The format of the data is specified
+ by "<code>annotation_set_item</code>" below.
+ </td>
+</tr>
+<tr>
+ <td>fields_size</td>
+ <td>uint</td>
+ <td>count of fields annotated by this item</td>
+</tr>
+<tr>
+ <td>annotated_methods_off</td>
+ <td>uint</td>
+ <td>count of methods annotated by this item</td>
+</tr>
+<tr>
+ <td>annotated_parameters_off</td>
+ <td>uint</td>
+ <td>count of method parameter lists annotated by this item</td>
+</tr>
+<tr>
+ <td>field_annotations</td>
+ <td>field_annotation[fields_size] <i>(optional)</i></td>
+ <td>list of associated field annotations. The elements of the list must
+ be sorted in increasing order, by <code>field_idx</code>.
+ </td>
+</tr>
+<tr>
+ <td>method_annotations</td>
+ <td>method_annotation[methods_size] <i>(optional)</i></td>
+ <td>list of associated method annotations. The elements of the list must
+ be sorted in increasing order, by <code>method_idx</code>.
+ </td>
+</tr>
+<tr>
+ <td>parameter_annotations</td>
+ <td>parameter_annotation[parameters_size] <i>(optional)</i></td>
+ <td>list of associated method parameter annotations. The elements of the
+ list must be sorted in increasing order, by <code>method_idx</code>.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<p><b>Note:</b> All elements' <code>field_id</code>s and
+<code>method_id</code>s must refer to the same defining class.</p>
+
+<h3><code>field_annotation</code> Format</h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>field_idx</td>
+ <td>uint</td>
+ <td>index into the <code>field_ids</code> list for the identity of the
+ field being annotated
+ </td>
+</tr>
+<tr>
+ <td>annotations_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the list of annotations for
+ the field. The offset should be to a location in the <code>data</code>
+ section. The format of the data is specified by
+ "<code>annotation_set_item</code>" below.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h3><code>method_annotation</code> Format</h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>method_idx</td>
+ <td>uint</td>
+ <td>index into the <code>method_ids</code> list for the identity of the
+ method being annotated
+ </td>
+</tr>
+<tr>
+ <td>annotations_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the list of annotations for
+ the method. The offset should be to a location in the
+ <code>data</code> section. The format of the data is specified by
+ "<code>annotation_set_item</code>" below.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h3><code>parameter_annotation</code> Format</h2>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>method_idx</td>
+ <td>uint</td>
+ <td>index into the <code>method_ids</code> list for the identity of the
+ method whose parameters are being annotated
+ </td>
+</tr>
+<tr>
+ <td>annotations_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the list of annotations for
+ the method parameters. The offset should be to a location in the
+ <code>data</code> section. The format of the data is specified by
+ "<code>annotation_set_ref_list</code>" below.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>annotation_set_ref_list</code></h2>
+<h4>referenced from <code>parameter_annotations_item</code></h4>
+<h4>appears in the <code>data</code> section</h4>
+<h4>alignment: 4 bytes</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>size</td>
+ <td>uint</td>
+ <td>size of the list, in entries</td>
+</tr>
+<tr>
+ <td>list</td>
+ <td>annotation_set_ref_item[size]</td>
+ <td>elements of the list</td>
+</tr>
+</tbody>
+</table>
+
+<h3><code>annotation_set_ref_item</code> Format</h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>annotations_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to the referenced annotation set
+ or <code>0</code> if there are no annotations for this element.
+ The offset, if non-zero, should be to a location in the <code>data</code>
+ section. The format of the data is specified by
+ "<code>annotation_set_item</code>" below.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>annotation_set_item</code></h2>
+<h4>referenced from <code>annotations_directory_item</code>,
+<code>field_annotations_item</code>,
+<code>method_annotations_item</code>, and
+<code>annotation_set_ref_item</code></h4>
+<h4>appears in the <code>data</code> section</h4>
+<h4>alignment: 4 bytes</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>size</td>
+ <td>uint</td>
+ <td>size of the set, in entries</td>
+</tr>
+<tr>
+ <td>entries</td>
+ <td>annotation_off_item[size]</td>
+ <td>elements of the set. The elements must be sorted in increasing order,
+ by <code>type_idx</code>.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h3><code>annotation_off_item</code> Format</h3>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>annotation_off</td>
+ <td>uint</td>
+ <td>offset from the start of the file to an annotation.
+ The offset should be to a location in the <code>data</code> section,
+ and the format of the data at that location is specified by
+ "<code>annotation_item</code>" below.
+ </td>
+</tr>
+</tbody>
+</table>
+
+
+<h2><code>annotation_item</code></h2>
+<h4>referenced from <code>annotation_set_item</code></h4>
+<h4>appears in the <code>data</code> section</h4>
+<h4>alignment: none (byte-aligned)</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>visibility</td>
+ <td>ubyte</td>
+ <td>intended visibility of this annotation (see below)</td>
+</tr>
+<tr>
+ <td>annotation</td>
+ <td>encoded_annotation</td>
+ <td>encoded annotation contents, in the format described by
+ "<code>encoded_annotation</code> Format" under
+ "<code>encoded_value</code> Encoding" above.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h3>Visibility values</h3>
+
+<p>These are the options for the <code>visibility</code> field in an
+<code>annotation_item</code>:</p>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Value</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>VISIBILITY_BUILD</td>
+ <td>0x00</td>
+ <td>intended only to be visible at build time (e.g., during compilation
+ of other code)
+ </td>
+</tr>
+<tr>
+ <td>VISIBILITY_RUNTIME</td>
+ <td>0x01</td>
+ <td>intended to visible at runtime</td>
+</tr>
+<tr>
+ <td>VISIBILITY_SYSTEM</td>
+ <td>0x02</td>
+ <td>intended to visible at runtime, but only to the underlying system
+ (and not to regular user code)
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>encoded_array_item</code></h2>
+<h4>referenced from <code>class_def_item</code></h4>
+<h4>appears in the <code>data</code> section</h4>
+<h4>alignment: none (byte-aligned)</h4>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>value</td>
+ <td>encoded_array</td>
+ <td>bytes representing the encoded array value, in the format specified
+ by "<code>encoded_array</code> Format" under "<code>encoded_value</code>
+ Encoding" above.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h1>System Annotations</h1>
+
+<p>System annotations are used to represent various pieces of reflective
+information about classes (and methods and fields). This information is
+generally only accessed indirectly by client (non-system) code.</p>
+
+<p>System annotations are represented in <code>.dex</code> files as
+annotations with visibility set to <code>VISIBILITY_SYSTEM</code>.
+
+<h2><code>dalvik.annotation.AnnotationDefault</code></h2>
+<h4>appears on methods in annotation interfaces</h4>
+
+<p>An <code>AnnotationDefault</code> annotation is attached to each
+annotation interface which wishes to indicate default bindings.</p>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>value</td>
+ <td>Annotation</td>
+ <td>the default bindings for this annotation, represented as an annotation
+ of this type. The annotation need not include all names defined by the
+ annotation; missing names simply do not have defaults.
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>dalvik.annotation.EnclosingClass</code></h2>
+<h4>appears on classes</h4>
+
+<p>An <code>EnclosingClass</code> annotation is attached to each class
+which is either defined as a member of another class, per se, or is
+anonymous but not defined within a method body (e.g., a synthetic
+inner class). Every class that has this annotation must also have an
+<code>InnerClass</code> annotation. Additionally, a class may not have
+both an <code>EnclosingClass</code> and an
+<code>EnclosingMethod</code> annotation.</p>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>value</td>
+ <td>Class</td>
+ <td>the class which most closely lexically scopes this class</td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>dalvik.annotation.EnclosingMethod</code></h2>
+<h4>appears on classes</h4>
+
+<p>An <code>EnclosingMethod</code> annotation is attached to each class
+which is defined inside a method body. Every class that has this
+annotation must also have an <code>InnerClass</code> annotation.
+Additionally, a class may not have both an <code>EnclosingClass</code>
+and an <code>EnclosingMethod</code> annotation.</p>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>value</td>
+ <td>Method</td>
+ <td>the method which most closely lexically scopes this class</td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>dalvik.annotation.InnerClass</code></h2>
+<h4>appears on classes</h4>
+
+<p>An <code>InnerClass</code> annotation is attached to each class
+which is defined in the lexical scope of another class's definition.
+Any class which has this annotation must also have <i>either</i> an
+<code>EnclosingClass</code> annotation <i>or</i> an
+<code>EnclosingMethod</code> annotation.</p>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>name</td>
+ <td>String</td>
+ <td>the originally declared simple name of this class (not including any
+ package prefix). If this class is anonymous, then the name is
+ <code>null</code>.
+ </td>
+</tr>
+<tr>
+ <td>accessFlags</td>
+ <td>int</td>
+ <td>the originally declared access flags of the class (which may differ
+ from the effective flags because of a mismatch between the execution
+ models of the source language and target virtual machine)
+ </td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>dalvik.annotation.MemberClasses</code></h2>
+<h4>appears on classes</h4>
+
+<p>A <code>MemberClasses</code> annotation is attached to each class
+which declares member classes. (A member class is a direct inner class
+that has a name.)</p>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>value</td>
+ <td>Class[]</td>
+ <td>array of the member classes</td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>dalvik.annotation.Signature</code></h2>
+<h4>appears on classes, fields, and methods</h4>
+
+<p>A <code>Signature</code> annotation is attached to each class,
+field, or method which is defined in terms of a more complicated type
+than is representable by a <code>type_id_item</code>. The
+<code>.dex</code> format does not define the format for signatures; it
+is merely meant to be able to represent whatever signatures a source
+language requires for successful implementation of that language's
+semantics. As such, signatures are not generally parsed (or verified)
+by virtual machine implementations. The signatures simply get handed
+off to higher-level APIs and tools (such as debuggers). Any use of a
+signature, therefore, should be written so as not to make any
+assumptions about only receiving valid signatures, explicitly guarding
+itself against the possibility of coming across a syntactically
+invalid signature.</p>
+
+<p>Because signature strings tend to have a lot of duplicated content,
+a <code>Signature</code> annotation is defined as an <i>array</i> of
+strings, where duplicated elements naturally refer to the same
+underlying data, and the signature is taken to be the concatenation of
+all the strings in the array. There are no rules about how to pull
+apart a signature into separate strings; that is entirely up to the
+tools that generate <code>.dex</code> files.</p>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>value</td>
+ <td>String[]</td>
+ <td>the signature of this class or member, as an array of strings that
+ is to be concatenated together</td>
+</tr>
+</tbody>
+</table>
+
+<h2><code>dalvik.annotation.Throws</code></h2>
+<h4>appears on methods</h4>
+
+<p>A <code>Throws</code> annotation is attached to each method which is
+declared to throw one or more exception types.</p>
+
+<table class="format">
+<thead>
+<tr>
+ <th>Name</th>
+ <th>Format</th>
+ <th>Description</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>value</td>
+ <td>Class[]</td>
+ <td>the array of exception types thrown</td>
+</tr>
+</tbody>
+</table>
+
+</body>
+</html>
diff --git a/docs/dexopt.html b/docs/dexopt.html
new file mode 100644
index 0000000..7f0b4bc
--- /dev/null
+++ b/docs/dexopt.html
@@ -0,0 +1,326 @@
+<html>
+<head>
+ <title>Dalvik Optimization and Verification</title>
+</head>
+
+<body>
+<h1>Dalvik Optimization and Verification With <i>dexopt</i></h1>
+
+<p>
+The Dalvik virtual machine was designed specifically for the Android
+mobile platform. The target systems have little RAM, store data on slow
+internal flash memory, and generally have the performance characteristics
+of decade-old desktop systems. They also run Linux, which provides
+virtual memory, processes and threads, and UID-based security mechanisms.
+<p>
+The features and limitations caused us to focus on certain goals:
+
+<ul>
+ <li>Class data, notably bytecode, must be shared between multiple
+ processes to minimize total system memory usage.
+ <li>The overhead in launching a new app must be minimized to keep
+ the device responsive.
+ <li>Storing class data in individual files results in a lot of
+ redundancy, especially with respect to strings. To conserve disk
+ space we need to factor this out.
+ <li>Parsing class data fields adds unnecessary overhead during
+ class loading. Accessing data values (e.g. integers and strings)
+ directly as C types is better.
+ <li>Bytecode verification is necessary, but slow, so we want to verify
+ as much as possible outside app execution.
+ <li>Bytecode optimization (quickened instructions, method pruning) is
+ important for speed and battery life.
+ <li>For security reasons, processes may not edit shared code.
+</ul>
+
+<p>
+The typical VM implementation uncompresses individual classes from a
+compressed archive and stores them on the heap. This implies a separate
+copy of each class in every process, and slows application startup because
+the code must be uncompressed (or at least read off disk in many small
+pieces). On the other hand, having the bytecode on the local heap makes
+it easy to rewrite instructions on first use, facilitating a number of
+different optimizations.
+<p>
+The goals led us to make some fundamental decisions:
+
+<ul>
+ <li>Multiple classes are aggregated into a single "DEX" file.
+ <li>DEX files are mapped read-only and shared between processes.
+ <li>Byte ordering and word alignment are adjusted to suit the local
+ system.
+ <li>Bytecode verification is mandatory for all classes, but we want
+ to "pre-verify" whatever we can.
+ <li>Optimizations that require rewriting bytecode must be done ahead
+ of time.
+</ul>
+
+<p>
+The consequences of these decisions are explained in the following sections.
+
+
+<h2>VM Operation</h2>
+
+<p>
+Application code is delivered to the system in a <code>.jar</code>
+or <code>.apk</code> file. These are really just <code>.zip</code>
+archives with some meta-data files added. The Dalvik DEX data file
+is always called <code>classes.dex</code>.
+<p>
+The bytecode cannot be memory-mapped and executed directly from the zip
+file, because the data is compressed and the start of the file is not
+guaranteed to be word-aligned. These problems could be addressed by
+storing <code>classes.dex</code> without compression and padding out the zip
+file, but that would increase the size of the package sent across the
+data network.
+<p>
+We need to extract <code>classes.dex</code> from the zip archive before
+we can use it. While we have the file available, we might as well perform
+some of the other actions (realignment, optimization, verification) described
+earlier. This raises a new question however: who is responsible for doing
+this, and where do we keep the output?
+
+<h3>Preparation</h3>
+
+<p>
+There are at least three different ways to create a "prepared" DEX file,
+sometimes known as "ODEX" (for Optimized DEX):
+<ol>
+ <li>The VM does it "just in time". The output goes into a special
+ <code>dalvik-cache</code> directory. This works on the desktop and
+ engineering-only device builds where the permissions on the
+ <code>dalvik-cache</code> directory are not restricted. On production
+ devices, this is not allowed.
+ <li>The system installer does it when an application is first added.
+ It has the privileges required to write to <code>dalvik-cache</code>.
+ <li>The build system does it ahead of time. The relevant <code>jar</code>
+ / <code>apk</code> files are present, but the <code>classes.dex</code>
+ is stripped out. The optimized DEX is stored next to the original
+ zip archive, not in <code>dalvik-cache</code>, and is part of the
+ system image.
+</ol>
+<p>
+The <code>dalvik-cache</code> directory is more accurately
+<code>$ANDROID_DATA/data/dalvik-cache</code>. The files inside it have
+names derived from the full path of the source DEX. On the device the
+directory is owned by <code>system</code> / <code>system</code>
+and has 0771 permissions, and the optimized DEX files stored there are
+owned by <code>system</code> and the
+application's group, with 0644 permissions. DRM-locked applications will
+use 640 permissions to prevent other user applications from examining them.
+The bottom line is that you can read your own DEX file and those of most
+other applications, but you cannot create, modify, or remove them.
+<p>
+Preparation of the DEX file for the "just in time" and "system installer"
+approaches proceeds in three steps:
+<p>
+First, the dalvik-cache file is created. This must be done in a process
+with appropriate privileges, so for the "system installer" case this is
+done within <code>installd</code>, which runs as root.
+<p>
+Second, the <code>classes.dex</code> entry is extracted from the the zip
+archive. A small amount of space is left at the start of the file for
+the ODEX header.
+<p>
+Third, the file is memory-mapped for easy access and tweaked for use on
+the current system. This includes byte-swapping and structure realigning,
+but no meaningful changes to the DEX file. We also do some basic
+structure checks, such as ensuring that file offsets and data indices
+fall within valid ranges.
+<p>
+The build system uses a hairy process that involves starting the
+emulator, forcing just-in-time optimization of all relevant DEX files,
+and then extracting the results from <code>dalvik-cache</code>. The
+reasons for doing this, rather than using a tool that runs on the desktop,
+will become more apparent when the optimizations are explained.
+<p>
+Once the code is byte-swapped and aligned, we're ready to go. We append
+some pre-computed data, fill in the ODEX header at the start of the file,
+and start executing. (The header is filled in last, so that we don't
+try to use a partial file.) If we're interested in verification and
+optimization, however, we need to insert a step after the initial prep.
+
+<h3>dexopt</h3>
+
+<p>
+We want to verify and optimize all of the classes in the DEX file. The
+easiest and safest way to do this is to load all of the classes into
+the VM and run through them. Anything that fails to load is simply not
+verified or optimized. Unfortunately, this can cause allocation of some
+resources that are difficult to release (e.g. loading of native shared
+libraries), so we don't want to do it in the same virtual machine that
+we're running applications in.
+<p>
+The solution is to invoke a program called <code>dexopt</code>, which
+is really just a back door into the VM. It performs an abbreviated VM
+initialization, loads zero or more DEX files from the bootstrap class
+path, and then sets about verifying and optimizing whatever it can from
+the target DEX. On completion, the process exits, freeing all resources.
+<p>
+It is possible for multiple VMs to want the same DEX file at the same
+time. File locking is used to ensure that dexopt is only run once.
+
+
+<h2>Verification</h2>
+
+<p>
+The bytecode verification process involves scanning through the instructions
+in every method in every class in a DEX file. The goal is to identify
+illegal instruction sequences so that we don't have to check for them at
+run time. Many of the computations involved are also necessary for "exact"
+garbage collection. See
+<a href="verifier.html">Dalvik Bytecode Verifier Notes</a> for more
+information.
+<p>
+For performance reasons, the optimizer (described in the next section)
+assumes that the verifier has run successfully, and makes some potentially
+unsafe assumptions. By default, Dalvik insists upon verifying all classes,
+and only optimizes classes that have been verified. If you want to
+disable the verifier, you can use command-line flags to do so. See also
+<a href="embedded-vm-control.html"> Controlling the Embedded VM</a>
+for instructions on controlling these
+features within the Android application framework.
+<p>
+Reporting of verification failures is a tricky issue. For example,
+calling a package-scope method on a class in a different package is
+illegal and will be caught by the verifier. We don't necessarily want
+to report it during verification though -- we actually want to throw
+an exception when the method call is attempted. Checking the access
+flags on every method call is expensive though. The
+<a href="verifier.html">Dalvik Bytecode Verifier Notes</a> document
+addresses this issue.
+<p>
+Classes that have been verified successfully have a flag set in the ODEX.
+They will not be re-verified when loaded. The Linux access permissions
+are expected to prevent tampering; if you can get around those, installing
+faulty bytecode is far from the easiest line of attack. The ODEX file has
+a 32-bit checksum, but that's chiefly present as a quick check for
+corrupted data.
+
+
+<h2>Optimization</h2>
+
+<p>
+Virtual machine interpreters typically perform certain optimizations the
+first time a piece of code is used. Constant pool references are replaced
+with pointers to internal data structures, operations that always succeed
+or always work a certain way are replaced with simpler forms. Some of
+these require information only available at runtime, others can be inferred
+statically when certain assumptions are made.
+<p>
+The Dalvik optimizer does the following:
+<ul>
+ <li>For virtual method calls, replace the method index with a
+ vtable index.
+ <li>For instance field get/put, replace the field index with
+ a byte offset. Also, merge the boolean / byte / char / short
+ variants into a single 32-bit form (less code in the interpreter
+ means more room in the CPU I-cache).
+ <li>Replace a handful of high-volume calls, like String.length(),
+ with "inline" replacements. This skips the usual method call
+ overhead, directly switching from the interpreter to a native
+ implementation.
+ <li>Prune empty methods. The simplest example is
+ <code>Object.<init></code>, which does nothing, but must be
+ called whenever any object is allocated. The instruction is
+ replaced with a new version that acts as a no-op unless a debugger
+ is attached.
+ <li>Append pre-computed data. For example, the VM wants to have a
+ hash table for lookups on class name. Instead of computing this
+ when the DEX file is loaded, we can compute it now, saving heap
+ space and computation time in every VM where the DEX is loaded.
+</ul>
+
+<p>
+All of the instruction modifications involve replacing the opcode with
+one not defined by the Dalvik specification. This allows us to freely
+mix optimized and unoptimized instructions. The set of optimized
+instructions, and their exact representation, is tied closely to the VM
+version.
+<p>
+Most of the optimizations are obvious "wins". The use of raw indices
+and offsets not only allows us to execute more quickly, we can also
+skip the initial symbolic resolution. Pre-computation eats up
+disk space, and so must be done in moderation.
+<p>
+There are a couple of potential sources of trouble with these
+optimizations. First, vtable indices and byte offsets are subject to
+change if the VM is updated. Second, if a superclass is in a different
+DEX, and that other DEX is updated, we need to ensure that our optimized
+indices and offsets are updated as well. A similar but more subtle
+problem emerges when user-defined class loaders are employed: the class
+we actually call may not be the one we expected to call.
+<p>These problems are addressed with dependency lists and some limitations
+on what can be optimized.
+
+
+<h2>Dependencies and Limitations</h2>
+
+<p>
+The optimized DEX file includes a list of dependencies on other DEX files,
+plus the CRC-32 and modification date from the originating
+<code>classes.dex</code> zip file entry. The dependency list includes the
+full path to the <code>dalvik-cache</code> file, and the file's SHA-1
+signature. The timestamps of files on the device are unreliable and
+not used. The dependency area also includes the VM version number.
+<p>
+An optimized DEX is dependent upon all of the DEX files in the bootstrap
+class path. DEX files that are part of the bootstrap class path depend
+upon the DEX files that appeared earlier. To ensure that nothing outside
+the dependent DEX files is available, <code>dexopt</code> only loads the
+bootstrap classes. References to classes in other DEX files fail, which
+causes class loading and/or verification to fail, and classes with
+external dependencies are simply not optimized.
+<p>
+This means that splitting code out into many separate DEX files has a
+disadvantage: virtual method calls and instance field lookups between
+non-boot DEX files can't be optimized. Because verification is pass/fail
+with class granularity, no method in a class that has any reliance on
+classes in external DEX files can be optimized. This may be a bit
+heavy-handed, but it's the only way to guarantee that nothing breaks
+when individual pieces are updated.
+<p>
+Another negative consequence: any change to a bootstrap DEX will result
+in rejection of all optimized DEX files. This makes it hard to keep
+system updates small.
+<p>
+Despite our caution, there is still a possibility that a class in a DEX
+file loaded by a user-defined class loader could ask for a bootstrap class
+(say, String) and be given a different class with the same name. If a
+class in the DEX file being processed has the same name as a class in the
+bootstrap DEX files, the class will be flagged as ambiguous and references
+to it will not be resolved during verification / optimization. The class
+linking code in the VM does additional checks to plug another hole;
+see the verbose description in the VM sources for details (vm/oo/Class.c).
+<p>
+If one of the dependencies is updated, we need to re-verify and
+re-optimize the DEX file. If we can do a just-in-time <code>dexopt</code>
+invocation, this is easy. If we have to rely on the installer daemon, or
+the DEX was shipped only in ODEX, then the VM has to reject the DEX.
+<p>
+The output of <code>dexopt</code> is byte-swapped and struct-aligned
+for the host, and contains indices and offsets that are highly VM-specific
+(both version-wise and platform-wise). For this reason it's tricky to
+write a version of <code>dexopt</code> that runs on the desktop but
+generates output suitable for a particular device. The safest way to
+invoke it is on the target device, or on an emulator for that device.
+
+
+<h2>Generated DEX</h2>
+
+<p>
+Some languages and frameworks rely on the ability to generate bytecode
+and execute it. The rather heavy <code>dexopt</code> verification and
+optimization model doesn't work well with that.
+<p>
+We intend to support this in a future release, but the exact method is
+to be determined. We may allow individual classes to be added or whole
+DEX files; may allow Java bytecode or Dalvik bytecode in instructions;
+may perform the usual set of optimizations, or use a separate interpreter
+that performs on-first-use optimizations directly on the bytecode (which
+won't be mapped read-only, since it's locally defined).
+
+<address>Copyright © 2008 The Android Open Source Project</address>
+
+</body>
+</html>
diff --git a/docs/embedded-vm-control.html b/docs/embedded-vm-control.html
new file mode 100644
index 0000000..f90f0e5
--- /dev/null
+++ b/docs/embedded-vm-control.html
@@ -0,0 +1,240 @@
+<html>
+<head>
+ <title>Controlling the Embedded VM</title>
+ <link rel=stylesheet href="android.css">
+</head>
+
+<body>
+<h1>Controlling the Embedded VM</h1>
+
+<ul>
+ <li><a href="#overview">Overview</a>
+ <li><a href="#checkjni">Extended JNI Checks</a>
+ <li><a href="#assertions">Assertions</a>
+ <li><a href="#verifier">Bytecode Verification and Optimization</a>
+ <li><a href="#execmode">Execution Mode</a>
+ <li><a href="#dp">Deadlock Prediction</a>
+ <li><a href="#stackdump">Stack Dumps</a>
+</ul>
+
+<h2><a name="overview">Overview</a></h2>
+
+<p>The Dalvik VM supports a variety of command-line arguments
+(use <code>adb shell dalvikvm -help</code> to get a summary), but
+it's not possible to pass arbitrary arguments through the
+Android application runtime. It is, however, possible to affect the
+VM behavior through certain system properties.
+
+<p>For all of the features described below, you would set the system property
+with <code>setprop</code>,
+issuing a shell command on the device like this:
+<pre>adb shell setprop <name> <value></pre>
+
+<p>The Android runtime must be restarted before the changes will take
+effect (<code>adb shell stop; adb shell start</code>). This is because the
+settings are processed in the "zygote" process, which starts early and stays
+around "forever".
+
+<p>You could also add a line to <code>/data/local.prop</code> that looks like:
+<pre><name> = <value></pre>
+
+<p>Such changes will survive reboots, but will be removed by anything
+that wipes the data partition. (Hint: create a <code>local.prop</code>
+on your workstation, then <code>adb push local.prop /data</code> .)
+
+
+<h2><a name="checkjni">Extended JNI Checks</a></h2>
+
+<p>JNI, the Java Native Interface, provides a way for code written in the
+Java programming language
+interact with native (C/C++) code. The extended JNI checks will cause
+the system to run more slowly, but they can spot a variety of nasty bugs
+before they have a chance to cause problems.
+
+<p>There are two system properties that affect this feature, which is
+enabled with the <code>-Xcheck:jni</code> command-line argument. The
+first is <code>ro.kernel.android.checkjni</code>. This is set by the
+Android build system for development builds. (It may also be set by
+the Android emulator unless the <code>-nojni</code> flag is provided on the
+emulator command line.) Because this is an "ro." property, the value cannot
+be changed once the device has started.
+
+<p>To allow toggling of the CheckJNI flag, a second
+property, <code>dalvik.vm.checkjni</code>, is also checked. The value
+of this overrides the value from <code>ro.kernel.android.checkjni</code>.
+
+<p>If neither property is defined, or <code>dalvik.vm.checkjni</code>
+is set to <code>false</code>, the <code>-Xcheck:jni</code> flag is
+not passed in, and JNI checks will be disabled.
+
+<p>To enable JNI checking:
+<pre>adb shell setprop dalvik.vm.checkjni true</pre>
+
+<p>You can also pass JNI-checking options into the VM through a system
+property. The value set for <code>dalvik.vm.jniopts</code> will
+be passed in as the <code>-Xjniopts</code> argument.
+
+<p>For more information about JNI checks, see
+<a href="jni-tips.html">JNI Tips</a>.
+
+
+<h2><a name="assertions">Assertions</a></h2>
+
+<p>Dalvik VM supports the Java programming language "assert" statement.
+By default they are off, but the <code>dalvik.vm.enableassertions</code>
+property provides a way to set the value for a <code>-ea</code> argument.
+
+<p>The argument behaves the same as it does in other desktop VMs. You
+can provide a class name, a package name (followed by "..."), or the
+special value "all".
+
+<p>For example, this:
+<pre>adb shell setprop dalvik.vm.enableassertions all</pre>
+enables assertions in all non-system classes.
+
+<p>The system property is much more limited than the full command line.
+It is not possible to specify more than one <code>-ea</code> entry, and there
+is no way to specify a <code>-da</code> entry. There is presently no
+equivalent for <code>-esa</code>/<code>-dsa</code>.
+
+
+<h2><a name="verifier">Bytecode Verification and Optimization</a></h2>
+
+<p>The system tries to pre-verify all classes in a DEX file to reduce
+class load overhead, and performs a series of optimizations to improve
+runtime performance. Both of these are done by the <code>dexopt</code>
+command, either in the build system or by the installer. On a development
+device, <code>dexopt</code> may be run the first time a DEX file is used
+and whenever it or one of its dependencies is updated ("just-in-time"
+optimization and verification).
+
+<p>There are two command-line flags that control the just-in-time
+verification and optimization,
+<code>-Xverify</code> and <code>-Xdexopt</code>. The Android framework
+configures these based on the <code>dalvik.vm.dexopt-flags</code>
+property.
+
+<p>If you set:
+<pre>adb shell setprop dalvik.vm.dexopt-flags v=a,o=v</pre>
+then the framework will pass <code>-Xverify:all -Xdexopt:verified</code>
+to the VM. This enables verification, and only optimizes classes that
+successfully verified. This is the safest setting, and is the default.
+<p>You could also set <code>dalvik.vm.dexopt-flags</code> to <code>v=n</code>
+to have the framework pass <code>-Xverify:none -Xdexopt:verified</code>
+to disable verification. (We could pass in <code>-Xdexopt:all</code> to
+allow optimization, but that wouldn't necessarily optimize more of the
+code, since classes that fail verification may well be skipped by the
+optimizer for the same reasons.) Classes will not be verified by
+<code>dexopt</code>, and unverified code will be loaded and executed.
+
+<p>Enabling verification will make the <code>dexopt</code> command
+take significantly longer, because the verification process is fairly slow.
+Once the verified and optimized DEX files have been prepared, verification
+incurs no additional overhead except when loading classes that failed
+to pre-verify.
+
+<p>If your DEX files are processed with verification disabled, and you
+later turn the verifier on, application loading will be noticeably
+slower (perhaps 40% or more) as classes are verified on first use.
+
+<p>For best results you should force a re-dexopt of all DEX files when
+this property changes. You can do this with:
+<pre>adb shell "rm /data/dalvik-cache/*"</pre>
+This removes the cached versions of the DEX files. Remember to
+stop and restart the runtime (<code>adb shell stop; adb shell start</code>).
+
+<p>(Previous version of the runtime supported the boolean
+<code>dalvik.vm.verify-bytecode</code> property, but that has been
+superceded by <code>dalvik.vm.dexopt-flags</code>.)</p>
+
+
+<h2><a name="execmode">Execution Mode</a></h2>
+
+<p>The current implementation of the Dalvik VM includes three distinct
+interpreter cores. These are referred to as "fast", "portable", and
+"debug". The "fast" interpreter is optimized for the current
+platform, and might consist of hand-optimized assembly routines. In
+constrast, the "portable" interpreter is written in C and expected to
+run on a broad range of platforms. The "debug" interpreter is a variant
+of "portable" that includes support for profiling and single-stepping.
+
+<p>The VM allows you to choose between "fast" and "portable" with an
+extended form of the <code>-Xint</code> argument. The value of this
+argument can be set through the <code>dalvik.vm.execution-mode</code>
+system property.
+
+<p>To select the "portable" interpreter, you would use:
+<pre>adb shell setprop dalvik.vm.execution-mode int:portable</pre>
+If the property is not specified, the most appropriate interpreter
+will be selected automatically. At some point this mechanism may allow
+selection of other modes, such as JIT compilation.
+
+<p>Not all platforms have an optimized implementation. In such cases,
+the "fast" interpreter is generated as a series of C stubs, and the
+result will be slower than the
+"portable" version. (When we have optimized versions for all popular
+architectures the naming convention will be more accurate.)
+
+<p>If profiling is enabled or a debugger is attached, the VM
+switches to the "debug" interpreter. When profiling ends or the debugger
+disconnects, the original interpreter is resumed. (The "debug" interpreter
+is substantially slower, something to keep in mind when evaluating
+profiling data.)
+
+
+<h2><a name="dp">Deadlock Prediction</a></h2>
+
+<p>If the VM is built with <code>WITH_DEADLOCK_PREDICTION</code>, the deadlock
+predictor can be enabled with the <code>-Xdeadlockpredict</code> argument.
+(The output from <code>dalvikvm -help</code> will tell you if the VM was
+built appropriately -- look for <code>deadlock_prediction</code> on the
+<code>Configured with:</code> line.)
+This feature tells the VM to keep track of the order in which object
+monitor locks are acquired. If the program attempts to acquire a set
+of locks in a different order from what was seen earlier, the VM logs
+a warning and optionally throws an exception.
+
+<p>The command-line argument is set based on the
+<code>dalvik.vm.deadlock-predict</code> property. Valid values are
+<code>off</code> to disable it (default), <code>warn</code> to log the
+problem but continue executing, <code>err</code> to cause a
+<code>dalvik.system.PotentialDeadlockError</code> to be thrown from the
+<code>monitor-enter</code> instruction, and <code>abort</code> to have
+the entire VM abort.
+
+<p>You will usually want to use:
+<pre>adb shell setprop dalvik.vm.deadlock-predict err</pre>
+unless you are keeping an eye on the logs as they scroll by.
+
+<p>Please note that this feature is deadlock prediction, not deadlock
+detection -- in the current implementation, the computations are performed
+after the lock is acquired (this simplifies the code, reducing the
+overhead added to every mutex operation). You can spot a deadlock in a
+hung process by sending a <code>kill -3</code> and examining the stack
+trace written to the log.
+
+<p>This only takes monitors into account. Native mutexes and other resources
+can also be the cause of deadlocks, but will not be detected by this.
+
+
+<h2><a name="stackdump">Stack Dumps</a></h2>
+
+<p>Like other desktop VMs, when the Dalvik VM receives a SIGQUIT
+(Ctrl-\ or <code>kill -3</code>), it dumps stack traces for all threads.
+By default this goes to the Android log, but it can also be written to a file.
+
+<p>The <code>dalvik.vm.stack-trace-file</code> property allows you to
+specify the name of the file where the thread stack traces will be written.
+The file will be created (world writable) if it doesn't exist, and the
+new information will be appended to the end of the file. The filename
+is passed into the VM via the <code>-Xstacktracefile</code> argument.
+
+<p>For example:
+<pre>adb shell setprop dalvik.vm.stack-trace-file /tmp/stack-traces.txt</pre>
+
+<p>If the property is not defined, the VM will write the stack traces to
+the Android log when the signal arrives.
+
+<address>Copyright © 2008 The Android Open Source Project</address>
+
+</body></html>
diff --git a/docs/instruction-formats.css b/docs/instruction-formats.css
new file mode 100644
index 0000000..ee23c5c
--- /dev/null
+++ b/docs/instruction-formats.css
@@ -0,0 +1,129 @@
+h1 {
+ font-family: serif;
+ color: #222266;
+}
+
+h2 {
+ font-family: serif;
+ border-top-style: solid;
+ border-top-width: 2px;
+ border-color: #ccccdd;
+ padding-top: 12px;
+ margin-top: 48px;
+ margin-bottom: 2px;
+ color: #222266;
+}
+
+h3 {
+ font-family: serif;
+ color: #222266;
+}
+
+@media print {
+ table {
+ font-size: 8pt;
+ }
+}
+
+@media screen {
+ table {
+ font-size: 10pt;
+ }
+}
+
+table th {
+ font-family: sans-serif;
+ background: #aaaaff;
+}
+
+table {
+ border-collapse: collapse;
+}
+
+table td {
+ font-family: sans-serif;
+ border-top-style: solid;
+ border-bottom-style: solid;
+ border-width: 1px;
+ border-color: #aaaaff;
+ padding-top: 4px;
+ padding-bottom: 4px;
+ padding-left: 2px;
+ padding-right: 2px;
+ background: #eeeeff;
+}
+
+
+/* the mnemonic guide */
+
+table.letters {
+ margin-top: 24px;
+ margin-bottom: 24px;
+ margin-left: 48px;
+ margin-right: 48px;
+}
+
+table.letters td:first-child {
+ font-family: monospace;
+ width: 10%;
+ text-align: center;
+}
+
+table.letters td:first-child + td {
+ width: 10%;
+ text-align: center;
+}
+
+table.letters td:first-child + td + td {
+ width: 80%;
+}
+
+
+/* the formats, per se */
+
+table.format {
+ background: #aaaaaa;
+ border-collapse: collapse;
+ margin-top: 24px;
+ margin-bottom: 24px;
+ margin-left: 48px;
+ margin-right: 48px;
+}
+
+table.format td {
+ font-family: monospace;
+}
+
+table.format td + td i {
+ font-family: sans-serif;
+}
+
+table.format td sub {
+ font-family: sans-serif;
+}
+
+table.format td sub {
+ font-family: sans-serif;
+ font-style: italic;
+ font-size: 70%
+}
+
+table.format th:first-child {
+ width: 28%;
+}
+
+table.format th:first-child + th {
+ width: 5%;
+}
+
+table.format th:first-child + th + th {
+ width: 45%;
+}
+
+table.format th:first-child + th + th + th {
+ width: 22%;
+}
+
+table.format p {
+ margin-bottom: 0pt;
+}
\ No newline at end of file
diff --git a/docs/instruction-formats.html b/docs/instruction-formats.html
new file mode 100644
index 0000000..d7bf690
--- /dev/null
+++ b/docs/instruction-formats.html
@@ -0,0 +1,430 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>Dalvik VM Instruction Formats</title>
+<link rel=stylesheet href="instruction-formats.css">
+</head>
+
+<body>
+
+<h1>Dalvik VM Instruction Formats</h1>
+<p>Copyright © 2007 The Android Open Source Project
+
+<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 the eight-bit
+opcode within the format. A slashed zero ("<code>Ø</code>") is
+used to indicate that all bits should be zero in the indicated
+position.</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>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.</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. These prelinked values include "<code>vtaboff</code>"
+(vtable offset), "<code>fieldoff</code>" (field offset), and
+"<code>iface</code>" (interface pool index).</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>[B=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>ØØ|<i>op</i></td>
+ <td>10x</td>
+ <td><i><code>op</code></i></td>
+ <td> </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> </td>
+</tr>
+<tr>
+ <td>11n</td>
+ <td><i><code>op</code></i> vA, #+B</td>
+ <td> </td>
+</tr>
+<tr>
+ <td rowspan="2">AA|<i>op</i></td>
+ <td>11x</td>
+ <td><i><code>op</code></i> vAA</td>
+ <td> </td>
+</tr>
+<tr>
+ <td>10t</td>
+ <td><i><code>op</code></i> +AA</td>
+ <td>goto</td>
+</tr>
+<tr>
+ <td>ØØ|<i>op</i> AAAA</td></td>
+ <td>20t</td>
+ <td><i><code>op</code></i> +AAAA</td>
+ <td>goto/16</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> </td>
+</tr>
+<tr>
+ <td>21t</td>
+ <td><i><code>op</code></i> vAA, +BBBB</td>
+ <td> </td>
+</tr>
+<tr>
+ <td>21s</td>
+ <td><i><code>op</code></i> vAA, #+BBBB</td>
+ <td> </td>
+</tr>
+<tr>
+ <td>21h</td>
+ <td><i><code>op</code></i> vAA, #+BBBB0000<br/>
+ <i><code>op</code></i> vAA, #+BBBB000000000000
+ </td>
+ <td> </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> </td>
+</tr>
+<tr>
+ <td>22b</td>
+ <td><i><code>op</code></i> vAA, vBB, #+CC</td>
+ <td> </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> </td>
+</tr>
+<tr>
+ <td>22s</td>
+ <td><i><code>op</code></i> vA, vB, #+CCCC</td>
+ <td> </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>ØØ|<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>ØØ|<i>op</i> AAAA BBBB</td>
+ <td>32x</td>
+ <td><i><code>op</code></i> vAAAA, vBBBB</td>
+ <td> </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> </td>
+</tr>
+<tr>
+ <td>31t</td>
+ <td><i><code>op</code></i> vAA, +BBBBBBBB</td>
+ <td> </td>
+</tr>
+<tr>
+ <td>31c</td>
+ <td><i><code>op</code></i> vAA, string@BBBBBBBB</td>
+ <td>const-string/jumbo</td>
+</tr>
+<tr>
+ <td>B|A|<i>op</i> CCCC G|F|E|D</td>
+ <td>35c</td>
+ <td><i>[<code>B=5</code>] <code>op</code></i> {vD, vE, vF, vG, vA},
+ meth@CCCC<br/>
+ <i>[<code>B=5</code>] <code>op</code></i> {vD, vE, vF, vG, vA},
+ type@CCCC<br/>
+ <i>[<code>B=4</code>] <code>op</code></i> {vD, vE, vF, vG},
+ <i><code>kind</code></i>@CCCC<br/>
+ <i>[<code>B=3</code>] <code>op</code></i> {vD, vE, vF},
+ <i><code>kind</code></i>@CCCC<br/>
+ <i>[<code>B=2</code>] <code>op</code></i> {vD, vE},
+ <i><code>kind</code></i>@CCCC<br/>
+ <i>[<code>B=1</code>] <code>op</code></i> {vD},
+ <i><code>kind</code></i>@CCCC<br/>
+ <i>[<code>B=0</code>] <code>op</code></i> {},
+ <i><code>kind</code></i>@CCCC
+ </td>
+ <td> </td>
+</tr>
+<tr>
+ <td>B|A|<i>op</i> CCCC G|F|E|D</td>
+ <td>35ms</td>
+
+ <td><i>[<code>B=5</code>] <code>op</code></i> {vD, vE, vF, vG, vA},
+ vtaboff@CCCC<br/>
+ <i>[<code>B=4</code>] <code>op</code></i> {vD, vE, vF, vG},
+ vtaboff@CCCC<br/>
+ <i>[<code>B=3</code>] <code>op</code></i> {vD, vE, vF},
+ vtaboff@CCCC<br/>
+ <i>[<code>B=2</code>] <code>op</code></i> {vD, vE},
+ vtaboff@CCCC<br/>
+ <i>[<code>B=1</code>] <code>op</code></i> {vD},
+ vtaboff@CCCC<br/>
+ </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>B|A|<i>op</i> DDCC H|G|F|E</td>
+ <td>35fs</td>
+ <td><i>[<code>B=5</code>] <code>op</code></i> {vE, vF, vG, vH, vA},
+ vtaboff@CC, iface@DD<br/>
+ <i>[<code>B=4</code>] <code>op</code></i> {vE, vF, vG, vH},
+ vtaboff@CC, iface@DD<br/>
+ <i>[<code>B=3</code>] <code>op</code></i> {vE, vF, vG},
+ vtaboff@CC, iface@DD<br/>
+ <i>[<code>B=2</code>] <code>op</code></i> {vE, vF},
+ vtaboff@CC, iface@DD<br/>
+ <i>[<code>B=1</code>] <code>op</code></i> {vE},
+ vtaboff@CC, iface@DD<br/>
+ </td>
+ <td><i>(suggested format for statically linked <code>invoke-interface</code>
+ instructions of format 35c)</i>
+ </td>
+</tr>
+<tr>
+ <td>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> </td>
+</tr>
+<tr>
+ <td>AA|<i>op</i> BBBB CCCC</td>
+ <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>AA|<i>op</i> CCBB DDDD</td>
+ <td>3rfs</td>
+ <td><i><code>op</code></i> {vDDDD .. vNNNN}, vtaboff@BB,
+ iface@CC<br/>
+ <p><i>(where <code>NNNN = DDDD+AA-1</code>, that is <code>A</code>
+ determines the count <code>0..255</code>, and <code>D</code>
+ determines the first register)</i></p>
+ </td>
+ <td><i>(suggested format for statically linked <code>invoke-interface</code>
+ instructions of format <code>3rc</code>)</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>
+
+</body>
+</html>
diff --git a/docs/java-bytecode.css b/docs/java-bytecode.css
new file mode 100644
index 0000000..6075c0d
--- /dev/null
+++ b/docs/java-bytecode.css
@@ -0,0 +1,54 @@
+@media print {
+ table {
+ font-size: 8pt;
+ }
+}
+
+@media screen {
+ table {
+ font-size: 10pt;
+ }
+}
+
+h1 {
+ text-align: center;
+}
+
+table {
+ vertical-align: top;
+ border-collapse: collapse;
+ font-family: sans-serif;
+}
+
+td {
+ vertical-align: top;
+ background: #f8f8f8;
+ border-width: 0;
+}
+
+td.outer {
+ width: 25%;
+ padding: 0;
+}
+
+td.outer table {
+ width: 100%;
+}
+
+td.outer td {
+ border-width: 0;
+ background: #f8f8f8;
+ padding: 1pt;
+ padding-left: 10pt;
+ padding-right: 2pt;
+}
+
+tr.d td {
+ background: #dddddd;
+}
+
+td.outer td + td + td {
+ font-family: monospace;
+ font-weight: bold;
+ padding-right: 5pt;
+}
\ No newline at end of file
diff --git a/docs/java-bytecode.html b/docs/java-bytecode.html
new file mode 100644
index 0000000..691ae54
--- /dev/null
+++ b/docs/java-bytecode.html
@@ -0,0 +1,228 @@
+<!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 At A Glance</title>
+<link rel="stylesheet" href="java-bytecode.css">
+</head>
+
+<body>
+
+<h1>Java Bytecode At A Glance</h1>
+
+<table align="center">
+<tr><td class="outer"><table>
+<tr><td>0x00</td><td>0</td><td>nop</td></tr>
+<tr><td>0x01</td><td>1</td><td>aconst_null</td></tr>
+<tr class="d"><td>0x02</td><td>2</td><td>iconst_m1</td></tr>
+<tr class="d"><td>0x03</td><td>3</td><td>iconst_0</td></tr>
+<tr><td>0x04</td><td>4</td><td>iconst_1</td></tr>
+<tr><td>0x05</td><td>5</td><td>iconst_2</td></tr>
+<tr class="d"><td>0x06</td><td>6</td><td>iconst_3</td></tr>
+<tr class="d"><td>0x07</td><td>7</td><td>iconst_4</td></tr>
+<tr><td>0x08</td><td>8</td><td>iconst_5</td></tr>
+<tr><td>0x09</td><td>9</td><td>lconst_0</td></tr>
+<tr class="d"><td>0x0a</td><td>10</td><td>lconst_1</td></tr>
+<tr class="d"><td>0x0b</td><td>11</td><td>fconst_0</td></tr>
+<tr><td>0x0c</td><td>12</td><td>fconst_1</td></tr>
+<tr><td>0x0d</td><td>13</td><td>fconst_2</td></tr>
+<tr class="d"><td>0x0e</td><td>14</td><td>dconst_0</td></tr>
+<tr class="d"><td>0x0f</td><td>15</td><td>dconst_1</td></tr>
+<tr><td>0x10</td><td>16</td><td>bipush</td></tr>
+<tr><td>0x11</td><td>17</td><td>sipush</td></tr>
+<tr class="d"><td>0x12</td><td>18</td><td>ldc</td></tr>
+<tr class="d"><td>0x13</td><td>19</td><td>ldc_w</td></tr>
+<tr><td>0x14</td><td>20</td><td>ldc2_w</td></tr>
+<tr><td>0x15</td><td>21</td><td>iload</td></tr>
+<tr class="d"><td>0x16</td><td>22</td><td>lload</td></tr>
+<tr class="d"><td>0x17</td><td>23</td><td>fload</td></tr>
+<tr><td>0x18</td><td>24</td><td>dload</td></tr>
+<tr><td>0x19</td><td>25</td><td>aload</td></tr>
+<tr class="d"><td>0x1a</td><td>26</td><td>iload_0</td></tr>
+<tr class="d"><td>0x1b</td><td>27</td><td>iload_1</td></tr>
+<tr><td>0x1c</td><td>28</td><td>iload_2</td></tr>
+<tr><td>0x1d</td><td>29</td><td>iload_3</td></tr>
+<tr class="d"><td>0x1e</td><td>30</td><td>lload_0</td></tr>
+<tr class="d"><td>0x1f</td><td>31</td><td>lload_1</td></tr>
+<tr><td>0x20</td><td>32</td><td>lload_2</td></tr>
+<tr><td>0x21</td><td>33</td><td>lload_3</td></tr>
+<tr class="d"><td>0x22</td><td>34</td><td>fload_0</td></tr>
+<tr class="d"><td>0x23</td><td>35</td><td>fload_1</td></tr>
+<tr><td>0x24</td><td>36</td><td>fload_2</td></tr>
+<tr><td>0x25</td><td>37</td><td>fload_3</td></tr>
+<tr class="d"><td>0x26</td><td>38</td><td>dload_0</td></tr>
+<tr class="d"><td>0x27</td><td>39</td><td>dload_1</td></tr>
+<tr><td>0x28</td><td>40</td><td>dload_2</td></tr>
+<tr><td>0x29</td><td>41</td><td>dload_3</td></tr>
+<tr class="d"><td>0x2a</td><td>42</td><td>aload_0</td></tr>
+<tr class="d"><td>0x2b</td><td>43</td><td>aload_1</td></tr>
+<tr><td>0x2c</td><td>44</td><td>aload_2</td></tr>
+<tr><td>0x2d</td><td>45</td><td>aload_3</td></tr>
+<tr class="d"><td>0x2e</td><td>46</td><td>iaload</td></tr>
+<tr class="d"><td>0x2f</td><td>47</td><td>laload</td></tr>
+<tr><td>0x30</td><td>48</td><td>faload</td></tr>
+<tr><td>0x31</td><td>49</td><td>daload</td></tr>
+<tr class="d"><td>0x32</td><td>50</td><td>aaload</td></tr>
+</table></td>
+<td class="outer"><table>
+<tr><td>0x33</td><td>51</td><td>baload</td></tr>
+<tr><td>0x34</td><td>52</td><td>caload</td></tr>
+<tr class="d"><td>0x35</td><td>53</td><td>saload</td></tr>
+<tr class="d"><td>0x36</td><td>54</td><td>istore</td></tr>
+<tr><td>0x37</td><td>55</td><td>lstore</td></tr>
+<tr><td>0x38</td><td>56</td><td>fstore</td></tr>
+<tr class="d"><td>0x39</td><td>57</td><td>dstore</td></tr>
+<tr class="d"><td>0x3a</td><td>58</td><td>astore</td></tr>
+<tr><td>0x3b</td><td>59</td><td>istore_0</td></tr>
+<tr><td>0x3c</td><td>60</td><td>istore_1</td></tr>
+<tr class="d"><td>0x3d</td><td>61</td><td>istore_2</td></tr>
+<tr class="d"><td>0x3e</td><td>62</td><td>istore_3</td></tr>
+<tr><td>0x3f</td><td>63</td><td>lstore_0</td></tr>
+<tr><td>0x40</td><td>64</td><td>lstore_1</td></tr>
+<tr class="d"><td>0x41</td><td>65</td><td>lstore_2</td></tr>
+<tr class="d"><td>0x42</td><td>66</td><td>lstore_3</td></tr>
+<tr><td>0x43</td><td>67</td><td>fstore_0</td></tr>
+<tr><td>0x44</td><td>68</td><td>fstore_1</td></tr>
+<tr class="d"><td>0x45</td><td>69</td><td>fstore_2</td></tr>
+<tr class="d"><td>0x46</td><td>70</td><td>fstore_3</td></tr>
+<tr><td>0x47</td><td>71</td><td>dstore_0</td></tr>
+<tr><td>0x48</td><td>72</td><td>dstore_1</td></tr>
+<tr class="d"><td>0x49</td><td>73</td><td>dstore_2</td></tr>
+<tr class="d"><td>0x4a</td><td>74</td><td>dstore_3</td></tr>
+<tr><td>0x4b</td><td>75</td><td>astore_0</td></tr>
+<tr><td>0x4c</td><td>76</td><td>astore_1</td></tr>
+<tr class="d"><td>0x4d</td><td>77</td><td>astore_2</td></tr>
+<tr class="d"><td>0x4e</td><td>78</td><td>astore_3</td></tr>
+<tr><td>0x4f</td><td>79</td><td>iastore</td></tr>
+<tr><td>0x50</td><td>80</td><td>lastore</td></tr>
+<tr class="d"><td>0x51</td><td>81</td><td>fastore</td></tr>
+<tr class="d"><td>0x52</td><td>82</td><td>dastore</td></tr>
+<tr><td>0x53</td><td>83</td><td>aastore</td></tr>
+<tr><td>0x54</td><td>84</td><td>bastore</td></tr>
+<tr class="d"><td>0x55</td><td>85</td><td>castore</td></tr>
+<tr class="d"><td>0x56</td><td>86</td><td>sastore</td></tr>
+<tr><td>0x57</td><td>87</td><td>pop</td></tr>
+<tr><td>0x58</td><td>88</td><td>pop2</td></tr>
+<tr class="d"><td>0x59</td><td>89</td><td>dup</td></tr>
+<tr class="d"><td>0x5a</td><td>90</td><td>dup_x1</td></tr>
+<tr><td>0x5b</td><td>91</td><td>dup_x2</td></tr>
+<tr><td>0x5c</td><td>92</td><td>dup2</td></tr>
+<tr class="d"><td>0x5d</td><td>93</td><td>dup2_x1</td></tr>
+<tr class="d"><td>0x5e</td><td>94</td><td>dup2_x2</td></tr>
+<tr><td>0x5f</td><td>95</td><td>swap</td></tr>
+<tr><td>0x60</td><td>96</td><td>iadd</td></tr>
+<tr class="d"><td>0x61</td><td>97</td><td>ladd</td></tr>
+<tr class="d"><td>0x62</td><td>98</td><td>fadd</td></tr>
+<tr><td>0x63</td><td>99</td><td>dadd</td></tr>
+<tr><td>0x64</td><td>100</td><td>isub</td></tr>
+<tr class="d"><td>0x65</td><td>101</td><td>lsub</td></tr>
+</table></td>
+<td class="outer"><table>
+<tr><td>0x66</td><td>102</td><td>fsub</td></tr>
+<tr><td>0x67</td><td>103</td><td>dsub</td></tr>
+<tr class="d"><td>0x68</td><td>104</td><td>imul</td></tr>
+<tr class="d"><td>0x69</td><td>105</td><td>lmul</td></tr>
+<tr><td>0x6a</td><td>106</td><td>fmul</td></tr>
+<tr><td>0x6b</td><td>107</td><td>dmul</td></tr>
+<tr class="d"><td>0x6c</td><td>108</td><td>idiv</td></tr>
+<tr class="d"><td>0x6d</td><td>109</td><td>ldiv</td></tr>
+<tr><td>0x6e</td><td>110</td><td>fdiv</td></tr>
+<tr><td>0x6f</td><td>111</td><td>ddiv</td></tr>
+<tr class="d"><td>0x70</td><td>112</td><td>irem</td></tr>
+<tr class="d"><td>0x71</td><td>113</td><td>lrem</td></tr>
+<tr><td>0x72</td><td>114</td><td>frem</td></tr>
+<tr><td>0x73</td><td>115</td><td>drem</td></tr>
+<tr class="d"><td>0x74</td><td>116</td><td>ineg</td></tr>
+<tr class="d"><td>0x75</td><td>117</td><td>lneg</td></tr>
+<tr><td>0x76</td><td>118</td><td>fneg</td></tr>
+<tr><td>0x77</td><td>119</td><td>dneg</td></tr>
+<tr class="d"><td>0x78</td><td>120</td><td>ishl</td></tr>
+<tr class="d"><td>0x79</td><td>121</td><td>lshl</td></tr>
+<tr><td>0x7a</td><td>122</td><td>ishr</td></tr>
+<tr><td>0x7b</td><td>123</td><td>lshr</td></tr>
+<tr class="d"><td>0x7c</td><td>124</td><td>iushr</td></tr>
+<tr class="d"><td>0x7d</td><td>125</td><td>lushr</td></tr>
+<tr><td>0x7e</td><td>126</td><td>iand</td></tr>
+<tr><td>0x7f</td><td>127</td><td>land</td></tr>
+<tr class="d"><td>0x80</td><td>128</td><td>ior</td></tr>
+<tr class="d"><td>0x81</td><td>129</td><td>lor</td></tr>
+<tr><td>0x82</td><td>130</td><td>ixor</td></tr>
+<tr><td>0x83</td><td>131</td><td>lxor</td></tr>
+<tr class="d"><td>0x84</td><td>132</td><td>iinc</td></tr>
+<tr class="d"><td>0x85</td><td>133</td><td>i2l</td></tr>
+<tr><td>0x86</td><td>134</td><td>i2f</td></tr>
+<tr><td>0x87</td><td>135</td><td>i2d</td></tr>
+<tr class="d"><td>0x88</td><td>136</td><td>l2i</td></tr>
+<tr class="d"><td>0x89</td><td>137</td><td>l2f</td></tr>
+<tr><td>0x8a</td><td>138</td><td>l2d</td></tr>
+<tr><td>0x8b</td><td>139</td><td>f2i</td></tr>
+<tr class="d"><td>0x8c</td><td>140</td><td>f2l</td></tr>
+<tr class="d"><td>0x8d</td><td>141</td><td>f2d</td></tr>
+<tr><td>0x8e</td><td>142</td><td>d2i</td></tr>
+<tr><td>0x8f</td><td>143</td><td>d2l</td></tr>
+<tr class="d"><td>0x90</td><td>144</td><td>d2f</td></tr>
+<tr class="d"><td>0x91</td><td>145</td><td>i2b</td></tr>
+<tr><td>0x92</td><td>146</td><td>i2c</td></tr>
+<tr><td>0x93</td><td>147</td><td>i2s</td></tr>
+<tr class="d"><td>0x94</td><td>148</td><td>lcmp</td></tr>
+<tr class="d"><td>0x95</td><td>149</td><td>fcmpl</td></tr>
+<tr><td>0x96</td><td>150</td><td>fcmpg</td></tr>
+<tr><td>0x97</td><td>151</td><td>dcmpl</td></tr>
+<tr class="d"><td>0x98</td><td>152</td><td>dcmpg</td></tr>
+</table></td>
+<td class="outer"><table>
+<tr><td>0x99</td><td>153</td><td>ifeq</td></tr>
+<tr><td>0x9a</td><td>154</td><td>ifne</td></tr>
+<tr class="d"><td>0x9b</td><td>155</td><td>iflt</td></tr>
+<tr class="d"><td>0x9c</td><td>156</td><td>ifge</td></tr>
+<tr><td>0x9d</td><td>157</td><td>ifgt</td></tr>
+<tr><td>0x9e</td><td>158</td><td>ifle</td></tr>
+<tr class="d"><td>0x9f</td><td>159</td><td>if_icmpeq</td></tr>
+<tr class="d"><td>0xa0</td><td>160</td><td>if_icmpne</td></tr>
+<tr><td>0xa1</td><td>161</td><td>if_icmplt</td></tr>
+<tr><td>0xa2</td><td>162</td><td>if_icmpge</td></tr>
+<tr class="d"><td>0xa3</td><td>163</td><td>if_icmpgt</td></tr>
+<tr class="d"><td>0xa4</td><td>164</td><td>if_icmple</td></tr>
+<tr><td>0xa5</td><td>165</td><td>if_acmpeq</td></tr>
+<tr><td>0xa6</td><td>166</td><td>if_acmpne</td></tr>
+<tr class="d"><td>0xa7</td><td>167</td><td>goto</td></tr>
+<tr class="d"><td>0xa8</td><td>168</td><td>jsr</td></tr>
+<tr><td>0xa9</td><td>169</td><td>ret</td></tr>
+<tr><td>0xaa</td><td>170</td><td>tableswitch</td></tr>
+<tr class="d"><td>0xab</td><td>171</td><td>lookupswitch</td></tr>
+<tr class="d"><td>0xac</td><td>172</td><td>ireturn</td></tr>
+<tr><td>0xad</td><td>173</td><td>lreturn</td></tr>
+<tr><td>0xae</td><td>174</td><td>freturn</td></tr>
+<tr class="d"><td>0xaf</td><td>175</td><td>dreturn</td></tr>
+<tr class="d"><td>0xb0</td><td>176</td><td>areturn</td></tr>
+<tr><td>0xb1</td><td>177</td><td>return</td></tr>
+<tr><td>0xb2</td><td>178</td><td>getstatic</td></tr>
+<tr class="d"><td>0xb3</td><td>179</td><td>putstatic</td></tr>
+<tr class="d"><td>0xb4</td><td>180</td><td>getfield</td></tr>
+<tr><td>0xb5</td><td>181</td><td>putfield</td></tr>
+<tr><td>0xb6</td><td>182</td><td>invokevirtual</td></tr>
+<tr class="d"><td>0xb7</td><td>183</td><td>invokespecial</td></tr>
+<tr class="d"><td>0xb8</td><td>184</td><td>invokestatic</td></tr>
+<tr><td>0xb9</td><td>185</td><td>invokeinterface</td></tr>
+<tr><td>0xba</td><td>186</td><td><i>(unused)</i></td></tr>
+<tr class="d"><td>0xbb</td><td>187</td><td>new</td></tr>
+<tr class="d"><td>0xbc</td><td>188</td><td>newarray</td></tr>
+<tr><td>0xbd</td><td>189</td><td>anewarray</td></tr>
+<tr><td>0xbe</td><td>190</td><td>arraylength</td></tr>
+<tr class="d"><td>0xbf</td><td>191</td><td>athrow</td></tr>
+<tr class="d"><td>0xc0</td><td>192</td><td>checkcast</td></tr>
+<tr><td>0xc1</td><td>193</td><td>instanceof</td></tr>
+<tr><td>0xc2</td><td>194</td><td>monitorenter</td></tr>
+<tr class="d"><td>0xc3</td><td>195</td><td>monitorexit</td></tr>
+<tr class="d"><td>0xc4</td><td>196</td><td>wide</td></tr>
+<tr><td>0xc5</td><td>197</td><td>multianewarray</td></tr>
+<tr><td>0xc6</td><td>198</td><td>ifnull</td></tr>
+<tr class="d"><td>0xc7</td><td>199</td><td>ifnonnull</td></tr>
+<tr class="d"><td>0xc8</td><td>200</td><td>goto_w</td></tr>
+<tr><td>0xc9</td><td>201</td><td>jsr_w</td></tr>
+</table></td></tr>
+</table>
+
+</body>
+</html>
diff --git a/docs/java-constraints.css b/docs/java-constraints.css
new file mode 100644
index 0000000..a315a73
--- /dev/null
+++ b/docs/java-constraints.css
@@ -0,0 +1,59 @@
+h1 {
+ font-family: serif;
+ color: #222266;
+}
+
+h2 {
+ font-family: serif;
+ border-top-style: solid;
+ border-top-width: 2px;
+ border-color: #ccccdd;
+ padding-top: 12px;
+ margin-top: 48px;
+ margin-bottom: 2px;
+ color: #222266;
+}
+
+@media print {
+ table {
+ font-size: 8pt;
+ }
+}
+
+@media screen {
+ table {
+ font-size: 10pt;
+ }
+}
+
+
+/* general for all tables */
+
+table {
+ border-collapse: collapse;
+ margin-top: 24px;
+ margin-bottom: 24px;
+ margin-left: 48px;
+ margin-right: 48px;
+}
+
+table th {
+ font-family: sans-serif;
+ background: #aabbff;
+ text-align: left;
+}
+
+table td {
+ font-family: sans-serif;
+ border-top-style: solid;
+ border-bottom-style: solid;
+ border-width: 1px;
+ border-color: #aaaaff;
+ padding-top: 4px;
+ padding-bottom: 4px;
+ padding-left: 4px;
+ padding-right: 6px;
+ background: #eeeeff;
+ margin-top: 4pt;
+ margin-bottom: 0pt;
+}
diff --git a/docs/java-constraints.html b/docs/java-constraints.html
new file mode 100644
index 0000000..2410a1e
--- /dev/null
+++ b/docs/java-constraints.html
@@ -0,0 +1,1080 @@
+<!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>
diff --git a/docs/jni-tips.html b/docs/jni-tips.html
new file mode 100644
index 0000000..e2c3b85
--- /dev/null
+++ b/docs/jni-tips.html
@@ -0,0 +1,512 @@
+<html>
+ <head>
+ <title>Android JNI Tips</title>
+ <link rel=stylesheet href="android.css">
+ </head>
+
+ <body>
+ <h1><a name="JNI_Tips"></a>Android JNI Tips</h1>
+<p>
+</p><p>
+</p><ul>
+<li> <a href="#What_s_JNI_">What's JNI?</a>
+</li>
+<li> <a href="#JavaVM_and_JNIEnv">JavaVM and JNIEnv</a>
+
+</li>
+<li> <a href="#jclassID_jmethodID_and_jfieldID">jclassID, jmethodID, and jfieldID</a>
+</li>
+<li> <a href="#local_vs_global_references">Local vs. Global References</a>
+</li>
+<li> <a href="#UTF_8_and_UTF_16_strings">UTF-8 and UTF-16 Strings</a>
+</li>
+<li> <a href="#Arrays">Primitive Arrays</a>
+</li>
+<li> <a href="#RegionCalls">Region Calls</a>
+</li>
+<li> <a href="#Exceptions">Exceptions</a>
+</li>
+
+<li> <a href="#Extended_checking">Extended Checking</a>
+</li>
+<li> <a href="#Native_Libraries">Native Libraries</a>
+</li>
+<li> <a href="#64bit">64-bit Considerations</a>
+</li>
+
+<li> <a href="#Unsupported">Unsupported Features</a>
+</ul>
+<p>
+<noautolink>
+</noautolink></p><p>
+</p><h2><a name="What_s_JNI_"> </a> What's JNI? </h2>
+<p>
+
+JNI is the Java Native Interface. It defines a way for code written in the
+Java programming language to interact with native
+code, e.g. functions written in C/C++. It's VM-neutral, has support for loading code from
+dynamic shared libraries, and while cumbersome at times is reasonably efficient.
+</p><p>
+You really should read through the
+<a href="http://java.sun.com/javase/6/docs/technotes/guides/jni/spec/jniTOC.html">JNI spec for J2SE 1.6</a>
+to get a sense for how JNI works and what features are available. Some
+aspects of the interface aren't immediately obvious on
+first reading, so you may find the next few sections handy.
+The more detailed <i>JNI Programmer's Guide and Specification</i> can be found
+<a href="http://java.sun.com/docs/books/jni/html/jniTOC.html">here</a>.
+</p><p>
+</p><p>
+</p><h2><a name="JavaVM_and_JNIEnv"> </a> JavaVM and JNIEnv </h2>
+<p>
+JNI defines two key data structures, "JavaVM" and "JNIEnv". Both of these are essentially
+pointers to pointers to function tables. (In the C++ version, it's a class whose sole member
+is a pointer to a function table.) The JavaVM provides the "invocation interface" functions,
+which allow you to create and destroy the VM. In theory you can have multiple VMs per process,
+but Android's VMs only allow one.
+</p><p>
+The JNIEnv provides most of the JNI functions. Your native functions all receive a JNIEnv as
+the first argument.
+</p><p>
+
+On some VMs, the JNIEnv is used for thread-local storage. For this reason, <strong>you cannot share a JNIEnv between threads</strong>.
+If a piece of code has no other way to get its JNIEnv, you should share
+the JavaVM, and use JavaVM->GetEnv to discover the thread's JNIEnv.
+</p><p>
+The C and C++ declarations of JNIEnv and JavaVM are different. "jni.h" provides different typedefs
+depending on whether it's included into ".c" or ".cpp". For this reason it's a bad idea to
+include JNIEnv arguments in header files included by both languages. (Put another way: if your
+header file requires "#ifdef __cplusplus", you may have to do some extra work if anything in
+that header refers to JNIEnv.)
+</p><p>
+</p><p>
+</p><h2><a name="jclassID_jmethodID_and_jfieldID"> jclassID, jmethodID, and jfieldID </a></h2>
+<p>
+If you want to access an object's field from native code, you would do the following:
+</p><p>
+</p><ul>
+<li> Get the class object reference for the class with <code>FindClass</code>
+</li>
+<li> Get the field ID for the field with <code>GetFieldID</code>
+</li>
+<li> Get the contents of the field with something appropriate, e.g.
+<code>GetIntField</code>
+</li>
+</ul>
+<p>
+Similarly, to call a method, you'd first get a class object reference and then a method ID. The IDs are often just
+pointers to internal VM data structures. Looking them up may require several string
+comparisons, but once you have them the actual call to get the field or invoke the method
+is very quick.
+</p><p>
+If performance is important, it's useful to look the values up once and cache the results
+in your native code. Because we are limiting ourselves to one VM per process, it's reasonable
+to store this data in a static local structure.
+</p><p>
+The class references, field IDs, and method IDs are guaranteed valid until the class is unloaded. Classes
+are only unloaded if all classes associated with a ClassLoader can be garbage collected,
+which is rare but will not be impossible in our system. The jclassID
+is a class reference and <strong>must be protected</strong> with a call
+to <code>NewGlobalRef</code> (see the next section).
+</p><p>
+If you would like to cache the IDs when a class is loaded, and automatically re-cache them
+if the class is ever unloaded and reloaded, the correct way to initialize
+the IDs is to add a piece of code that looks like this to the appropriate class:
+</p><p>
+
+</p><pre> /*
+ * We use a class initializer to allow the native code to cache some
+ * field offsets.
+ */
+
+ /*
+ * A native function that looks up and caches interesting
+ * class/field/method IDs for this class. Returns false on failure.
+ */
+ native private static boolean nativeClassInit();
+
+ /*
+ * Invoke the native initializer when the class is loaded.
+ */
+ static {
+ if (!nativeClassInit())
+ throw new RuntimeException("native init failed");
+ }
+</pre>
+<p>
+Create a nativeClassInit method in your C/C++ code that performs the ID lookups. The code
+will be executed once, when the class is initialized. If the class is ever unloaded and
+then reloaded, it will be executed again. (See the implementation of java.io.FileDescriptor
+for an example in our source tree.)
+</p><p>
+</p><p>
+</p><p>
+</p><h2><a name="local_vs_global_references"> Local vs. Global References </a></h2>
+<p>
+Every object that JNI returns is a "local reference". This means that it's valid for the
+duration of the current native method in the current thread.
+<strong>Even if the object itself continues to live on after the native method returns, the reference is not valid.</strong>
+This applies to all sub-classes of jobject, including jclass and jarray.
+(Dalvik VM will warn you about this when -Xcheck:jni is enabled.)
+</p><p>
+
+If you want to hold on to a reference for a longer period, you must use a "global" reference.
+The <code>NewGlobalRef</code> function takes the local reference as
+an argument and returns a global one:
+
+<p><pre>jobject* localRef = [...];
+jobject* globalRef;
+globalRef = env->NewGlobalRef(localRef);
+</pre>
+
+The global reference is guaranteed to be valid until you call
+<code>DeleteGlobalRef</code>.
+</p><p>
+All JNI methods accept both local and global references as arguments.
+</p><p>
+Programmers are required to "not excessively allocate" local references. In practical terms this means
+that if you're creating large numbers of local references, perhaps while running through an array of
+Objects, you should free them manually with
+<code>DeleteLocalRef</code> instead of letting JNI do it for you. The
+VM is only required to reserve slots for
+16 local references, so if you need more than that you should either delete as you go or use
+<code>EnsureLocalCapacity</code> to reserve more.
+</p><p>
+Note: method and field IDs are just 32-bit identifiers, not object
+references, and should not be passed to <code>NewGlobalRef</code>. The raw data
+pointers returned by functions like <code>GetStringUTFChars</code>
+and <code>GetByteArrayElements</code> are also not objects.
+</p><p>
+One unusual case deserves separate mention. If you attach a native
+thread to the VM with AttachCurrentThread, the code you are running will
+never "return" to the VM until the thread detaches from the VM. Any local
+references you create will have to be deleted manually unless the thread
+is about to exit or detach.
+</p><p>
+</p><p>
+</p><p>
+</p><h2><a name="UTF_8_and_UTF_16_strings"> </a> UTF-8 and UTF-16 Strings </h2>
+<p>
+The Java programming language uses UTF-16. For convenience, JNI provides methods that work with "modified UTF-8" encoding
+as well. (Some VMs use the modified UTF-8 internally to store strings; ours do not.) The
+modified encoding only supports the 8- and 16-bit forms, and stores ASCII NUL values in a 16-bit encoding.
+The nice thing about it is that you can count on having C-style zero-terminated strings,
+suitable for use with standard libc string functions. The down side is that you cannot pass
+arbitrary UTF-8 data into the VM and expect it to work correctly.
+</p><p>
+It's usually best to operate with UTF-16 strings. With our current VMs, the
+<code>GetStringChars</code> method
+does not require a copy, whereas <code>GetStringUTFChars</code> requires a malloc and a UTF conversion. Note that
+<strong>UTF-16 strings are not zero-terminated</strong>, and \u0000 is allowed,
+so you need to hang on to the string length as well as
+the string pointer.
+
+</p><p>
+<strong>Don't forget to Release the strings you Get</strong>. The string functions return <code>jchar*</code> or <code>jbyte*</code>, which
+are pointers to primitive types rather than local references. They are
+guaranteed valid until Release is called, which means they are not
+released when the native method returns.
+</p><p>
+</p><p>
+
+
+</p><h2><a name="Arrays"> </a> Primitive Arrays </h2>
+<p>
+JNI provides functions for accessing the contents of array objects.
+While arrays of objects must be accessed one entry at a time, arrays of
+primitives can be read and written directly as if they were declared in C.
+</p><p>
+To make the interface as efficient as possible without constraining
+the VM implementation,
+the <code>Get<PrimitiveType>ArrayElements</code> family of calls
+allows the VM to either return a pointer to the actual elements, or
+allocate some memory and make a copy. Either way, the raw pointer returned
+is guaranteed to be valid until the corresponding <code>Release</code> call
+is issued (which implies that, if the data wasn't copied, the array object
+will be pinned down and can't be relocated as part of compacting the heap).
+<strong>You must Release every array you Get.</strong> Also, if the Get
+call fails, you must ensure that your code doesn't try to Release a NULL
+pointer later.
+</p><p>
+You can determine whether or not the data was copied by passing in a
+non-NULL pointer for the <code>isCopy</code> argument. This is rarely
+useful.
+</p><p>
+The <code>Release</code> call takes a <code>mode</code> argument that can
+have one of three values. The actions performed by the VM depend upon
+whether it returned a pointer to the actual data or a copy of it:
+<ul>
+ <li><code>0</code>
+ <ul>
+ <li>Actual: the array object is un-pinned.
+ <li>Copy: data is copied back. The buffer with the copy is freed.
+ </ul>
+ <li><code>JNI_COMMIT</code>
+ <ul>
+ <li>Actual: does nothing.
+ <li>Copy: data is copied back. The buffer with the copy
+ <strong>is not freed</strong>.
+ </ul>
+ <li><code>JNI_ABORT</code>
+ <ul>
+ <li>Actual: the array object is un-pinned. Earlier
+ writes are <strong>not</strong> aborted.
+ <li>Copy: the buffer with the copy is freed; any changes to it are lost.
+ </ul>
+</ul>
+</p><p>
+One reason for checking the <code>isCopy</code> flag is to know if
+you need to call <code>Release</code> with <code>JNI_COMMIT</code>
+after making changes to an array -- if you're alternating between making
+changes and executing code that uses the contents of the array, you may be
+able to
+skip the no-op commit. Another possible reason for checking the flag is for
+efficient handling of <code>JNI_ABORT</code>. For example, you might want
+to get an array, modify it in place, pass pieces to other functions, and
+then discard the changes. If you know that JNI is making a new copy for
+you, there's no need to create another "editable" copy. If JNI is passing
+you the original, then you do need to make your own copy.
+</p><p>
+Some have asserted that you can skip the <code>Release</code> call if
+<code>*isCopy</code> is false. This is not the case. If no copy buffer was
+allocated, then the original memory must be pinned down and can't be moved by
+the garbage collector.
+</p><p>
+Also note that the <code>JNI_COMMIT</code> flag does NOT release the array,
+and you will need to call <code>Release</code> again with a different flag
+eventually.
+</p><p>
+</p><p>
+
+
+</p><h2><a name="RegionCalls"> Region Calls </a></h2>
+
+<p>
+There is an alternative to calls like <code>Get<Type>ArrayElements</code>
+and <code>GetStringChars</code> that may be very helpful when all you want
+to do is copy data in or out. Consider the following:
+<pre>
+ jbyte* data = env->GetByteArrayElements(array, NULL);
+ if (data != NULL) {
+ memcpy(buffer, data, len);
+ env->ReleaseByteArrayElements(array, data, JNI_ABORT);
+ }
+</pre>
+<p>
+This grabs the array, copies the first <code>len</code> byte
+elements out of it, and then releases the array. Depending upon the VM
+policies the <code>Get</code> call will either pin or copy the array contents.
+We copy the data (for perhaps a second time), then call Release; in this case
+we use <code>JNI_ABORT</code> so there's no chance of a third copy.
+</p><p>
+We can accomplish the same thing with this:
+<pre>
+ env->GetByteArrayRegion(array, 0, len, buffer);
+</pre>
+</p><p>
+This accomplishes the same thing, with several advantages:
+<ul>
+ <li>Requires one JNI call instead of 3, reducing overhead.
+ <li>Doesn't require pinning or extra data copies.
+ <li>Reduces the risk of programmer error -- no need to match up
+ <code>Get</code> and <code>Release</code> calls.
+</ul>
+</p><p>
+Similarly, you can use the <code>Set<Type>ArrayRegion</code> call
+to copy data into an array, and <code>GetStringRegion</code> or
+<code>GetStringUTFRegion</code> to copy characters out of a
+<code>String</code>.
+
+
+</p><h2><a name="Exceptions"> Exceptions </a></h2>
+<p>
+<strong>You may not call most JNI functions while an exception is pending.</strong>
+Your code is expected to notice the exception (via the function's return value,
+<code>ExceptionCheck()</code>, or <code>ExceptionOccurred()</code>) and return,
+or clear the exception and handle it.
+</p><p>
+The only JNI functions that you are allowed to call while an exception is
+pending are:
+<font size="-1"><ul>
+ <li>DeleteGlobalRef
+ <li>DeleteLocalRef
+ <li>DeleteWeakGlobalRef
+ <li>ExceptionCheck
+ <li>ExceptionClear
+ <li>ExceptionDescribe
+ <li>ExceptionOccurred
+ <li>MonitorExit
+ <li>PopLocalFrame
+ <li>PushLocalFrame
+ <li>Release<PrimitiveType>ArrayElements
+ <li>ReleasePrimitiveArrayCritical
+ <li>ReleaseStringChars
+ <li>ReleaseStringCritical
+ <li>ReleaseStringUTFChars
+</ul></font>
+</p><p>
+Note that exceptions thrown by interpreted code do not "leap over" native code,
+and C++ exceptions thrown by native code are not handled by Dalvik.
+The JNI <code>Throw</code> and <code>ThrowNew</code> instructions just
+set an exception pointer in the current thread. Upon returning to the VM from
+native code, the exception will be noted and handled appropriately.
+</p><p>
+Native code can "catch" an exception by calling <code>ExceptionCheck</code> or
+<code>ExceptionOccurred</code>, and clear it with
+<code>ExceptionClear</code>. As usual,
+discarding exceptions without handling them can lead to problems.
+</p><p>
+There are no built-in functions for manipulating the Throwable object
+itself, so if you want to (say) get the exception string you will need to
+find the Throwable class, look up the method ID for
+<code>getMessage "()Ljava/lang/String;"</code>, invoke it, and if the result
+is non-NULL use <code>GetStringUTFChars</code> to get something you can
+hand to printf or a LOG macro.
+
+</p><p>
+</p><p>
+</p><h2><a name="Extended_checking"> Extended Checking </a></h2>
+<p>
+JNI does very little error checking. Calling <code>SetFieldInt</code>
+on an Object field will succeed, even if the field is marked
+<code>private</code> and <code>final</code>. The
+goal is to minimize the overhead on the assumption that, if you've written it in native code,
+you probably did it for performance reasons.
+</p><p>
+Some VMs support extended checking with the "<code>-Xcheck:jni</code>" flag. If the flag is set, the VM
+puts a different table of functions into the JavaVM and JNIEnv pointers. These functions do
+an extended series of checks before calling the standard implementation.
+
+</p><p>
+Some things that may be verified:
+</p><p>
+</p>
+<ul>
+<li> Check for null pointers where not allowed.
+<li>
+<li> Verify argument type correctness (jclass is a class object,
+jfieldID points to field data, jstring is a java.lang.String).
+</li>
+<li> Field type correctness, e.g. don't store a HashMap in a String field.
+</li>
+<li> Check to see if an exception is pending on calls where pending exceptions are not legal.
+</li>
+<li> Check for calls to inappropriate functions between Critical get/release calls.
+</li>
+<li> Check that JNIEnv structs aren't being shared between threads.
+
+</li>
+<li> Make sure local references aren't used outside their allowed lifespan.
+</li>
+<li> UTF-8 strings contain valid "modified UTF-8" data.
+</li>
+</ul>
+<p>Accessibility of methods and fields (i.e. public vs. private) is not
+checked.
+<p>
+The Dalvik VM supports the <code>-Xcheck:jni</code> flag. For a
+description of how to enable it for Android apps, see
+<a href="embedded-vm-control.html">Controlling the Embedded VM</a>.
+It's currently enabled by default in the Android emulator and on
+"engineering" device builds.
+
+</p><p>
+JNI checks can be modified with the <code>-Xjniopts</code> command-line
+flag. Currently supported values include:
+</p>
+<blockquote><dl>
+<dt>forcecopy
+<dd>When set, any function that can return a copy of the original data
+(array of primitive values, UTF-16 chars) will always do so. The buffers
+are over-allocated and surrounded with a guard pattern to help identify
+code writing outside the buffer, and the contents are erased before the
+storage is freed to trip up code that uses the data after calling Release.
+<dt>warnonly
+<dd>By default, JNI "warnings" cause the VM to abort. With this flag
+it continues on.
+</dl></blockquote>
+
+
+</p><p>
+</p><h2><a name="Native_Libraries"> Native Libraries </a></h2>
+<p>
+You can load native code from shared libraries with the standard
+<code>System.loadLibrary()</code> call. The
+preferred way to get at your native code is:
+</p><p>
+</p><ul>
+<li> Call <code>System.loadLibrary()</code> from a static class initializer. (See the earlier example, where one is used to call nativeClassInit().) The argument is the "undecorated" library name, e.g. to load "libfubar.so" you would pass in "fubar".
+
+</li>
+<li> Provide a native function: <code><strong>jint JNI_OnLoad(JavaVM* vm, void* reserved)</strong></code>
+</li>
+<li>In <code>JNI_OnLoad</code>, register all of your native methods. You
+should declare
+the methods "static" so the names don't take up space in the symbol table
+on the device.
+</li>
+</ul>
+<p>
+The <code>JNI_OnLoad</code> function should look something like this if
+written in C:
+</p><blockquote><pre>jint JNI_OnLoad(JavaVM* vm, void* reserved)
+{
+ JNIEnv* env;
+ if ((*vm)->GetEnv(vm, (void**) &env, JNI_VERSION_1_4) != JNI_OK)
+ return -1;
+
+ /* get class with (*env)->FindClass */
+ /* register methods with (*env)->RegisterNatives */
+
+ return JNI_VERSION_1_4;
+}
+</pre></blockquote>
+</p><p>
+You can also call <code>System.load()</code> with the full path name of the
+shared library. For Android apps, you can get the full path to the
+application's private data storage area from the context object.
+</p><p>
+Dalvik does support "discovery" of native methods that are named in a
+specific way (see <a href="http://java.sun.com/javase/6/docs/technotes/guides/jni/spec/design.html#wp615">
+ the JNI spec</a> for details), but this is a less desirable
+approach. It requires more space in the shared object symbol table,
+loading is slower because it requires string searches through all of the
+loaded shared libraries, and if a method signature is wrong you won't know
+about it until the first time the method is actually used.
+</p><p>
+
+
+</p><h2><a name="64bit"> 64-bit Considerations </a></h2>
+
+<p>
+Android is currently expected to run on 32-bit platforms. In theory it
+could be built for a 64-bit system, but that is not a goal at this time.
+For the most part this isn't something that you will need to worry about
+when interacting with native code,
+but it becomes significant if you plan to store pointers to native
+structures in integer fields in an object. To support architectures
+that use 64-bit pointers, <strong>you need to stash your native pointers in a
+<code>long</code> field rather than an <code>int</code></strong>.
+
+
+</p><h2><a name="Unsupported"> Unsupported Features </a></h2>
+<p>All JNI 1.6 features are supported, with the following exceptions:
+<ul>
+ <li><code>DefineClass</code> is not implemented. Dalvik does not use
+ Java bytecodes or class files, so passing in binary class data
+ doesn't work. Translation facilities may be added in a future
+ version of the VM.</li>
+ <li><code>NewWeakGlobalRef</code> and <code>DeleteWeakGlobalRef</code>
+ are not implemented. The
+ VM supports weak references, but not JNI "weak global" references.
+ These will be supported in a future release.</li>
+ <li><code>GetObjectRefType</code> (new in 1.6) is implemented but not fully
+ functional -- it can't always tell the difference between "local" and
+ "global" references.</li>
+</ul>
+
+</p>
+
+<address>Copyright © 2008 The Android Open Source Project</address>
+
+ </body>
+</html>
diff --git a/docs/libraries.html b/docs/libraries.html
new file mode 100644
index 0000000..e1c3035
--- /dev/null
+++ b/docs/libraries.html
@@ -0,0 +1,165 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
+ "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+
+<title>Dalvik Libraries</title>
+
+<link rel=stylesheet href="dex-format.css">
+<link href="prettify.css" type="text/css" rel="stylesheet" />
+<script type="text/javascript" src="prettify.js"></script>
+
+<style>
+ul.code li {
+ font-family: monospace;
+}
+</style>
+
+</head>
+
+<body onload="prettyPrint()">
+
+<h1 class="title">Dalvik Libraries</h1>
+
+<p>The Dalvik Libraries, also known as the <i>Android core libraries</i>,
+implement general purpose APIs used by code written in the Java programming
+language. While the libraries themselves don't depend on Android, they do form
+the foundation of the Android framework. Android applications use the Dalvik
+libraries both directly and indirectly for data structures, networking,
+concurrency, I/O, and more.</p>
+
+<p>The Dalvik libraries break down into two categories:</p>
+
+<ul>
+ <li><a href="#vm-specific">Dalvik VM-specific libraries</a></li>
+ <li><a href="#interop">Java programming language interoperability
+ libraries</a></li>
+</ul>
+
+<p>Any system claiming to be Android-compatible must implement these libraries.
+Unless otherwise noted, both the signatures and the behavior of such a system
+need to conform to the Android 1.0 reference implementation. Both types of
+conformance will be checked by the upcoming Android Compatibility Test Suite
+(CTS).</p>
+
+<a name="vm-specific"/><h2>Dalvik VM-specific libraries</h2>
+
+<p>The VM-specific libraries enable requesting or modifying VM-specific
+information. Code that uses these classes is only portable across Dalvik-based
+systems. The VM-specific Dalvik packages include:</p>
+
+<ul class="code">
+ <li>dalvik.annotation</li>
+ <li>dalvik.bytecode</li>
+ <li>dalvik.system</li>
+</ul>
+
+<a name="interop"/><h2>Java programming language interoperability libraries</h2>
+
+<p>This category of library provides a familiar environment for programmers
+writing code in the Java programming language. Much of the implementation of
+this code comes from <a href="http://harmony.apache.org/">Apache Harmony</a>.
+Sometimes, we have to change the Harmony code to make it more suitable for the
+memory and CPU-constrained environments targeted by Dalvik. We delineate
+Dalvik-specific changes like so:
+
+<pre class="prettyprint">
+ private static final long serialVersionUID = 8683452581122892189L;
+
+// BEGIN android-added
+ /** zero-element array */
+ private static final Object[] emptyArray = new Object[0];
+// END android-added
+
+ private transient int firstIndex;
+</pre>
+
+<p>If you change existing Harmony code instead of just inserting new code, use
+<code>android-changed</code> instead of <code>android-added</code>. These
+markers help us keep track of our own changes when we pull down updates from
+Harmony.</p>
+
+<p>Packages in this category include:</p>
+
+<ul class="code">
+ <li>java.io</li>
+ <li>java.lang</li>
+ <li>java.lang.annotation</li>
+ <li>java.lang.ref</li>
+ <li>java.lang.reflect</li>
+ <li>java.math</li>
+ <li>java.net</li>
+ <li>java.nio</li>
+ <li>java.nio.channels</li>
+ <li>java.nio.channels.spi</li>
+ <li>java.nio.charset</li>
+ <li>java.nio.charset.spi</li>
+ <li>java.security</li>
+ <li>java.security.acl</li>
+ <li>java.security.cert</li>
+ <li>java.security.interfaces</li>
+ <li>java.security.spec</li>
+ <li>java.sql</li>
+ <li>java.text</li>
+ <li>java.util</li>
+ <li>java.util.concurrent</li>
+ <li>java.util.concurrent.atomic</li>
+ <li>java.util.concurrent.locks</li>
+ <li>java.util.jar</li>
+ <li>java.util.logging</li>
+ <li>java.util.prefs</li>
+ <li>java.util.regex</li>
+ <li>java.util.zip</li>
+ <li>javax.crypto</li>
+ <li>javax.crypto.interfaces</li>
+ <li>javax.crypto.spec</li>
+ <li>javax.net</li>
+ <li>javax.net.ssl</li>
+ <li>javax.security.auth</li>
+ <li>javax.security.auth.callback</li>
+ <li>javax.security.auth.login</li>
+ <li>javax.security.auth.x500</li>
+ <li>javax.security.cert</li>
+ <li>javax.sql</li>
+ <li>javax.xml</li>
+ <li>javax.xml.parsers</li>
+ <li>org.w3c.dom</li>
+ <li>org.xml.sax</li>
+ <li>org.xml.sax.ext</li>
+ <li>org.xml.sax.helpers</li>
+</ul>
+
+<p>We only provide the core functionality of <code>XMLParser</code> and
+<code>DocumentBuilder</code> in the XML packages. Some methods dealing with XML
+schema were left out because we don't provide the corresponding packages.</p>
+
+<p>In addition to the aforementioned packages, we plan to support the following
+packages some time in the future. We currently have an unfinished
+implementation of 2D drawing and image processing.</p>
+
+<ul class="code">
+ <li>java.awt</li>
+ <li>java.awt.color</li>
+ <li>java.awt.event</li>
+ <li>java.awt.font</li>
+ <li>java.awt.geom</li>
+ <li>java.awt.im</li>
+ <li>java.awt.im.spi</li>
+ <li>java.awt.image</li>
+ <li>java.awt.image.renderable</li>
+ <li>javax.imageio</li>
+ <li>javax.imageio.event</li>
+ <li>javax.imageio.metadata</li>
+ <li>javax.imageio.plugins.bmp</li>
+ <li>javax.imageio.plugins.jpeg</li>
+ <li>javax.imageio.spi</li>
+ <li>javax.imageio.stream</li>
+</ul>
+
+<p style="margin-top: 50px">Copyright © 2008 The Android Open Source
+Project</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-00-nop.html b/docs/opcodes/opcode-00-nop.html
new file mode 100644
index 0000000..726f560
--- /dev/null
+++ b/docs/opcodes/opcode-00-nop.html
@@ -0,0 +1,59 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>nop</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>nop</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Waste cycles.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>00 10x</td>
+ <td>nop</td>
+ <td> </td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<p>
+None.
+</p>
+
+<h2>Behavior</h2>
+
+<p>
+No externally observable effects, that is, all registers and object state(s)
+stay the same. The program counter silently advances to the next instruction.
+</p>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-01-move.html b/docs/opcodes/opcode-01-move.html
new file mode 100644
index 0000000..13c1150
--- /dev/null
+++ b/docs/opcodes/opcode-01-move.html
@@ -0,0 +1,92 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>move</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>move</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Move the contents of one non-object register to another.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>01 12x</td>
+ <td>move vA, vB</td>
+ <td><code>A:</code> destination register (4 bits)<br/>
+ <code>B:</code> source register (4 bits)</td>
+</tr>
+<tr>
+ <td>02 22x</td>
+ <td>move/from16 vAA, vBBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> source register (16 bits)</td>
+</tr>
+<tr>
+ <td>03 32x</td>
+ <td>move/16 vAAAA, vBBBB</td>
+ <td><code>A:</code> destination register (16 bits)<br/>
+ <code>B:</code> source register (16 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ Both A and B must be valid register indices in the current stack frame.
+ </li>
+ <li>
+ Register vB must be defined.
+ </li>
+ <li>
+ Register vB must not contain a reference value.
+ </li>
+ <li>
+ Register vB must not be part of a register pair.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The value of register vB is moved to register vA, that is, vA' = vB.
+ </li>
+ <li>
+ If register v(A-1) is the first half of a register pair, register v(A-1)'
+ becomes undefined.
+ </li>
+ <li>
+ If register v(A+1) is the second half of a register pair, register v(A+1)'
+ becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-04-move-wide.html b/docs/opcodes/opcode-04-move-wide.html
new file mode 100644
index 0000000..42202be
--- /dev/null
+++ b/docs/opcodes/opcode-04-move-wide.html
@@ -0,0 +1,106 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>move-wide</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>move</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Move the contents of one register-pair to another.
+</p>
+<p>
+Note: It is legal to move from vN to either vN-1 or vN+1, so implementations
+must arrange for both halves of a register pair to be read before anything is
+written.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>04 12x</td>
+ <td>move-wide vA, vB</td>
+ <td><code>A:</code> destination register pair (4 bits)<br/>
+ <code>B:</code> source register pair (4 bits)</td>
+</tr>
+<tr>
+ <td>05 22x</td>
+ <td>move-wide/from16 vAA, vBBBB</td>
+ <td><code>A:</code> destination register pair (8 bits)<br/>
+ <code>B:</code> source register pair (16 bits)</td>
+</tr>
+<tr>
+ <td>06 32x</td>
+ <td>move-wide/16 vAAAA, vBBBB</td>
+ <td><code>A:</code> destination register pair (16 bits)<br/>
+ <code>B:</code> source register pair (16 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ Both A+1 and B+1 must be valid register indices in the current stackframe
+ (which includes A and B being valid).
+ </li>
+ <li>
+ Register vB must be the lower half of a register pair (which excludes the
+ case of it containing a reference).
+ </li>
+ <li>
+ Both register vB and v(B+1) must be defined.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The value of register vB is moved to register vA, that is, vA' = vB.
+ </li>
+ <li>
+ The value of register v(B+1) is moved to register v(A+1), that is, v(A+1)'
+ = v(B+1).
+ </li>
+ <li>
+ If register v(A-1) is the lower half of a register pair, then v(A-1)'
+ becomes undefined.
+ </li>
+ <li>
+ If register v(A+2) is the upper half of a register pair, then v(A+2)'
+ becomes undefined.
+ </li>
+ <li>
+ If A = B-1, then v(B+1)' becomes undefined.
+ </li>
+ <li>
+ If A = B+1, then v(B)' becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-07-move-object.html b/docs/opcodes/opcode-07-move-object.html
new file mode 100644
index 0000000..e62750d
--- /dev/null
+++ b/docs/opcodes/opcode-07-move-object.html
@@ -0,0 +1,90 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>move-object</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>move-object</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Move the contents of one object-bearing register to another.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>07 12x</td>
+ <td>move-object vA, vB</td>
+ <td><code>A:</code> destination register (4 bits)<br/>
+ <code>B:</code> source register (4 bits)</td>
+</tr>
+<tr>
+ <td>08 22x</td>
+ <td>move-object/from16 vAA, vBBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> source register (16 bits)</td>
+</tr>
+<tr>
+ <td>09 32x</td>
+ <td>move-object/16 vAAAA, vBBBB</td>
+ <td><code>A:</code> destination register (16 bits)<br/>
+ <code>B:</code> source register (16 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ Both A and B must be legal register indices in the current stack frame.
+ </li>
+ <li>
+ Register vB must be defined.
+ </li>
+ <li>
+ Register vB must contain a reference value (which excludes the case of it
+ being part of a register pair).
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The value of register vB is moved to register vA, that is, vA' = vB.
+ </li>
+ <li>
+ If register v(A-1) is the lower half of a register pair, register v(A-1)'
+ becomes undefined.
+ </li>
+ <li>
+ If register v(A+1) is the upper half of a register pair, register v(A+1)'
+ becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-0a-move-result.html b/docs/opcodes/opcode-0a-move-result.html
new file mode 100644
index 0000000..f043d13
--- /dev/null
+++ b/docs/opcodes/opcode-0a-move-result.html
@@ -0,0 +1,96 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>move-result</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>move-result</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Move the single-word non-object result of the most recent invoke-kind into the
+indicated register. This must be done as the instruction immediately after an
+invoke-kind whose (single-word, non-object) result is not to be ignored;
+anywhere else is invalid.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>0a 11x</td>
+ <td>move-result vAA</td>
+ <td><code>A:</code> destination register (8 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ The instruction must be immediately preceded (in the code array) by an
+ invoke-kind instruction.
+ </li>
+ <li>
+ The instruction must be immediately reached (in the actual control flow)
+ through returning from this invoke-kind instruction (it must not be jumped
+ to).
+ </li>
+ <li>
+ The result delivered by the invoke-kind instruction must not be a reference
+ value or require a register pair.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The result delivered by the invoke-kind instruction is moved to register
+ vA, that is, vA' = result.
+ </li>
+ <li>
+ If register v(A-1) is the lower half of a register pair, register v(A-1)'
+ becomes undefined.
+ </li>
+ <li>
+ If register v(A+1) is the upper half of a register pair, register v(A+1)'
+ becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+<h2>Notes</h2>
+
+<p>
+This instruction can also be thought of as reading the contents of a special
+"result" register that is made valid and defined by executing a non-void return
+instruction or a filled-new-array instruction. The execution of any other
+instruction (including this one) renders this special register invalid.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-0b-move-result-wide.html b/docs/opcodes/opcode-0b-move-result-wide.html
new file mode 100644
index 0000000..c3e5779
--- /dev/null
+++ b/docs/opcodes/opcode-0b-move-result-wide.html
@@ -0,0 +1,101 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>move-result-wide</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>move-result-wide</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Move the double-word result of the most recent invoke-kind into the indicated
+register pair. This must be done as the instruction immediately after an
+invoke-kind whose (double-word) result is not to be ignored; anywhere else is
+invalid.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>0b 11x</td>
+ <td>move-result-wide vAA</td>
+ <td><code>A:</code> destination register pair (8 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A+1 must be a valid register index in the current stack frame (which
+ includes A itself being valid).
+ </li>
+ <li>
+ The instruction must be immediately preceded (in the code array) by an
+ invoke-kind instruction.
+ </li>
+ <li>
+ The instruction must be immediately reached (in the actual control flow)
+ through returning from this invoke-kind instruction (it must not be jumped
+ to).
+ </li>
+ <li>
+ The result delivered by the invoke-kind instruction must be either a long
+ or a double value.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The upper 32 bits of the result delivered by the invoke-kind instruction are
+ moved to register vA, that is, vA' = result >> 0x20.
+ </li>
+ <li>
+ The lower 32 bits of the result delivered by the invoke-kind instruction are
+ moved to register v(A+1), that is, v(A+1)' = result & 0xffffffff.
+ </li>
+ <li>
+ If register v(A-1) is the lower half of a register pair, register v(A-1)'
+ becomes undefined.
+ </li>
+ <li>
+ If register v(A+2) is the upper half of a register pair, register v(A+2)'
+ becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+<h2>Notes</h2>
+
+<p>
+This instruction can also be thought of as reading the contents of a special
+"result" register that is made valid and defined by executing a non-void return
+instruction or a filled-new-array instruction. The execution of any other
+instruction (including this one) renders this special register invalid.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-0c-move-result-object.html b/docs/opcodes/opcode-0c-move-result-object.html
new file mode 100644
index 0000000..f065fed
--- /dev/null
+++ b/docs/opcodes/opcode-0c-move-result-object.html
@@ -0,0 +1,97 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>move-result-object</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>move-result-object</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Move the object result of the most recent invoke-kind into the indicated
+register. This must be done as the instruction immediately after an invoke-kind
+or filled-new-array whose (object) result is not to be ignored; anywhere else
+is invalid.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>0c 11x</td>
+ <td>move-result-object vAA</td>
+ <td><code>A:</code> destination register (8 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ The instruction must be immediately preceded (in the code array) by an
+ invoke-kind, filled-new-array, or filled-new-array/range instruction.
+ </li>
+ <li>
+ The instruction must be immediately reached (in the actual control flow)
+ through returning from this invoke-kind instruction or by passing a
+ filled-new-array or filled-new-array/range instruction (it must not be
+ jumped to).
+ </li>
+ <li>
+ The result delivered by the invoke-kind instruction must be a reference
+ value (which excludes the case of a long and double values).
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The result delivered by the invoke-kind instruction is moved to register
+ vA, that, is vA' = result.
+ </li>
+ <li>
+ If register v(A-1) is the lower half of a register pair, register v(A-1)'
+ becomes undefined.
+ </li>
+ <li>
+ If register v(A+1) is the upper half of a register pair, register v(A+1)'
+ becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+<h2>Notes</h2>
+
+<p>
+This instruction can also be thought of as reading the contents of a special
+"result" register that is made valid and defined by executing a non-void return
+instruction or a filled-new-array instruction. The execution of any other
+instruction (including this one) renders this special register invalid.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-0d-move-exception.html b/docs/opcodes/opcode-0d-move-exception.html
new file mode 100644
index 0000000..3c3bc83
--- /dev/null
+++ b/docs/opcodes/opcode-0d-move-exception.html
@@ -0,0 +1,79 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>move-exception</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>move-exception</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Save a just-caught exception into the given register. This should be the first
+instruction of any exception handler whose caught exception is not to be
+ignored, and this instruction may only ever occur as the first instruction of an
+exception handler; anywhere else is invalid.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>0d 11x</td>
+ <td>move-exception vAA</td>
+ <td><code>A:</code> destination register (8 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ The instruction must be the first instruction (in the code array) of an
+ instruction handler, that is, its offset in the code array must match one of
+ the handlers defined for the method in the Dex file.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The active exception of the current thread is moved to register vA, that is,
+ vA' = exception.
+ </li>
+ <li>
+ If register v(A-1) is the lower half of a register pair, register v(A-1)'
+ becomes undefined.
+ </li>
+ <li>
+ If register v(A+1) is the upper half of a register pair, register v(A+1)'
+ becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-0e-return-void.html b/docs/opcodes/opcode-0e-return-void.html
new file mode 100644
index 0000000..7a9431e
--- /dev/null
+++ b/docs/opcodes/opcode-0e-return-void.html
@@ -0,0 +1,84 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>return-void</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>return-void</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Return from a void method.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>0e 10x</td>
+ <td>return-void</td>
+ <td> </td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ The return type of the current method must be void.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ If the method is synchronized, the object's monitor is released in a way
+ similar to the monitor-exit instruction.
+ </li>
+ <li>
+ The stack frame of the current method invocation is removed from the stack.
+ This includes all its registers becoming invalid.
+ </li>
+ <li>
+ If the stack is now empty, the current thread terminates.
+ </li>
+ <li>
+ Otherwise, the following happens:
+ <ul>
+ <li>
+ The stack frame that caused this method invocation becomes valid. This
+ includes all its registers and their old values.
+ </li>
+ <li>
+ Execution continues at the bytecode instruction immediately following
+ the invoke-kind or invoke-kind/range instruction that caused this method
+ invocation.
+ </li>
+ </ul>
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-0f-return.html b/docs/opcodes/opcode-0f-return.html
new file mode 100644
index 0000000..33e0b7c
--- /dev/null
+++ b/docs/opcodes/opcode-0f-return.html
@@ -0,0 +1,100 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>return</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>return</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Return from a single-width (32-bit) non-object value-returning method.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>0f 11x</td>
+ <td>return vAA</td>
+ <td><code>A:</code> return value register (8 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ The return type of the current method must not be double, long, or a
+ reference.
+ </li>
+ <li>
+ A must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ Register vA must not be part of a register pair.
+ </li>
+ <li>
+ The type of vA must match the return type of the method.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ If the method is synchronized, the object's monitor is released in a way
+ similar to the monitor-exit instruction.
+ </li>
+ <li>
+ The stack frame of the current method invocation is removed from the stack.
+ This includes all its registers becoming invalid.
+ </li>
+ <li>
+ If the stack is now empty, the current thread terminates.
+ </li>
+ <li>
+ Otherwise, the following happens:
+ <ul>
+ <li>
+ The stack frame that caused this method invocation becomes valid. This
+ includes all its registers and their old values.
+ </li>
+ <li>
+ Execution continues at the bytecode instruction immediately following
+ the invoke-kind or invoke-kind/range instruction that caused this
+ method invocation.
+ </li>
+ <li>
+ The return value can be consumed by (exactly) the first instruction
+ following the invoke-kind or invoke-kind/range instruction that caused
+ this method invocation, and this instructions needs to be a move-result
+ instruction.
+ </li>
+ </ul>
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-10-return-wide.html b/docs/opcodes/opcode-10-return-wide.html
new file mode 100644
index 0000000..f1a6b83
--- /dev/null
+++ b/docs/opcodes/opcode-10-return-wide.html
@@ -0,0 +1,99 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>return-wide</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>return-wide</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Return from a double-width (64-bit) value-returning method.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>10 11x</td>
+ <td>return-wide vAA</td>
+ <td><code>A:</code> return value register-pair (8 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ The return type of the current method must be double or long.
+ </li>
+ <li>
+ A+1 must be a valid register index in the current stack frame (which
+ includes A being valid).
+ </li>
+ <li>
+ Register vA must be the lower half of a register pair.
+ </li>
+ <li>
+ The type of vA must match the return type of the method.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ If the method is synchronized, the object's monitor is released in a way
+ similar to the monitor-exit instruction.
+ </li>
+ <li>
+ The stack frame of the current method invocation is removed from the stack.
+ This includes all its registers becoming invalid.
+ </li>
+ <li>
+ If the stack is now empty, the current thread terminates.
+ </li>
+ <li>
+ Otherwise, the following happens:
+ <ul>
+ <li>
+ The stack frame that caused this method invocation becomes valid. This
+ includes all its registers and their old values.
+ </li>
+ <li>
+ Execution continues at the bytecode instruction immediately following
+ the invoke instruction that caused this method invocation.
+ </li>
+ <li>
+ The return value can be consumed by (exactly) the first instruction
+ following the invoke-kind or invoke-kind/range instruction that caused
+ this method invocation, and this instructions needs to be a
+ move-result-wide instruction.
+ </li>
+ </ul>
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-11-return-object.html b/docs/opcodes/opcode-11-return-object.html
new file mode 100644
index 0000000..369233c
--- /dev/null
+++ b/docs/opcodes/opcode-11-return-object.html
@@ -0,0 +1,96 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>return-object</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>return-object</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Return from an object-returning method.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>11 11x</td>
+ <td>return-object vAA</td>
+ <td><code>A:</code> return value register (8 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ The return type of the current method must be a reference.
+ </li>
+ <li>
+ A must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ Register vA must be known to be reference-bearing.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ If the method is synchronized, the object's monitor is released in a way
+ similar to the monitor-exit instruction.
+ </li>
+ <li>
+ The stack frame of the current method invocation is removed from the stack.
+ This includes all its registers becoming invalid.
+ </li>
+ <li>
+ If the stack is now empty, the current thread terminates.
+ </li>
+ <li>
+ Otherwise, the following happens:
+ </li>
+ <ul>
+ <li>
+ The stack frame that caused this method invocation becomes valid. This
+ includes all its registers and their old values.
+ </li>
+ <li>
+ Execution continues at the bytecode instruction immediately following
+ the invoke instruction that caused this method invocation.
+ </li>
+ <li>
+ The return value can be consumed by (exactly) the first instruction
+ following the invoke-kind or invoke-kind/range instruction that caused
+ this method invocation, and this instructions needs to be a
+ move-result-object instruction.
+ </li>
+ </ul>
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-12-const.html b/docs/opcodes/opcode-12-const.html
new file mode 100644
index 0000000..b2d894f
--- /dev/null
+++ b/docs/opcodes/opcode-12-const.html
@@ -0,0 +1,102 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>const</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>const</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Move the given literal value (sign-extended to 32 bits, if necessary) into the
+specified register.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>12 11n</td>
+ <td>const/4 vA, #+B</td>
+ <td><code>A:</code> destination register (4 bits)<br/>
+ <code>B:</code> signed int (4 bits)</td>
+</tr>
+<tr>
+ <td>13 21s</td>
+ <td>const/16 vAA, #+BBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> signed int (16 bits)</td>
+</tr>
+<tr>
+ <td>14 31i</td>
+ <td>const vAA, #+BBBBBBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> arbitrary 32-bit constant</td>
+</tr>
+<tr>
+ <td>15 21h</td>
+ <td>const/high16 vAA, #+BBBB0000</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> signed int (16 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index in the current stackframe.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ First, an adjusted value B' is determined as follows:
+ <ul>
+ <li>
+ If we are executing the /high16 variant, then B is left-shifted by 16
+ bits, that is, B'=B << 0x10
+ <li>
+ Otherwise, if B is a 4 bit or 16 bit constant, it is sign-extended to 32
+ bits, that is, B'=sign-extended(B).
+ </li>
+ <li>
+ Otherwise, B'=B.
+ </li>
+ </ul>
+ <li>
+ Then, the adjusted value B' is moved into the register A, that is, vA'=B'
+ </li>
+ <li>
+ If v(A-1) is the lower half of a register pair, v(A-1)' becomes undefined.
+ </li>
+ <li>
+ If v(A+1) is the upper half of a register pair, v(A+1)' becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-16-const-wide.html b/docs/opcodes/opcode-16-const-wide.html
new file mode 100644
index 0000000..ac92b04
--- /dev/null
+++ b/docs/opcodes/opcode-16-const-wide.html
@@ -0,0 +1,110 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>const-wide</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>const-wide</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Move the given literal value (sign-extended to 64 bits) into the specified
+register-pair.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>16 21s</td>
+ <td>const-wide/16 vAA, #+BBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> signed int (16 bits)</td>
+</tr>
+<tr>
+ <td>17 31i</td>
+ <td>const-wide/32 vAA, #+BBBBBBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> signed int (32 bits)</td>
+</tr>
+<tr>
+ <td>18 51l</td>
+ <td>const-wide vAA, #+BBBBBBBBBBBBBBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> arbitrary double-width (64-bit) constant</td>
+</tr>
+<tr>
+ <td>19 21h</td>
+ <td>const-wide/high16 vAA, #+BBBB000000000000</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> signed int (16 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index in the current stack frame.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ First, an adjusted value B' is determined as follows:
+ <ul>
+ <li>
+ If we are executing the /high16 variant, then B is left-shifted by 40
+ bits, that is, B'=B << 0x28
+ <li>
+ Otherwise, if B is a 16 bit or 32 bit constant, it is sign-extended to
+ 64 bits, that is, B'=sign-extended(B).
+ </li>
+ <li>
+ Otherwise, B'=B.
+ </li>
+ </ul>
+ <li>
+ The immediate value B is moved into the register pair (vA, v(A+1)), that is,
+ <ul>
+ <li>
+ vA' = B << 0x20
+ </li>
+ <li>
+ v(A+1)' = B & 0xffffffff
+ </li>
+ </ul>
+ </li>
+ <li>
+ If v(A-1) is the lower half of a register pair, v(A-1)' becomes undefined.
+ </li>
+ <li>
+ If v(A+2) is the upper half of a register pair, v(A+2)' becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-1a-const-string.html b/docs/opcodes/opcode-1a-const-string.html
new file mode 100644
index 0000000..bd89b70
--- /dev/null
+++ b/docs/opcodes/opcode-1a-const-string.html
@@ -0,0 +1,85 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>const-string</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>const-string</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Move a reference to the string specified by the given index into the specified
+register.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>1a 21c</td>
+ <td>const-string vAA, string@BBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> string index</td>
+</tr>
+<tr>
+ <td>1b 31c</td>
+ <td>const-string/jumbo vAA, string@BBBBBBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> string index</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ B must be a valid index into the string constant pool.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ A new java.lang.String object S is allocated on the heap and filled with the
+ contents of string pool entry B.
+ </li>
+ <li>
+ A reference to an internalized version of the new object is moved into
+ register vA, that is, the instruction behaves as if vA' = S.intern() was
+ called.
+ </li>
+ <li>
+ If v(A-1) is the lower half of a register pair, v(A-1)' becomes undefined.
+ </li>
+ <li>
+ If v(A+1) is the upper half of a register pair, v(A+1)' becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-1b-const-class.html b/docs/opcodes/opcode-1b-const-class.html
new file mode 100644
index 0000000..7a33e77
--- /dev/null
+++ b/docs/opcodes/opcode-1b-const-class.html
@@ -0,0 +1,92 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>const-class</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>const-class</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Move a reference to the class specified by the given index into the specified
+register. In the case where the indicated type is primitive, this will store a
+reference to the primitive type's degenerate class.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>1c 21c</td>
+ <td>const-class vAA, type@BBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> type index</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ B must be a valid index into the type constant pool.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ An attempt is made to get a reference to the class C the name of which is
+ contained in type pool entry B.
+ </li>
+ <li>
+ If B refers to a primitive type, the corresponding degenerate class is used
+ instead.
+ </li>
+ <li>
+ If C has not been loaded and resolved before, it is being loaded and
+ resolved. All exceptions that are possible during class loading can occur at
+ this point.
+ </li>
+ <li>
+ A reference to C is moved into register vA, that is, vA' = C.
+ </li>
+ <li>
+ If v(A-1) is the lower half of a register pair, v(A-1)' becomes undefined.
+ </li>
+ <li>
+ If v(A+1) is the upper half of a register pair, v(A+1)' becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ ClassNotFoundException is thrown if the class does not exist at all.
+ </li>
+ <li>
+ VerifyError is thrown if the class does exist, but could not be verified.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-1d-monitor-enter.html b/docs/opcodes/opcode-1d-monitor-enter.html
new file mode 100644
index 0000000..cecc939
--- /dev/null
+++ b/docs/opcodes/opcode-1d-monitor-enter.html
@@ -0,0 +1,93 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>monitor-enter</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>monitor-enter</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Acquire the monitor for the indicated object.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>1d 11x</td>
+ <td>monitor-enter vAA</td>
+ <td><code>A:</code> reference-bearing register (8 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index for the current stack frame.
+ </li>
+ <li>
+ Register vA must contain a reference to an object.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ An attempt is made for the current thread to acquire the monitor of the
+ indicated object. Various results are possible:
+ <ul>
+ <li>
+ If the monitor is not owned by any thread at this point, then the
+ current thread becomes owner of the monitor. The entry count of the
+ indicated object is set to 1.
+ </li>
+ <li>
+ Otherwise, if the monitor is owned by the same thread that attempts the
+ acquiration, then the entry count of the indicated object is increased
+ by 1.
+ </li>
+ <li>
+ Otherwise the monitor is owned by a different thread. The current thread
+ sleeps until the monitor of the object is released. Once that happens, a
+ new attempt to acquire the monitor is made, as described here. There is
+ no guarantee that the second attempt (or any subsequent attempt) will be
+ successful.
+ </li>
+ </ul>
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ NullPointerException if vA is null.
+ </li>
+ <li>
+ IllegalMonitorStateException if the entry count exceeds an
+ (implementation-dependent) upper bound for recursive monitor entries. Note
+ that it is unlikely this bound is ever hit, since for most implementations
+ the call stack will overflow before.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-1e-monitor-exit.html b/docs/opcodes/opcode-1e-monitor-exit.html
new file mode 100644
index 0000000..bee711d
--- /dev/null
+++ b/docs/opcodes/opcode-1e-monitor-exit.html
@@ -0,0 +1,101 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>monitor-exit</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>monitor-exit</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Release the monitor for the indicated object.
+</p>
+<p>
+Note: If this instruction needs to throw an exception, it must do so as if the
+pc has already advanced past the instruction. It may be useful to think of this
+as the instruction successfully executing (in a sense), and the exception
+getting thrown after the instruction but before the next one gets a chance to
+run. This definition makes it possible for a method to use a monitor cleanup
+catch-all (e.g., finally) block as the monitor cleanup for that block itself,
+as a way to handle the arbitrary exceptions that might get thrown due to the
+historical implementation of Thread.stop(), while still managing to have proper
+monitor hygiene.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>1e 11x</td>
+ <td>monitor-exit vAA</td>
+ <td><code>A:</code> reference-bearing register (8 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index for the current stack frame.
+ </li>
+ <li>
+ Register vA must contain a reference to an object.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ An attempt is made for the current thread to release the monitor of the
+ indicated object.
+ </li>
+ <li>
+ If the current thread is the owner, the following happens:
+ <ul>
+ <li>
+ The monitor's entry count is decreased by one.
+ </li>
+ <li>
+ If the entry count has reached zero, the monitor is released. Other
+ threads waiting for the same monitor have a chance to acquire it.
+ </li>
+ </ul>
+ </li>
+ <li>
+ Any exception that gets thrown by this instruction bears the PC of the
+ instruction following the monitor-exit. That is, from the point of view of
+ an exception handler it cannot be distinguished from the same type of
+ exception being thrown immediately after the monitor-exit instruction.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ NullPointerException is thrown if vA is null.
+ </li>
+ <li>
+ IllegalMonitorStateException is thrown if the current thread is not the
+ owner of that monitor.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-1f-check-cast.html b/docs/opcodes/opcode-1f-check-cast.html
new file mode 100644
index 0000000..6325ca4
--- /dev/null
+++ b/docs/opcodes/opcode-1f-check-cast.html
@@ -0,0 +1,95 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>check-cast</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>check-cast</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Throw if the reference in the given register cannot be cast to the indicated
+type. The type must be a reference type (not a primitive type).
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>1f 21c</td>
+ <td>check-cast vAA, type@BBBB</td>
+ <td><code>A:</code> reference-bearing register (8 bits)<br/>
+ <code>B:</code> type index (16 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ Register vA must contain a reference value.
+ </li>
+ <li>
+ B must be a valid index into the type pool.
+ </li>
+ <li>
+ Type pool entry B must contain a valid type descriptor for a reference type.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ An attempt is made to get a reference to the class C the name of which is
+ contained in type pool entry B.
+ </li>
+ <li>
+ If C has not been loaded and resolved before, it is being loaded and
+ resolved. All exceptions that are possible during class loading can occur at
+ this point.
+ </li>
+ <li>
+ The run-time type of the object reference vA is compared against C.
+ <ul>
+ <li>
+ If vA is null, the instruction succeeds (without further effects).
+ </li>
+ <li>
+ If vA is assignment compatible with C according to the usual rules of
+ the Java programming language, the instruction succeeds (without further
+ effects).
+ </li>
+ </ul>
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ ClassCastException is thrown if vA is either not null or not assignment
+ compatible with C.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-20-instance-of.html b/docs/opcodes/opcode-20-instance-of.html
new file mode 100644
index 0000000..5dbfef7
--- /dev/null
+++ b/docs/opcodes/opcode-20-instance-of.html
@@ -0,0 +1,100 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>instance-of</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>instance-of</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Store in the given destination register 1 if the indicated reference is an
+instance of the given type, or 0 if not. The type must be a reference type (not
+a primitive type).
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>20 22c</td>
+ <td>instance-of vA, vB, type@CCCC</td>
+ <td><code>A:</code> destination register (4 bits)<br/>
+ <code>B:</code> reference-bearing register (4 bits)<br/>
+ <code>C:</code> type index (16 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ Both A and B must be a valid register indices for the current stack frame.
+ </li>
+ <li>
+ Register vB must contain a reference value.
+ </li>
+ <li>
+ C must be a valid index into the type constant pool.
+ </li>
+ <li>
+ Type constant pool entry C must contain a valid type descriptor for a
+ reference type.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ An attempt is made to get a reference to the class K the name of which is
+ contained in type pool entry C.
+ </li>
+ <li>
+ If K has not been loaded and resolved before, it is being loaded and
+ resolved. All exceptions that are possible during class loading can occur at
+ this point.
+ </li>
+ <li>
+ The run-time type of the object reference vB is compared against K. The
+ register vA reflects the result:
+ <ul>
+ <li>
+ vA' = 1 if (and only if) vB is not null and vB is assignment compatible
+ with K according to the usual rules of the Java programming language.
+ </li>
+ <li>
+ vA' = 0 otherwise
+ </li>
+ </ul>
+ </li>
+ <li>
+ If v(A-1) is the lower half of a register pair, v(A-1)' becomes undefined.
+ </li>
+ <li>
+ If v(A+1) is the upper half of a register pair, v(A+1)' becomes undefined.
+ </li>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-21-array-length.html b/docs/opcodes/opcode-21-array-length.html
new file mode 100644
index 0000000..5767742
--- /dev/null
+++ b/docs/opcodes/opcode-21-array-length.html
@@ -0,0 +1,78 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>array-length</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>array-length</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Store in the given destination register the length of the indicated array,
+in entries.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>21 12x</td>
+ <td>array-length vA, vB</td>
+ <td><code>A:</code> destination register (4 bits)<br/>
+ <code>B:</code> array reference-bearing register (4 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ Both A and B must be valid register indices for the current stack frame.
+ </li>
+ <li>
+ Register vB must contain a reference to an array.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The length of the array referenced by vB is stored in vA, that is
+ vA' = length(vB).
+ </li>
+ <li>
+ If register v(A-1) is the lower half of a register pair, register v(A-1)'
+ becomes undefined.
+ </li>
+ <li>
+ If register v(A+1) is the upper half of a register pair, register v(A+1)'
+ becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ NullPointerException is thrown if the value of register vB is null.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-22-new-instance.html b/docs/opcodes/opcode-22-new-instance.html
new file mode 100644
index 0000000..ec0a347
--- /dev/null
+++ b/docs/opcodes/opcode-22-new-instance.html
@@ -0,0 +1,95 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>new-instance</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>new-instance</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Construct a new instance of the indicated type, storing a reference to it in the
+destination. The type must refer to a non-array class.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>22 21c</td>
+ <td>new-instance vAA, type@BBBB</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> type index</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index for the current stack frame.
+ </li>
+ <li>
+ B must be a valid index into the type pool.
+ </li>
+ <li>
+ Type constant pool entry B must contain a valid type descriptor for a
+ non-array class.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ An attempt is made to get a reference to the class C the name of which is
+ contained in type pool entry B.
+ </li>
+ <li>
+ If C has not been loaded and resolved before, it is being loaded and
+ resolved. All exceptions that are possible during class loading can occur at
+ this point.
+ </li>
+ <li>
+ An attempt is made to create a new instance I of C. All exceptions that are
+ possible during instantiation can occur at this point.
+ </li>
+ <li>
+ A reference to the new instance is stored in register vA, that is vA' = I.
+ </li>
+ <li>
+ If v(A-1) is the lower part of a register pair, v(A-1)' becomes undefined.
+ </li>
+ <li>
+ If v(A+1) is the upper part of a register pair, v(A+1)' becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ All exceptions that are possible during class loading can occur.
+ </li>
+ <li>
+ All exceptions that are possible during instantiation can occur.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-23-new-array.html b/docs/opcodes/opcode-23-new-array.html
new file mode 100644
index 0000000..be46a49
--- /dev/null
+++ b/docs/opcodes/opcode-23-new-array.html
@@ -0,0 +1,108 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>new-array</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>new-array</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Construct a new array of the indicated type and size. The type must be an array
+type.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>23 22c</td>
+ <td>new-array vA, vB, type@CCCC</td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> size register<br/>
+ <code>C:</code> type index</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A and B must be valid register indices for the current stack frame.
+ </li>
+ <li>
+ Register vB must not contain a reference value.
+ </li>
+ <li>
+ Register vB must not be part of a register pair.
+ </li>
+ <li>
+ C must be a valid index into the type pool.
+ </li>
+ <li>
+ Type constant pool entry C must contain a valid array type descriptor.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ An attempt is made to get a reference to the class K the name of which is
+ contained in type pool entry C.
+ </li>
+ <li>
+ If K has not been loaded and resolved before, it is being loaded and
+ resolved. All exceptions that are possible during class loading can occur at
+ this point.
+ </li>
+ <li>
+ An attempt is made to create a new instance I of K and length B. All
+ exceptions that are possible during instantiation can occur at this point.
+ </li>
+ <li>
+ All elements of the new array are initialized to null (for object arrays) or
+ 0 (for numeric arrays) or false (for boolean arrays).
+ </li>
+ <li>
+ A reference to the new array is moved to register vA, that is, vA' = I.
+ </li>
+ <li>
+ If v(A-1) is the lower half of a register pair, v(A-1)' becomes undefined.
+ </li>
+ <li>
+ If v(A+1) is the upper half of a register pair, v(A+1)' becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ NegativeArraySizeException if vB < 0
+ </li>
+ <li>
+ All exceptions that are possible during class loading can occur.
+ </li>
+ <li>
+ All exceptions that are possible during instantiation can occur.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-24-filled-new-array.html b/docs/opcodes/opcode-24-filled-new-array.html
new file mode 100644
index 0000000..cba0c5d
--- /dev/null
+++ b/docs/opcodes/opcode-24-filled-new-array.html
@@ -0,0 +1,144 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>filled-new-array</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>filled-new-array</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Construct an array of the given type and size, filling it with the supplied
+contents. The type must be an array type. The array's contents must be
+single-word (that is, no arrays of long or double). The constructed instance is
+stored as a "result" in the same way that the method invocation instructions
+store their results, so the constructed instance must be moved to a register
+with a subsequent move-result-object instruction (if it is to be used).
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>24 35c</td>
+ <td>filled-new-array {vD, vE, vF, vG, vA}, type@CCCC</td>
+ <td><code>B:</code> array size and argument word count (4 bits)<br/>
+ <code>C:</code> type index (16 bits)<br/>
+ <code>D..G, A:</code> argument registers (4 bits each)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ The value B must not be greater than 5.
+ </li>
+ <li>
+ If B > 0, then D must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ If B > 1, then E must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ If B > 2, then F must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ If B > 3, then G must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ If B > 4, then A must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ C must be a valid index into the type pool.
+ </li>
+ <li>
+ The type denoted by C must be a valid array type descriptor.
+ </li>
+ <li>
+ The element size of the type denoted by C must be no larger than 32 bits.
+ </li>
+ <li>
+ If the element type is a primitive type, then all actual arguments
+ (vD .. vA, depending on B) must be primitive, too.
+ </li>
+ <li>
+ If the element type is a reference type, then all actual arguments
+ (vD .. vA, depending on B) must be references, too.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ An attempt is made to get a reference to the type T the name of which is
+ contained in type pool entry C.
+ </li>
+ <li>
+ If T is a reference type and it has not been loaded and resolved before, it
+ is being loaded and resolved. All exceptions that are possible during class
+ loading can occur at this point.
+ </li>
+ <li>
+ An attempt is made to create a new array R of type T and length B. All
+ exceptions that are possible during instantiation can occur at this point.
+ </li>
+ <li>
+ The elements of R are filled according to the following rules:
+ <ul>
+ <li>
+ If B > 0 then R[0] = vD
+ </li>
+ <li>
+ If B > 1 then R[1] = vE
+ </li>
+ <li>
+ If B > 2 then R[2] = vF
+ </li>
+ <li>
+ If B > 3 then R[3] = vG
+ </li>
+ <li>
+ If B > 4 then R[4] = vA
+ </li>
+ </ul>
+ </li>
+ <li>
+ No reference to R is stored in any register. Instead, R can be accessed by a
+ move-result-object instruction immediately following this filled-new-array
+ instruction.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ NegativeArraySizeException if vB < 0
+ </li>
+ <li>
+ All exceptions that are possible during class loading can occur.
+ </li>
+ <li>
+ All exceptions that are possible during instantiation can occur.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-25-filled-new-array-range.html b/docs/opcodes/opcode-25-filled-new-array-range.html
new file mode 100644
index 0000000..944a8b1
--- /dev/null
+++ b/docs/opcodes/opcode-25-filled-new-array-range.html
@@ -0,0 +1,125 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>filled-new-array/range</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>filled-new-array/range</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Construct an array of the given type and size, filling it with the supplied
+contents. Clarifications and restrictions are the same as filled-new-array,
+described above.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>25 3rc</td>
+ <td>filled-new-array/range {vCCCC .. vNNNN}, type@BBBB</td>
+ <td><code>A:</code> array size and argument word count (8 bits)<br/>
+ <code>B:</code> type index (16 bits)<br/>
+ <code>C:</code> first argument register (16 bits)<br/>
+ <code>N = A + C - 1</code></td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ vN must be a valid register index in the current stack frame (this means
+ everything below vN is valid, too).
+ </li>
+ <li>
+ For all values I in the interval [C .. N] the following must hold:
+ <ul>
+ <li>
+ vI must not be part of a register pair
+ </li>
+ <li>
+ If the array type is a simple type, vI must be a simple type, too.
+ </li>
+ <li>
+ If the array type is a reference type, vI must be a reference type, too.
+ </li>
+ </ul>
+ </li>
+ <li>
+ B must be a valid index into the type pool.
+ </li>
+ <li>
+ The type denoted by B must be an array type.
+ </li>
+ <li>
+ The element size of the type denoted by B must be no larger than 32 bits.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ An attempt is made to get a reference to the type T the name of which is
+ contained in type pool entry B.
+ </li>
+ <li>
+ If T has not been loaded and resolved before, it is being loaded and
+ resolved. All exceptions that are possible during class loading can occur at
+ this point.
+ </li>
+ <li>
+ An attempt is made to create a new instance J of type T and length vA. All
+ exceptions that are possible during instantiation can occur at this point.
+ </li>
+ <li>
+ The elements of R are filled according to the following rules:
+ <ul>
+ <li>
+ J[0] = vC
+ </li>
+ <li>
+ J[1] = v(C+1)
+ </li>
+ <li>
+ ...
+ </li>
+ <li>
+ J[vA] = vN
+ </li>
+ </ul>
+ </li>
+ <li>
+ No reference to J is stored in any register. Instead, J can be accessed by a
+ move-result-object instruction immediately following this filled-new-array
+ instruction.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ NegativeArraySizeException if vA < 0
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-26-fill-array-data.html b/docs/opcodes/opcode-26-fill-array-data.html
new file mode 100644
index 0000000..c5040a5
--- /dev/null
+++ b/docs/opcodes/opcode-26-fill-array-data.html
@@ -0,0 +1,96 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>fill-array-data</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>fill-array-data</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Fill the given array with the indicated data. The reference must be to an array
+of primitives, and the data table must match it in type and size.
+</p>
+<p>
+Note: The address of the table is guaranteed to be even (that is, 4-byte
+aligned). If the code size of the method is otherwise odd, then an extra code
+unit is inserted between the main code and the table whose value is the same as
+a nop.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+ <td>26 31t</td>
+ <td>fill-array-data vAA, +BBBBBBBB <i>(with supplemental data as specified
+ below in "<code>fill-array-data</code> Format")</i></td>
+ <td><code>A:</code> array reference (8 bits)<br/>
+ <code>B:</code> signed "branch" offset to table data (32 bits)</td>
+ </td>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ vA must be a reference-bearing register (according to data flow) and contain
+ an array-reference.
+ </li>
+ <li>
+ B must be branch offset in the same method.
+ </li>
+ <li>
+ The target address (PC+B) must be 4-byte aligned.
+ </li>
+ <li>
+ The target address must hold the pseudo-opcode 0x300.
+ </li>
+ <li>
+ The table entry size must match the size of the data type of the array.
+ </li>
+ <li>
+ The table size must be equal or smaller than the array length.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The contents of the array referenced by vA are filled with the table data,
+ starting from array index 0 and in the given order.
+ </li>
+ <li>
+ If there are less elements in the table than the array provides space for,
+ the remaining array elements stay untouched.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ NullPointerException if vA is null.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-27-throw.html b/docs/opcodes/opcode-27-throw.html
new file mode 100644
index 0000000..f4416ce
--- /dev/null
+++ b/docs/opcodes/opcode-27-throw.html
@@ -0,0 +1,80 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>throw</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>throw</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Throw the indicated exception.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>27 11x</td>
+ <td>throw vAA</td>
+ <td><code>A:</code> exception-bearing register (8 bits)<br/></td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ Register vA must be a reference-bearing register
+ </li>
+ <li>
+ Register vA must be assignment-compatible with java.lang.Throwable according
+ to the usual rules of the Java programming language.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ Throws the given exception vA, resulting in a search for a matching handler
+ according to the usual rules of the Java programming language.
+ </li>
+ <li>
+ If no matching handler is found for the current thread, the thread
+ terminates, possibly notifying its uncaught exception handler or thread
+ group before.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ NullPointerException if vA is null.
+ </li>
+ <li>
+ Otherwise, the indicated exception.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-28-goto.html b/docs/opcodes/opcode-28-goto.html
new file mode 100644
index 0000000..5ca92b6
--- /dev/null
+++ b/docs/opcodes/opcode-28-goto.html
@@ -0,0 +1,73 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>goto</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>goto</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Unconditionally jump to the indicated instruction.
+</p>
+<p>
+Note: The branch offset may not be 0. (A spin loop may be legally constructed
+either with goto/32 or by including a nop as a target before the branch.)
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>28 10t</td>
+ <td>goto +AA</td>
+ <td><code>A:</code> signed branch offset (8 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must point to a valid bytecode instruction inside the current method.
+ </li>
+ <li>
+ A must not be 0.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The (otherwise invisible) program counter PC is set to the address of the
+ instruction plus the given offset, that is, PC' = PC(goto) + A.
+ </li>
+ <li>
+ Executions resumes at PC'.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-29-goto-16.html b/docs/opcodes/opcode-29-goto-16.html
new file mode 100644
index 0000000..abe4aec
--- /dev/null
+++ b/docs/opcodes/opcode-29-goto-16.html
@@ -0,0 +1,73 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>goto/16</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>goto/16</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Unconditionally jump to the indicated instruction.
+</p>
+<p>
+Note: The branch offset may not be 0. (A spin loop may be legally constructed
+either with goto/32 or by including a nop as a target before the branch.)
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>29 20t</td>
+ <td>goto/16 +AAAA</td>
+ <td><code>A:</code> signed branch offset (16 bits)<br/></td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must point to a valid bytecode instruction inside the current method.
+ </li>
+ <li>
+ A must not be 0.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The (otherwise invisible) program counter PC is set to the address of the
+ instruction plus the given offset, that is, PC' = PC(goto) + A.
+ </li>
+ <li>
+ Executions resumes at PC'.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-2a-goto-32.html b/docs/opcodes/opcode-2a-goto-32.html
new file mode 100644
index 0000000..7d64ac9
--- /dev/null
+++ b/docs/opcodes/opcode-2a-goto-32.html
@@ -0,0 +1,66 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>goto/32</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>goto/32</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Unconditionally jump to the indicated instruction.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>2a 30t</td>
+ <td>goto/32 +AAAAAAAA</td>
+ <td><code>A:</code> signed branch offset (32 bits)<br/></td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must point to a valid bytecode instruction inside the current method.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The (otherwise invisible) program counter PC is set to the address of the
+ instruction plus the given offset, that is, PC' = PC(goto) + A.
+ </li>
+ <li>
+ Executions resumes at PC'.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-2b-packed-switch.html b/docs/opcodes/opcode-2b-packed-switch.html
new file mode 100644
index 0000000..73125bd
--- /dev/null
+++ b/docs/opcodes/opcode-2b-packed-switch.html
@@ -0,0 +1,103 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>packed-switch</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>packed-switch</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Jump to a new instruction based on the value in the given register, using a
+table of offsets corresponding to each value in a particular integral range, or
+fall through to the next instruction if there is no match.
+</p>
+<p>
+Note: The address of the table is guaranteed to be even (that is, 4-byte
+aligned). If the code size of the method is otherwise odd, then an extra code
+unit is inserted between the main code and the table whose value is the same as
+a nop.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>2b 31t</td>
+ <td>packed-switch vAA, +BBBBBBBB <i>(with supplemental data as
+ specified below in "<code>packed-switch</code> Format")</i></td>
+ <td><code>A:</code> register to test<br/>
+ <code>B:</code> signed "branch" offset to table data (32 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ Let PC be the address of the packed-switch instruction in the code array of
+ the current method. Then T = PC + B with the following properties:
+ <ul>
+ <li>
+ T must be 4-byte-aligned.
+ </li
+ <li>
+ T must be in the same method.
+ </li>
+ <li>
+ T must point to a packed-switch data table.
+ </li>
+ </ul>
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The value of vA is used as an index into the given table data.
+ </li>
+ <li>
+ If vA is in the range of the table, that is, if vA >= table.first_key and
+ vA < first_key + size, then the jump target is determined as follows:
+ <ul>
+ <li>
+ PC' = PC + table.targets[vA - table.firstKey].
+ </li>
+ <li>
+ Execution resumes at this address.
+ </li>
+ </ul>
+ </li>
+ <li>
+ Otherwise execution continues at the instruction following the packed-switch
+ statement.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-2c-sparse-switch.html b/docs/opcodes/opcode-2c-sparse-switch.html
new file mode 100644
index 0000000..cec581e
--- /dev/null
+++ b/docs/opcodes/opcode-2c-sparse-switch.html
@@ -0,0 +1,110 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>sparse-switch</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>sparse-switch</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Jump to a new instruction based on the value in the given register, using an
+ordered table of value-offset pairs, or fall through to the next instruction if
+there is no match.
+</p>
+<p>
+Note: The address of the table is guaranteed to be even (that is, 4-byte
+aligned). If the code size of the method is otherwise odd, then an extra code
+unit is inserted between the main code and the table whose value is the same as
+a nop.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>2c 31t</td>
+ <td>sparse-switch vAA, +BBBBBBBB <i>(with supplemental data as
+ specified below in "<code>sparse-switch</code> Format")</i></td>
+ <td><code>A:</code> register to test<br/>
+ <code>B:</code> signed "branch" offset to table data (32 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index in the current stack frame.
+ </li>
+ <li>
+ Let PC be the address of the packed-switch instruction in the code array of
+ the current method. Then T = PC + B with the following properties:
+ <ul>
+ <li>
+ T must be 4-byte-aligned.
+ </li>
+ <li>
+ T must be in the same method.
+ </li>
+ <li>
+ T must point to a sparse-switch data table.
+ </li>
+ </ul>
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The value of vA is used as a lookup key inside the sparse table data.
+ </li>
+ <li>
+ If there exists an I with 0 <= I < table.size such that table.keys[I] = vA,
+ then the jump target is determined as follows:
+ <ul>
+ <li>
+ PC' = PC + table.targets[I].
+ </li>
+ <li>
+ Execution will resume at this address.
+ </li>
+ </ul>
+ </li>
+ <li>
+ Otherwise execution continues at the instruction following the sparse-switch
+ statement.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+<h2>Notes</h2>
+
+<p>
+The low-to-high ordering of the keys allows the VM to employ binary search for
+the lookup, resulting in O(log table.size) comparisons.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-2d-cmp-kind.html b/docs/opcodes/opcode-2d-cmp-kind.html
new file mode 100644
index 0000000..431ccd4
--- /dev/null
+++ b/docs/opcodes/opcode-2d-cmp-kind.html
@@ -0,0 +1,121 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>cmp<kind></title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>cmp<kind></h1>
+
+<h2>Purpose</h2>
+
+<p>
+Perform the indicated floating point or long comparison, storing 0 if the two
+arguments are equal, 1 if the second argument is larger, or -1 if the first
+argument is larger. The "bias" listed for the floating point operations
+indicates how NaN comparisons are treated: "Gt bias" instructions return 1 for
+NaN comparisons, and "lt bias" instructions return -1.
+</p>
+<p>
+For example, to check to see if floating point a < b, then it is advisable to
+use cmpg-float; a result of -1 indicates that the test was true, and the other
+values indicate it was false either due to a valid comparison or because one
+or the other values was NaN.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>2d..31 23x</td>
+ <td>cmp<i>kind</i> vAA, vBB, vCC<br/>
+ 2d: cmpl-float <i>(lt bias)</i><br/>
+ 2e: cmpg-float <i>(gt bias)</i><br/>
+ 2f: cmpl-double <i>(lt bias)</i><br/>
+ 30: cmpg-double <i>(gt bias)</i><br/>
+ 31: cmp-long
+ </td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> first source register or pair<br/>
+ <code>C:</code> second source register or pair</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A, B and C must be valid register indices in the current stack frame.
+ </li>
+ <li>
+ For the two -float variants, both vB and vC must be of type float.
+ </li>
+ <li>
+ For the two -double variants, both vB and vC must be the lower part of a
+ register pair holding a double value.
+ </li>
+ <li>
+ For the -long variant, both both vB and vC must be the lower part of a
+ register pair holding a long value.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The values of registers vB and vC are compared. The result, which is stored
+ in vA, is one of the following three:
+ <ul>
+ <li>
+ If vB < vC, then vA'=-1.
+ </li>
+ <li>
+ If vB == vC, then vA'=0.
+ </li>
+ <li>
+ If vC > vC, then vA'=1.
+ </li>
+ </ul>
+ </li>
+ <li>
+ For the -float and -double variants, an addition "bias" specifies what
+ happens if one or both of the arguments are NaN:
+ <ul>
+ <li>
+ A "lt bias" results in vA'=-1.
+ </li>
+ <li>
+ A "gt bias" results in vA'=1.
+ </li>
+ </ul>
+ </li>
+ <li>
+ If v(A-1) is the lower half of a register pair, v(A-1)' becomes undefined.
+ </li>
+ <li>
+ If v(A+1) is the upper half of a register pair, v(A+1)' becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-32-if-test.html b/docs/opcodes/opcode-32-if-test.html
new file mode 100644
index 0000000..af0adb2
--- /dev/null
+++ b/docs/opcodes/opcode-32-if-test.html
@@ -0,0 +1,101 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>if-test</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>if-test</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Branch to the given destination if the given two registers' values compare as
+specified.
+</p>
+<p>
+Note: The branch offset may not be 0. (A spin loop may be legally constructed
+either by branching around a backward goto or by including a nop as a target
+before the branch.)
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>32..37 22t</td>
+ <td>if-<i>test</i> vA, vB, +CCCC<br/>
+ 32: if-eq<br/>
+ 33: if-ne<br/>
+ 34: if-lt<br/>
+ 35: if-ge<br/>
+ 36: if-gt<br/>
+ 37: if-le<br/>
+ </td>
+ <td><code>A:</code> first register to test (4 bits)<br/>
+ <code>B:</code> second register to test (4 bits)<br/>
+ <code>C:</code> signed branch offset (16 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A and B must be valid register indices for the current stack frame.
+ </li>
+ <li>
+ Registers vA and vB must not contain a reference value.
+ </li>
+ <li>
+ Registers vA and vB must not be part of a register pair.
+ </li>
+ <li>
+ Registers vA and vB must not contain a floating point value (???).
+ </li>
+ C must of a signed offset that, when added to the PC of the instruction,
+ points to a valid bytecode instruction inside the same method.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The values of registers vA and vB are compared according to the <test>
+ condition. Two results are possible:
+ <ul>
+ <li>
+ The condition holds. The value of C is used as a signed offset to the
+ address of the if-<test> instruction. Execution continues at the
+ resulting address.
+ </li>
+ <li>
+ The condition does not hold. Execution continues at the instruction
+ following the if-<test> instruction.
+ </li>
+ </ul>
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-38-if-testz.html b/docs/opcodes/opcode-38-if-testz.html
new file mode 100644
index 0000000..354e894
--- /dev/null
+++ b/docs/opcodes/opcode-38-if-testz.html
@@ -0,0 +1,100 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>if-<test>z</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>if-<test>z</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Branch to the given destination if the given register's value compares with 0
+as specified.
+</p>
+<p>
+ Note: The branch offset may not be 0. (A spin loop may be legally constructed
+ either by branching around a backward goto or by including a nop as a target
+ before the branch.)
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>38..3d 21t</td>
+ <td>if-<i>test</i>z vAA, +BBBB<br/>
+ 38: if-eqz<br/>
+ 39: if-nez<br/>
+ 3a: if-ltz<br/>
+ 3b: if-gez<br/>
+ 3c: if-gtz<br/>
+ 3d: if-lez<br/>
+ </td>
+ <td><code>A:</code> register to test (8 bits)<br/>
+ <code>B:</code> signed branch offset (16 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index for the current stackframe.
+ </li>
+ <li>
+ Register vA must not contain a reference value.
+ </li>
+ <li>
+ Register vA must not be part of a register pair.
+ </li>
+ <li>
+ Register vA must not contain a floating point value (???).
+ </li>
+ <li>
+ B must not be 0.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The value of register vA is compared to zero according to the <test>
+ condition. Two results are possible:
+ <ul>
+ <li>
+ The condition holds. The value of B is used as a signed offset to the
+ address of the if-<test>z instruction. Execution continues at the
+ resulting address.
+ </li>
+ <li>
+ The condition does not hold. Execution continues at the instruction
+ following the if-<test>z instruction.
+ </li>
+ </ul>
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<p>
+None.
+</p>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-44-aget.html b/docs/opcodes/opcode-44-aget.html
new file mode 100644
index 0000000..b9d6f23
--- /dev/null
+++ b/docs/opcodes/opcode-44-aget.html
@@ -0,0 +1,113 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>aget<kind></title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>aget<kind></h1>
+
+<h2>Purpose</h2>
+
+<p>
+Perform the identified array operation at the identified index of the given
+array, storing into the value register.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>44..51 23x</td>
+ <td><i>arrayop</i> vAA, vBB, vCC<br/>
+ 44: aget<br/>
+ 45: aget-wide<br/>
+ 46: aget-object<br/>
+ 47: aget-boolean<br/>
+ 48: aget-byte<br/>
+ 49: aget-char<br/>
+ 4a: aget-short<br/>
+ </td>
+ <td><code>A:</code> dest value register or pair; (8 bits)<br/>
+ <code>B:</code> array register (8 bits)<br/>
+ <code>C:</code> index register (8 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A, B and C must be valid register indices in the current stackframe.
+ </li>
+ <li>
+ For the aget-wide variant, also A+1 must be a valid register index in the
+ current stackframe.
+ </li>
+ <li>
+ Register vB must contain an array reference. The component type of the
+ array must match the variant of the instruction.
+ </li>
+ <li>
+ Register vC must contain an integer value.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ For all but the -wide variant, the array element at the given index is moved
+ into register vA, that is, vA'=array[index].
+ </li>
+ <li>
+ For the -wide variant, the array element at the given index is moved into
+ registers vA and v(A+1) as follows:
+ <ul>
+ <li>
+ vA'=array[index] >> 0x20
+ </li>
+ <li>
+ v(A+1)'=array[index] & 0xffffffff;
+ </li>
+ </ul>
+ </li>
+ <li>
+ If v(A-1) is the lower half of a register pair, v(A-1)' becomes undefined.
+ </li>
+ <li>
+ For all but the -wide variant, if v(A+1) is the upper half of a register
+ pair, v(A+1)' becomes undefined.
+ </li>
+ <li>
+ For the -wide variant, if v(A+2) is the upper half of a register pair,
+ v(A+2)' becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ NullPointerException if vB=null.
+ </li>
+ <li>
+ ArrayIndexOutOfBoundsException if vC < 0 or vC >= array.length.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-4b-aput.html b/docs/opcodes/opcode-4b-aput.html
new file mode 100644
index 0000000..089c1ca
--- /dev/null
+++ b/docs/opcodes/opcode-4b-aput.html
@@ -0,0 +1,98 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>aput<kind></title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>aput<kind></h1>
+
+<h2>Purpose</h2>
+
+<p>
+Waste cycles.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>4b..51 23x</td>
+ <td><i>arrayop</i> vAA, vBB, vCC<br/>
+ 4b: aput<br/>
+ 4c: aput-wide<br/>
+ 4d: aput-object<br/>
+ 4e: aput-boolean<br/>
+ 4f: aput-byte<br/>
+ 50: aput-char<br/>
+ 51: aput-short
+ </td>
+ <td><code>A:</code> source value register or pair; (8 bits)<br/>
+ <code>B:</code> array register (8 bits)<br/>
+ <code>C:</code> index register (8 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A, B and C must be valid register indices in the current stack frame.
+ </li>
+ <li>
+ For the aget-wide variant, also A+1 must be a valid register index in the
+ current stack frame.
+ </li>
+ <li>
+ Register vB must contain an array reference. The component type of the array
+ must match the variant of the instruction.
+ </li>
+ <li>
+ Register vC must contain an integer value.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ For all but the -wide variant, the value of register vA is move into the
+ array element at the given index, that is, array[index]'=vA.
+ </li>
+ <li>
+ For the -wide variant, the registers vA and v(A+1) are moved into the array
+ element at the given index as follows:
+ <ul>
+ <li>
+ array[index]' = vA << 0x20 | v(A+1)
+ </li>
+ </ul>
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ NullPointerException if vB=null.
+ </li>
+ <li>
+ ArrayIndexOutOfBoundsException if vC < 0 or vC >= array.length.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-52-iget.html b/docs/opcodes/opcode-52-iget.html
new file mode 100644
index 0000000..837b511
--- /dev/null
+++ b/docs/opcodes/opcode-52-iget.html
@@ -0,0 +1,109 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>iget<kind></title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>iget<kind></h1>
+
+<h2>Purpose</h2>
+
+<p>
+Perform the identified object instance field operation with the identified
+field, loading or storing into the value register.
+</p>
+<p>
+Note: These opcodes are reasonable candidates for static linking, altering the
+field argument to be a more direct offset.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>52..58 22c</td>
+ <td>i<i>instanceop</i> vA, vB, field@CCCC<br/>
+ 52: iget<br/>
+ 53: iget-wide<br/>
+ 54: iget-object<br/>
+ 55: iget-boolean<br/>
+ 56: iget-byte<br/>
+ 57: iget-char<br/>
+ 58: iget-short<br/>
+ </td>
+ <td><code>A:</code> dest value register or pair; (4 bits)<br/>
+ <code>B:</code> object register (4 bits)<br/>
+ <code>C:</code> instance field reference index (16 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A and B must be valid register indices in the current stackframe.
+ </li>
+ <li>
+ For the -wide variant, also A+1 must be a valid register index in the
+ current stackframe.
+ </li>
+ <li>
+ Register vB must contain an object reference.
+ </li>
+ <li>
+ C must be a valid index into the field reference pool.
+ </li>
+ <li>
+ The field must be an instance field. The type of the field denoted by C must
+ match the variant of the instruction.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The value of the given instance field is read from the given object and
+ moved into the given register vA, that is, vA'=<object>.<field>.
+ </li>
+ <li>
+ If v(A-1) is the lower half of a register pair, v(A-1)' becomes undefined.
+ </li>
+ <li>For all but the -wide variant, if v(A+1) is the upper half of a register
+ pair, v(A+1)' becomes undefined.
+ </li>
+ <li>
+ For the -wide variant, if v(A+2) is the upper half of a register pair,
+ v(A+2)' becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ NullPointerException if object is null.
+ </li>
+ <li>
+ IllegalAccessException if <object>.<field> is not visible from
+ the current context according to the usual visibility and access rules of
+ the Java programming language.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-59-iput.html b/docs/opcodes/opcode-59-iput.html
new file mode 100644
index 0000000..22a3479
--- /dev/null
+++ b/docs/opcodes/opcode-59-iput.html
@@ -0,0 +1,112 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>iget<kind></title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>iget<kind></h1>
+
+<h2>Purpose</h2>
+
+<p>
+Perform the identified object instance field operation with the identified
+field, loading or storing into the value register.
+</p>
+<p>
+Note: These opcodes are reasonable candidates for static linking, altering the
+field argument to be a more direct offset.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>59..5f 22c</td>
+ <td>i<i>instanceop</i> vA, vB, field@CCCC<br/>
+ 59: iput<br/>
+ 5a: iput-wide<br/>
+ 5b: iput-object<br/>
+ 5c: iput-boolean<br/>
+ 5d: iput-byte<br/>
+ 5e: iput-char<br/>
+ 5f: iput-short
+ </td>
+ <td><code>A:</code> source value register or pair; (4 bits)<br/>
+ <code>B:</code> object register (4 bits)<br/>
+ <code>C:</code> instance field reference index (16 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A and B must be valid register indices in the current stack frame.
+ </li>
+ <li>
+ For the -wide variant, also A+1 must be a valid register index in the
+ current stack frame.
+ </li>
+ <li>
+ Register vB must contain an object reference.
+ </li>
+ <li>
+ C must be a valid index into the field reference pool.
+ </li>
+ <li>
+ The field must be an instance field. The type of the field denoted by C must
+ match the variant of the instruction.
+ </li>
+ <li>
+ For the -object variant, the instance referenced by register vA must be
+ assignment-compatible to the type of the field.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ For all but the -wide variant, the value of register vA is move into the
+ field, that is, <object>.<field>'=vA.
+ </li>
+ <li>
+ For the -wide variant, the registers vA and v(A+1) are moved into the
+ field as follows:
+ <ul>
+ <li>
+ <object>.<field>' = vA << 0x20 | v(A+1)
+ </li>
+ </ul>
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ NullPointerException if vB=null.
+ </li>
+ <li>
+ IllegalAccessException if <object>.<field> is not visible from
+ the current context according to the usual visibility and access rules of
+ the Java programming language, or final.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-60-sget.html b/docs/opcodes/opcode-60-sget.html
new file mode 100644
index 0000000..820886e
--- /dev/null
+++ b/docs/opcodes/opcode-60-sget.html
@@ -0,0 +1,106 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>sget<kind></title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>sget<kind></h1>
+
+<h2>Purpose</h2>
+
+<p>
+Perform the identified object static field operation with the identified static
+field, loading or storing into the value register.
+</p>
+<p>
+Note: These opcodes are reasonable candidates for static linking, altering the
+field argument to be a more direct offset.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>60..6d 21c</td>
+ <td>s<i>staticop</i> vAA, field@BBBB<br/>
+ 60: sget<br/>
+ 61: sget-wide<br/>
+ 62: sget-object<br/>
+ 63: sget-boolean<br/>
+ 64: sget-byte<br/>
+ 65: sget-char<br/>
+ 66: sget-short<br/>
+ </td>
+ <td><code>A:</code> dest value register or pair; (8 bits)<br/>
+ <code>B:</code> static field reference index (16 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index in the current stackframe.
+ </li>
+ <li>
+ For the -wide variant, also A+1 must be a valid register index in the
+ current stackframe.
+ </li>
+ <li>
+ B must be a valid index into the field reference pool.
+ </li>
+ <li>
+ The field denoted by B must be static. The type of the field denoted by B
+ must match the variant of the instruction.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The value of the given instance field is read from the given object and
+ moved into the given register vA, that is, vA'=<class>.<field>.
+ </li>
+ <li>
+ If v(A-1) is the lower half of a register pair, v(A-1)' becomes undefined.
+ </li>
+ <li>
+ For all but the -wide variant, if v(A+1) is the upper half of a register
+ pair, v(A+1)' becomes undefined.
+ </li>
+ <li>
+ For the -wide variant, if v(A+2) is the upper half of a register pair,
+ v(A+2)' becomes undefined.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ NullPointerException if object is null.
+ </li>
+ <li>
+ IllegalAccessException if <object>.<field> is not visible from
+ the current context according to the usual visibility and access rules of
+ the Java programming language.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-67-sput.html b/docs/opcodes/opcode-67-sput.html
new file mode 100644
index 0000000..b4d88bb
--- /dev/null
+++ b/docs/opcodes/opcode-67-sput.html
@@ -0,0 +1,108 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>sput<kind></title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>sput<kind></h1>
+
+<h2>Purpose</h2>
+
+<p>
+Perform the identified object static field operation with the identified static
+field, loading or storing into the value register.
+</p>
+<p>
+Note: These opcodes are reasonable candidates for static linking, altering the
+field argument to be a more direct offset.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>67..6d 21c</td>
+ <td>s<i>staticop</i> vAA, field@BBBB<br/>
+ 67: sput<br/>
+ 68: sput-wide<br/>
+ 69: sput-object<br/>
+ 6a: sput-boolean<br/>
+ 6b: sput-byte<br/>
+ 6c: sput-char<br/>
+ 6d: sput-short
+ </td>
+ <td><code>A:</code> source value register or pair; (8 bits)<br/>
+ <code>B:</code> static field reference index (16 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ A must be a valid register index in the current stackframe.
+ </li>
+ <li>
+ For the -wide variant, also A+1 must be a valid register index in the
+ current stackframe.
+ </li>
+ <li>
+ B must be a valid index into the field reference pool.
+ </li>
+ <li>
+ The field must be static. The type of the field denoted by C must match the
+ variant of the instruction.
+ </li>
+ <li>
+ For the -object variant, the instance referenced by register vA must be
+ assignment-compatible to the type of the field.
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ For all but the -wide variant, the value of register vA is move into the
+ field, that is, <class>.<field>'=vA.
+ </li>
+ <li>
+ For the -wide variant, the registers vA and v(A+1) are moved into the field
+ as follows:
+ <ul>
+ <li>
+ <class>.<field>' = vA << 0x20 | v(A+1)
+ </li>
+ </ul>
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ NullPointerException if vB=null.
+ </li>
+ <li>
+ IllegalAccessException if <object>.<field> is not visible from
+ the current context according to the usual visibility and access rules of
+ the Java programming language, or final.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-7b-unop.html b/docs/opcodes/opcode-7b-unop.html
new file mode 100644
index 0000000..1068a25
--- /dev/null
+++ b/docs/opcodes/opcode-7b-unop.html
@@ -0,0 +1,108 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>unop</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>unop</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Perform the identified unary operation on the source register, storing the
+result in the destination register.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>7b..8f 12x</td>
+ <td><i>unop</i> vA, vB<br/>
+ 7b: neg-int<br/>
+ 7c: not-int<br/>
+ 7d: neg-long<br/>
+ 7e: not-long<br/>
+ 7f: neg-float<br/>
+ 80: neg-double<br/>
+ 81: int-to-long<br/>
+ 82: int-to-float<br/>
+ 83: int-to-double<br/>
+ 84: long-to-int<br/>
+ 85: long-to-float<br/>
+ 86: long-to-double<br/>
+ 87: float-to-int<br/>
+ 88: float-to-long<br/>
+ 89: float-to-double<br/>
+ 8a: double-to-int<br/>
+ 8b: double-to-long<br/>
+ 8c: double-to-float<br/>
+ 8d: int-to-byte<br/>
+ 8e: int-to-char<br/>
+ 8f: int-to-short
+ </td>
+ <td><code>A:</code> destination register or pair (4 bits)<br/>
+ <code>B:</code> source register or pair (4 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ Both A and B must be valid register indices for the current stackframe.
+ </li>
+ <li>
+ If the input type of <unop> is double or long, also B+1 must be a
+ valid register index in the current stackframe.
+ </li>
+ <li>
+ If the output type of <unop> is double or long, also A+1 must be a
+ valid register index in the current stackframe.
+ </li>
+ <li>
+ The type of register vB must match the source type of the instruction (this
+ probably needs more detail).
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The given operation <unop> is performed according to the semantics
+ specified in table XXX.
+ </li>
+ <li>
+ The result is stored in register vA, that is, vA'=<unop> vB.
+ </li>
+ <li>
+ It gets a bit messy if we want to describe all the combinations of input and
+ output with and without pairs here. Probably it's better to split it up.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ ArithmeticException if an arithmetic error occurs during the instruction.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-90-binop.html b/docs/opcodes/opcode-90-binop.html
new file mode 100644
index 0000000..cdc08a8
--- /dev/null
+++ b/docs/opcodes/opcode-90-binop.html
@@ -0,0 +1,120 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>binop</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>binop</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Perform the identified binary operation on the two source registers, storing
+the result in the first source register.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>90..af 23x</td>
+ <td><i>binop</i> vAA, vBB, vCC<br/>
+ 90: add-int<br/>
+ 91: sub-int<br/>
+ 92: mul-int<br/>
+ 93: div-int<br/>
+ 94: rem-int<br/>
+ 95: and-int<br/>
+ 96: or-int<br/>
+ 97: xor-int<br/>
+ 98: shl-int<br/>
+ 99: shr-int<br/>
+ 9a: ushr-int<br/>
+ 9b: add-long<br/>
+ 9c: sub-long<br/>
+ 9d: mul-long<br/>
+ 9e: div-long<br/>
+ 9f: rem-long<br/>
+ a0: and-long<br/>
+ a1: or-long<br/>
+ a2: xor-long<br/>
+ a3: shl-long<br/>
+ a4: shr-long<br/>
+ a5: ushr-long<br/>
+ a6: add-float<br/>
+ a7: sub-float<br/>
+ a8: mul-float<br/>
+ a9: div-float<br/>
+ aa: rem-float<br/>
+ ab: add-double<br/>
+ ac: sub-double<br/>
+ ad: mul-double<br/>
+ ae: div-double<br/>
+ af: rem-double
+ </td>
+ <td><code>A:</code> destination register or pair (8 bits)<br/>
+ <code>B:</code> first source register or pair (8 bits)<br/>
+ <code>C:</code> second source register or pair (8 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ All A, B and C must be valid register indices in the current stackframe.
+ </li>
+ <li>
+ For the -long and -double variants, also A+1, B+1 and C+1 must be valid
+ register indices.
+ </li>
+ <li>
+ Registers vB and vC must be defined. They must both contain values that
+ match the variant of the instruction (it's probably better to split this up
+ into multiple pages again).
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The given operation <binop> is performed according to the semantics
+ specified in table XXX.
+ </li>
+ <li>
+ The result is stored in register vA, that is, vA'=<biop> vB.
+ </li>
+ <li>
+ For the -double and -long variants, (vA+1) is also affected.
+ </li>
+ <li>
+ As usual, neighboring registers might get undefined, if vA (and vA+1) were
+ part of a register pair originally.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ ArithmeticException if an error occurs during the instruction.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-b0-binop-2addr.html b/docs/opcodes/opcode-b0-binop-2addr.html
new file mode 100644
index 0000000..b3374f4
--- /dev/null
+++ b/docs/opcodes/opcode-b0-binop-2addr.html
@@ -0,0 +1,120 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>binop/2addr</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>binop/2addr</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Perform the identified binary operation on the two source registers, storing the
+result in the first source register.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>b0..cf 12x</td>
+ <td><i>binop</i>/2addr vA, vB<br/>
+ b0: add-int/2addr<br/>
+ b1: sub-int/2addr<br/>
+ b2: mul-int/2addr<br/>
+ b3: div-int/2addr<br/>
+ b4: rem-int/2addr<br/>
+ b5: and-int/2addr<br/>
+ b6: or-int/2addr<br/>
+ b7: xor-int/2addr<br/>
+ b8: shl-int/2addr<br/>
+ b9: shr-int/2addr<br/>
+ ba: ushr-int/2addr<br/>
+ bb: add-long/2addr<br/>
+ bc: sub-long/2addr<br/>
+ bd: mul-long/2addr<br/>
+ be: div-long/2addr<br/>
+ bf: rem-long/2addr<br/>
+ c0: and-long/2addr<br/>
+ c1: or-long/2addr<br/>
+ c2: xor-long/2addr<br/>
+ c3: shl-long/2addr<br/>
+ c4: shr-long/2addr<br/>
+ c5: ushr-long/2addr<br/>
+ c6: add-float/2addr<br/>
+ c7: sub-float/2addr<br/>
+ c8: mul-float/2addr<br/>
+ c9: div-float/2addr<br/>
+ ca: rem-float/2addr<br/>
+ cb: add-double/2addr<br/>
+ cc: sub-double/2addr<br/>
+ cd: mul-double/2addr<br/>
+ ce: div-double/2addr<br/>
+ cf: rem-double/2addr
+ </td>
+ <td><code>A:</code> destination and first source register or pair
+ (4 bits)<br/>
+ <code>B:</code> second source register or pair (4 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ Both A and B must be valid register indices in the current stackframe.
+ </li>
+ <li>
+ For the -long and -double variants, also A+1 and B+1 must be valid register
+ indices.
+ </li>
+ <li>
+ Registers vA and vB must be defined. They must both contain values that
+ match the variant of the instruction (it's probably better to split this up
+ into multiple pages again).
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The given operation <binop> is performed according to the semantics
+ specified in table XXX.
+ </li>
+ <li>
+ The result is stored in register vA, that is, vA'=vA <binop> vB.
+ </li>
+ <li>
+ For the -double and -long variants, (vA+1) is also affected.
+ </li>
+ <li>
+ As usual, neighboring registers might get undefined, if vA (and vA+1) were
+ part of a register pair originally.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ ArithmeticException if an error occurs during the instruction.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-d0-binop-lit16.html b/docs/opcodes/opcode-d0-binop-lit16.html
new file mode 100644
index 0000000..f9d3327
--- /dev/null
+++ b/docs/opcodes/opcode-d0-binop-lit16.html
@@ -0,0 +1,94 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>binop/lit16</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>binop/lit16</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Perform the indicated binary op on the indicated register (first argument) and
+literal value (second argument), storing the result in the destination register.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>d0..d7 22s</td>
+ <td><i>binop</i>/lit16 vA, vB, #+CCCC<br/>
+ d0: add-int/lit16<br/>
+ d1: rsub-int (reverse subtract)<br/>
+ d2: mul-int/lit16<br/>
+ d3: div-int/lit16<br/>
+ d4: rem-int/lit16<br/>
+ d5: and-int/lit16<br/>
+ d6: or-int/lit16<br/>
+ d7: xor-int/lit16
+ </td>
+ <td><code>A:</code> destination register (4 bits)<br/>
+ <code>B:</code> source register (4 bits)<br/>
+ <code>C:</code> signed int constant (16 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ Both A and B must be valid register indices in the current stackframe.
+ </li>
+ <li>
+ Registers vA and vB must be defined. They must both contain integer values.
+ </li>
+ <li>
+ C is an immediate, signed integer constant taken from the instruction stream
+ (actually this means there are no special requirements for C at all).
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The given operation <binop> is performed according to the semantics
+ specified in table XXX.
+ </li>
+ <li>
+ Argument C is sign-extended to 32 bits before.
+ </li>
+ <li>
+ The result is stored in register vA, that is, vA'=vB <binop> vC.
+ </li>
+ <li>
+ As usual, neighboring registers might get undefined, if vA was part of a
+ register pair originally.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ ArithmeticException if an error occurs during the instruction.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode-d8-binop-lit8.html b/docs/opcodes/opcode-d8-binop-lit8.html
new file mode 100644
index 0000000..26005e9
--- /dev/null
+++ b/docs/opcodes/opcode-d8-binop-lit8.html
@@ -0,0 +1,97 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
+
+<html>
+
+<head>
+<title>binop/lit8</title>
+<link rel=stylesheet href="opcode.css">
+</head>
+
+<body>
+
+<h1>binop/lit8</h1>
+
+<h2>Purpose</h2>
+
+<p>
+Perform the indicated binary op on the indicated register (first argument) and
+literal value (second argument), storing the result in the destination register.
+</p>
+
+<h2>Details</h2>
+
+<table class="instruc">
+<thead>
+<tr>
+ <th>Op & Format</th>
+ <th>Mnemonic / Syntax</th>
+ <th>Arguments</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+ <td>d8..e2 22b</td>
+ <td><i>binop</i>/lit8 vAA, vBB, #+CC<br/>
+ d8: add-int/lit8<br/>
+ d9: rsub-int/lit8<br/>
+ da: mul-int/lit8<br/>
+ db: div-int/lit8<br/>
+ dc: rem-int/lit8<br/>
+ dd: and-int/lit8<br/>
+ de: or-int/lit8<br/>
+ df: xor-int/lit8<br/>
+ e0: shl-int/lit8<br/>
+ e1: shr-int/lit8<br/>
+ e2: ushr-int/lit8
+ </td>
+ <td><code>A:</code> destination register (8 bits)<br/>
+ <code>B:</code> source register (8 bits)<br/>
+ <code>C:</code> signed int constant (8 bits)</td>
+</tr>
+</tbody>
+</table>
+
+<h2>Constraints</h2>
+
+<ul>
+ <li>
+ Both A and B must be valid register indices in the current stackframe.
+ </li>
+ <li>
+ Registers vA and vB must be defined. They must both contain integer values.
+ </li>
+ <li>
+ C is an immediate, signed integer constant taken from the instruction stream
+ (actually this means there are no special requirements for C at all).
+ </li>
+</ul>
+
+<h2>Behavior</h2>
+
+<ul>
+ <li>
+ The given operation <binop> is performed according to the semantics
+ specified in table XXX.
+ </li>
+ <li>
+ Argument C is sign-extended to 32 bits before.
+ </li>
+ <li>
+ The result is stored in register vA, that is, vA'=vB <binop> vC.
+ </li>
+ <li>
+ As usual, neighboring registers might get undefined, if vA was part of a
+ register pair originally.
+ </li>
+</ul>
+
+<h2>Exceptions</h2>
+
+<ul>
+ <li>
+ ArithmeticException if an error occurs during the instruction.
+ </li>
+</ul>
+
+</body>
+</html>
diff --git a/docs/opcodes/opcode.css b/docs/opcodes/opcode.css
new file mode 100644
index 0000000..c3c1304
--- /dev/null
+++ b/docs/opcodes/opcode.css
@@ -0,0 +1,166 @@
+h1 {
+ font-family: serif;
+ color: #222266;
+}
+
+h2 {
+ font-family: serif;
+ border-top-style: solid;
+ border-top-width: 2px;
+ border-color: #ccccdd;
+ padding-top: 12px;
+ margin-top: 48px;
+ margin-bottom: 2px;
+ color: #222266;
+}
+
+@media print {
+ table {
+ font-size: 8pt;
+ }
+}
+
+@media screen {
+ table {
+ font-size: 10pt;
+ }
+}
+
+
+/* general for all tables */
+
+table {
+ border-collapse: collapse;
+ margin-top: 12px;
+}
+
+table th {
+ font-family: sans-serif;
+ background: #aabbff;
+ text-align: left;
+}
+
+table td {
+ font-family: sans-serif;
+ border-top-style: solid;
+ border-bottom-style: solid;
+ border-width: 1px;
+ border-color: #aaaaff;
+ padding-top: 4px;
+ padding-bottom: 4px;
+ padding-left: 4px;
+ padding-right: 6px;
+ background: #eeeeff;
+}
+
+table td p {
+ margin-top: 4pt;
+ margin-bottom: 0pt;
+}
+
+
+
+/* opcodes table */
+
+table.instruc {
+ margin-top: 24px;
+ margin-bottom: 24px;
+ margin-left: 48px;
+ margin-right: 48px;
+}
+
+table.instruc td {
+ font-family: sans-serif;
+ border-top-style: solid;
+ border-bottom-style: solid;
+ border-width: 1px;
+ padding-top: 4px;
+ padding-bottom: 4px;
+ padding-left: 2px;
+ padding-right: 2px;
+}
+
+table.instruc td:first-child {
+ font-family: monospace;
+ font-size: 90%;
+ vertical-align: top;
+ width: 12%;
+}
+
+table.instruc td:first-child + td {
+ font-family: monospace;
+ font-size: 90%;
+ vertical-align: top;
+ width: 23%;
+}
+
+table.instruc td:first-child + td i {
+ font-family: sans-serif;
+ font-size: 90%;
+}
+
+table.instruc td:first-child + td + td {
+ vertical-align: top;
+ width: 28%;
+}
+
+table.instruc td:first-child + td + td + td {
+ vertical-align: top;
+ width: 37%;
+}
+
+
+/* supplemental opcode format table */
+
+table.supplement {
+ margin-top: 24px;
+ margin-bottom: 24px;
+ margin-left: 48px;
+ margin-right: 48px;
+}
+
+table.supplement td:first-child {
+ font-family: monospace;
+ vertical-align: top;
+ width: 20%;
+}
+
+table.supplement td:first-child + td {
+ font-family: monospace;
+ vertical-align: top;
+ width: 20%;
+}
+
+table.supplement td:first-child + td + td {
+ font-family: sans-serif;
+ vertical-align: top;
+ width: 60%;
+}
+
+
+/* math details table */
+
+table.math {
+ margin-top: 24px;
+ margin-bottom: 24px;
+ margin-left: 48px;
+ margin-right: 48px;
+}
+
+table.math td:first-child {
+ font-family: monospace;
+ vertical-align: top;
+ width: 10%;
+}
+
+table.math td:first-child + td {
+ font-family: monospace;
+ vertical-align: top;
+ width: 30%;
+}
+
+table.math td:first-child + td + td {
+ font-family: sans-serif;
+ vertical-align: top;
+ width: 60%;
+}
diff --git a/docs/prettify.css b/docs/prettify.css
new file mode 100644
index 0000000..351152b
--- /dev/null
+++ b/docs/prettify.css
@@ -0,0 +1,27 @@
+/* Pretty printing styles. Used with prettify.js. */
+
+.str { color: #080; }
+.kwd { color: #008; }
+.com { color: #800; }
+.typ { color: #606; }
+.lit { color: #066; }
+.pun { color: #660; }
+.pln { color: #000; }
+.tag { color: #008; }
+.atn { color: #606; }
+.atv { color: #080; }
+.dec { color: #606; }
+pre.prettyprint { padding: 2px; border: 1px solid #888; }
+
+@media print {
+ .str { color: #060; }
+ .kwd { color: #006; font-weight: bold; }
+ .com { color: #600; font-style: italic; }
+ .typ { color: #404; font-weight: bold; }
+ .lit { color: #044; }
+ .pun { color: #440; }
+ .pln { color: #000; }
+ .tag { color: #006; font-weight: bold; }
+ .atn { color: #404; }
+ .atv { color: #060; }
+}
diff --git a/docs/prettify.js b/docs/prettify.js
new file mode 100644
index 0000000..c954118
--- /dev/null
+++ b/docs/prettify.js
@@ -0,0 +1,1280 @@
+// Copyright (C) 2006 Google Inc.
+//
+// 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.
+
+
+/**
+ * @fileoverview
+ * some functions for browser-side pretty printing of code contained in html.
+ *
+ * The lexer should work on a number of languages including C and friends,
+ * Java, Python, Bash, SQL, HTML, XML, CSS, Javascript, and Makefiles.
+ * It works passably on Ruby, PHP and Awk and a decent subset of Perl, but,
+ * because of commenting conventions, doesn't work on Smalltalk, Lisp-like, or
+ * CAML-like languages.
+ *
+ * If there's a language not mentioned here, then I don't know it, and don't
+ * know whether it works. If it has a C-like, Bash-like, or XML-like syntax
+ * then it should work passably.
+ *
+ * Usage:
+ * 1) include this source file in an html page via
+ * <script type="text/javascript" src="/path/to/prettify.js"></script>
+ * 2) define style rules. See the example page for examples.
+ * 3) mark the <pre> and <code> tags in your source with class=prettyprint.
+ * You can also use the (html deprecated) <xmp> tag, but the pretty printer
+ * needs to do more substantial DOM manipulations to support that, so some
+ * css styles may not be preserved.
+ * That's it. I wanted to keep the API as simple as possible, so there's no
+ * need to specify which language the code is in.
+ *
+ * Change log:
+ * cbeust, 2006/08/22
+ * Java annotations (start with "@") are now captured as literals ("lit")
+ */
+
+// JSLint declarations
+/*global console, document, navigator, setTimeout, window */
+
+/**
+ * Split {@code prettyPrint} into multiple timeouts so as not to interfere with
+ * UI events.
+ * If set to {@code false}, {@code prettyPrint()} is synchronous.
+ */
+var PR_SHOULD_USE_CONTINUATION = true;
+
+/** the number of characters between tab columns */
+var PR_TAB_WIDTH = 8;
+
+/** Walks the DOM returning a properly escaped version of innerHTML.
+ * @param {Node} node
+ * @param {Array.<string>} out output buffer that receives chunks of HTML.
+ */
+var PR_normalizedHtml;
+
+/** Contains functions for creating and registering new language handlers.
+ * @type {Object}
+ */
+var PR;
+
+/** Pretty print a chunk of code.
+ *
+ * @param {string} sourceCodeHtml code as html
+ * @return {string} code as html, but prettier
+ */
+var prettyPrintOne;
+/** find all the < pre > and < code > tags in the DOM with class=prettyprint
+ * and prettify them.
+ * @param {Function} opt_whenDone if specified, called when the last entry
+ * has been finished.
+ */
+var prettyPrint;
+
+/** browser detection. @extern */
+function _pr_isIE6() {
+ var isIE6 = navigator && navigator.userAgent &&
+ /\bMSIE 6\./.test(navigator.userAgent);
+ _pr_isIE6 = function () { return isIE6; };
+ return isIE6;
+}
+
+
+(function () {
+ /** Splits input on space and returns an Object mapping each non-empty part to
+ * true.
+ */
+ function wordSet(words) {
+ words = words.split(/ /g);
+ var set = {};
+ for (var i = words.length; --i >= 0;) {
+ var w = words[i];
+ if (w) { set[w] = null; }
+ }
+ return set;
+ }
+
+ // Keyword lists for various languages.
+ var FLOW_CONTROL_KEYWORDS =
+ "break continue do else for if return while ";
+ var C_KEYWORDS = FLOW_CONTROL_KEYWORDS + "auto case char const default " +
+ "double enum extern float goto int long register short signed sizeof " +
+ "static struct switch typedef union unsigned void volatile ";
+ var COMMON_KEYWORDS = C_KEYWORDS + "catch class delete false import " +
+ "new operator private protected public this throw true try ";
+ var CPP_KEYWORDS = COMMON_KEYWORDS + "alignof align_union asm axiom bool " +
+ "concept concept_map const_cast constexpr decltype " +
+ "dynamic_cast explicit export friend inline late_check " +
+ "mutable namespace nullptr reinterpret_cast static_assert static_cast " +
+ "template typeid typename typeof using virtual wchar_t where ";
+ var JAVA_KEYWORDS = COMMON_KEYWORDS +
+ "boolean byte extends final finally implements import instanceof null " +
+ "native package strictfp super synchronized throws transient ";
+ var CSHARP_KEYWORDS = JAVA_KEYWORDS +
+ "as base by checked decimal delegate descending event " +
+ "fixed foreach from group implicit in interface internal into is lock " +
+ "object out override orderby params readonly ref sbyte sealed " +
+ "stackalloc string select uint ulong unchecked unsafe ushort var ";
+ var JSCRIPT_KEYWORDS = COMMON_KEYWORDS +
+ "debugger eval export function get null set undefined var with " +
+ "Infinity NaN ";
+ var PERL_KEYWORDS = "caller delete die do dump elsif eval exit foreach for " +
+ "goto if import last local my next no our print package redo require " +
+ "sub undef unless until use wantarray while BEGIN END ";
+ var PYTHON_KEYWORDS = FLOW_CONTROL_KEYWORDS + "and as assert class def del " +
+ "elif except exec finally from global import in is lambda " +
+ "nonlocal not or pass print raise try with yield " +
+ "False True None ";
+ var RUBY_KEYWORDS = FLOW_CONTROL_KEYWORDS + "alias and begin case class def" +
+ " defined elsif end ensure false in module next nil not or redo rescue " +
+ "retry self super then true undef unless until when yield BEGIN END ";
+ var SH_KEYWORDS = FLOW_CONTROL_KEYWORDS + "case done elif esac eval fi " +
+ "function in local set then until ";
+ var ALL_KEYWORDS = (
+ CPP_KEYWORDS + CSHARP_KEYWORDS + JSCRIPT_KEYWORDS + PERL_KEYWORDS +
+ PYTHON_KEYWORDS + RUBY_KEYWORDS + SH_KEYWORDS);
+
+ // token style names. correspond to css classes
+ /** token style for a string literal */
+ var PR_STRING = 'str';
+ /** token style for a keyword */
+ var PR_KEYWORD = 'kwd';
+ /** token style for a comment */
+ var PR_COMMENT = 'com';
+ /** token style for a type */
+ var PR_TYPE = 'typ';
+ /** token style for a literal value. e.g. 1, null, true. */
+ var PR_LITERAL = 'lit';
+ /** token style for a punctuation string. */
+ var PR_PUNCTUATION = 'pun';
+ /** token style for a punctuation string. */
+ var PR_PLAIN = 'pln';
+
+ /** token style for an sgml tag. */
+ var PR_TAG = 'tag';
+ /** token style for a markup declaration such as a DOCTYPE. */
+ var PR_DECLARATION = 'dec';
+ /** token style for embedded source. */
+ var PR_SOURCE = 'src';
+ /** token style for an sgml attribute name. */
+ var PR_ATTRIB_NAME = 'atn';
+ /** token style for an sgml attribute value. */
+ var PR_ATTRIB_VALUE = 'atv';
+
+ /**
+ * A class that indicates a section of markup that is not code, e.g. to allow
+ * embedding of line numbers within code listings.
+ */
+ var PR_NOCODE = 'nocode';
+
+ function isWordChar(ch) {
+ return (ch >= 'a' && ch <= 'z') || (ch >= 'A' && ch <= 'Z');
+ }
+
+ /** Splice one array into another.
+ * Like the python <code>
+ * container[containerPosition:containerPosition + countReplaced] = inserted
+ * </code>
+ * @param {Array} inserted
+ * @param {Array} container modified in place
+ * @param {Number} containerPosition
+ * @param {Number} countReplaced
+ */
+ function spliceArrayInto(
+ inserted, container, containerPosition, countReplaced) {
+ inserted.unshift(containerPosition, countReplaced || 0);
+ try {
+ container.splice.apply(container, inserted);
+ } finally {
+ inserted.splice(0, 2);
+ }
+ }
+
+ /** A set of tokens that can precede a regular expression literal in
+ * javascript.
+ * http://www.mozilla.org/js/language/js20/rationale/syntax.html has the full
+ * list, but I've removed ones that might be problematic when seen in
+ * languages that don't support regular expression literals.
+ *
+ * <p>Specifically, I've removed any keywords that can't precede a regexp
+ * literal in a syntactically legal javascript program, and I've removed the
+ * "in" keyword since it's not a keyword in many languages, and might be used
+ * as a count of inches.
+ * @private
+ */
+ var REGEXP_PRECEDER_PATTERN = function () {
+ var preceders = [
+ "!", "!=", "!==", "#", "%", "%=", "&", "&&", "&&=",
+ "&=", "(", "*", "*=", /* "+", */ "+=", ",", /* "-", */ "-=",
+ "->", /*".", "..", "...", handled below */ "/", "/=", ":", "::", ";",
+ "<", "<<", "<<=", "<=", "=", "==", "===", ">",
+ ">=", ">>", ">>=", ">>>", ">>>=", "?", "@", "[",
+ "^", "^=", "^^", "^^=", "{", "|", "|=", "||",
+ "||=", "~" /* handles =~ and !~ */,
+ "break", "case", "continue", "delete",
+ "do", "else", "finally", "instanceof",
+ "return", "throw", "try", "typeof"
+ ];
+ var pattern = '(?:' +
+ '(?:(?:^|[^0-9.])\\.{1,3})|' + // a dot that's not part of a number
+ '(?:(?:^|[^\\+])\\+)|' + // allow + but not ++
+ '(?:(?:^|[^\\-])-)'; // allow - but not --
+ for (var i = 0; i < preceders.length; ++i) {
+ var preceder = preceders[i];
+ if (isWordChar(preceder.charAt(0))) {
+ pattern += '|\\b' + preceder;
+ } else {
+ pattern += '|' + preceder.replace(/([^=<>:&])/g, '\\$1');
+ }
+ }
+ pattern += '|^)\\s*$'; // matches at end, and matches empty string
+ return new RegExp(pattern);
+ // CAVEAT: this does not properly handle the case where a regular
+ // expression immediately follows another since a regular expression may
+ // have flags for case-sensitivity and the like. Having regexp tokens
+ // adjacent is not
+ // valid in any language I'm aware of, so I'm punting.
+ // TODO: maybe style special characters inside a regexp as punctuation.
+ }();
+
+ // Define regexps here so that the interpreter doesn't have to create an
+ // object each time the function containing them is called.
+ // The language spec requires a new object created even if you don't access
+ // the $1 members.
+ var pr_amp = /&/g;
+ var pr_lt = /</g;
+ var pr_gt = />/g;
+ var pr_quot = /\"/g;
+ /** like textToHtml but escapes double quotes to be attribute safe. */
+ function attribToHtml(str) {
+ return str.replace(pr_amp, '&')
+ .replace(pr_lt, '<')
+ .replace(pr_gt, '>')
+ .replace(pr_quot, '"');
+ }
+
+ /** escapest html special characters to html. */
+ function textToHtml(str) {
+ return str.replace(pr_amp, '&')
+ .replace(pr_lt, '<')
+ .replace(pr_gt, '>');
+ }
+
+
+ var pr_ltEnt = /</g;
+ var pr_gtEnt = />/g;
+ var pr_aposEnt = /'/g;
+ var pr_quotEnt = /"/g;
+ var pr_ampEnt = /&/g;
+ var pr_nbspEnt = / /g;
+ /** unescapes html to plain text. */
+ function htmlToText(html) {
+ var pos = html.indexOf('&');
+ if (pos < 0) { return html; }
+ // Handle numeric entities specially. We can't use functional substitution
+ // since that doesn't work in older versions of Safari.
+ // These should be rare since most browsers convert them to normal chars.
+ for (--pos; (pos = html.indexOf('&#', pos + 1)) >= 0;) {
+ var end = html.indexOf(';', pos);
+ if (end >= 0) {
+ var num = html.substring(pos + 3, end);
+ var radix = 10;
+ if (num && num.charAt(0) === 'x') {
+ num = num.substring(1);
+ radix = 16;
+ }
+ var codePoint = parseInt(num, radix);
+ if (!isNaN(codePoint)) {
+ html = (html.substring(0, pos) + String.fromCharCode(codePoint) +
+ html.substring(end + 1));
+ }
+ }
+ }
+
+ return html.replace(pr_ltEnt, '<')
+ .replace(pr_gtEnt, '>')
+ .replace(pr_aposEnt, "'")
+ .replace(pr_quotEnt, '"')
+ .replace(pr_ampEnt, '&')
+ .replace(pr_nbspEnt, ' ');
+ }
+
+ /** is the given node's innerHTML normally unescaped? */
+ function isRawContent(node) {
+ return 'XMP' === node.tagName;
+ }
+
+ function normalizedHtml(node, out) {
+ switch (node.nodeType) {
+ case 1: // an element
+ var name = node.tagName.toLowerCase();
+ out.push('<', name);
+ for (var i = 0; i < node.attributes.length; ++i) {
+ var attr = node.attributes[i];
+ if (!attr.specified) { continue; }
+ out.push(' ');
+ normalizedHtml(attr, out);
+ }
+ out.push('>');
+ for (var child = node.firstChild; child; child = child.nextSibling) {
+ normalizedHtml(child, out);
+ }
+ if (node.firstChild || !/^(?:br|link|img)$/.test(name)) {
+ out.push('<\/', name, '>');
+ }
+ break;
+ case 2: // an attribute
+ out.push(node.name.toLowerCase(), '="', attribToHtml(node.value), '"');
+ break;
+ case 3: case 4: // text
+ out.push(textToHtml(node.nodeValue));
+ break;
+ }
+ }
+
+ var PR_innerHtmlWorks = null;
+ function getInnerHtml(node) {
+ // inner html is hopelessly broken in Safari 2.0.4 when the content is
+ // an html description of well formed XML and the containing tag is a PRE
+ // tag, so we detect that case and emulate innerHTML.
+ if (null === PR_innerHtmlWorks) {
+ var testNode = document.createElement('PRE');
+ testNode.appendChild(
+ document.createTextNode('<!DOCTYPE foo PUBLIC "foo bar">\n<foo />'));
+ PR_innerHtmlWorks = !/</.test(testNode.innerHTML);
+ }
+
+ if (PR_innerHtmlWorks) {
+ var content = node.innerHTML;
+ // XMP tags contain unescaped entities so require special handling.
+ if (isRawContent(node)) {
+ content = textToHtml(content);
+ }
+ return content;
+ }
+
+ var out = [];
+ for (var child = node.firstChild; child; child = child.nextSibling) {
+ normalizedHtml(child, out);
+ }
+ return out.join('');
+ }
+
+ /** returns a function that expand tabs to spaces. This function can be fed
+ * successive chunks of text, and will maintain its own internal state to
+ * keep track of how tabs are expanded.
+ * @return {function (string) : string} a function that takes
+ * plain text and return the text with tabs expanded.
+ * @private
+ */
+ function makeTabExpander(tabWidth) {
+ var SPACES = ' ';
+ var charInLine = 0;
+
+ return function (plainText) {
+ // walk over each character looking for tabs and newlines.
+ // On tabs, expand them. On newlines, reset charInLine.
+ // Otherwise increment charInLine
+ var out = null;
+ var pos = 0;
+ for (var i = 0, n = plainText.length; i < n; ++i) {
+ var ch = plainText.charAt(i);
+
+ switch (ch) {
+ case '\t':
+ if (!out) { out = []; }
+ out.push(plainText.substring(pos, i));
+ // calculate how much space we need in front of this part
+ // nSpaces is the amount of padding -- the number of spaces needed
+ // to move us to the next column, where columns occur at factors of
+ // tabWidth.
+ var nSpaces = tabWidth - (charInLine % tabWidth);
+ charInLine += nSpaces;
+ for (; nSpaces >= 0; nSpaces -= SPACES.length) {
+ out.push(SPACES.substring(0, nSpaces));
+ }
+ pos = i + 1;
+ break;
+ case '\n':
+ charInLine = 0;
+ break;
+ default:
+ ++charInLine;
+ }
+ }
+ if (!out) { return plainText; }
+ out.push(plainText.substring(pos));
+ return out.join('');
+ };
+ }
+
+ // The below pattern matches one of the following
+ // (1) /[^<]+/ : A run of characters other than '<'
+ // (2) /<!--.*?-->/: an HTML comment
+ // (3) /<!\[CDATA\[.*?\]\]>/: a cdata section
+ // (3) /<\/?[a-zA-Z][^>]*>/ : A probably tag that should not be highlighted
+ // (4) /</ : A '<' that does not begin a larger chunk. Treated as 1
+ var pr_chunkPattern =
+ /(?:[^<]+|<!--[\s\S]*?-->|<!\[CDATA\[([\s\S]*?)\]\]>|<\/?[a-zA-Z][^>]*>|<)/g;
+ var pr_commentPrefix = /^<!--/;
+ var pr_cdataPrefix = /^<\[CDATA\[/;
+ var pr_brPrefix = /^<br\b/i;
+ var pr_tagNameRe = /^<(\/?)([a-zA-Z]+)/;
+
+ /** split markup into chunks of html tags (style null) and
+ * plain text (style {@link #PR_PLAIN}), converting tags which are
+ * significant for tokenization (<br>) into their textual equivalent.
+ *
+ * @param {string} s html where whitespace is considered significant.
+ * @return {Object} source code and extracted tags.
+ * @private
+ */
+ function extractTags(s) {
+ // since the pattern has the 'g' modifier and defines no capturing groups,
+ // this will return a list of all chunks which we then classify and wrap as
+ // PR_Tokens
+ var matches = s.match(pr_chunkPattern);
+ var sourceBuf = [];
+ var sourceBufLen = 0;
+ var extractedTags = [];
+ if (matches) {
+ for (var i = 0, n = matches.length; i < n; ++i) {
+ var match = matches[i];
+ if (match.length > 1 && match.charAt(0) === '<') {
+ if (pr_commentPrefix.test(match)) { continue; }
+ if (pr_cdataPrefix.test(match)) {
+ // strip CDATA prefix and suffix. Don't unescape since it's CDATA
+ sourceBuf.push(match.substring(9, match.length - 3));
+ sourceBufLen += match.length - 12;
+ } else if (pr_brPrefix.test(match)) {
+ // <br> tags are lexically significant so convert them to text.
+ // This is undone later.
+ sourceBuf.push('\n');
+ ++sourceBufLen;
+ } else {
+ if (match.indexOf(PR_NOCODE) >= 0 && isNoCodeTag(match)) {
+ // A <span class="nocode"> will start a section that should be
+ // ignored. Continue walking the list until we see a matching end
+ // tag.
+ var name = match.match(pr_tagNameRe)[2];
+ var depth = 1;
+ end_tag_loop:
+ for (var j = i + 1; j < n; ++j) {
+ var name2 = matches[j].match(pr_tagNameRe);
+ if (name2 && name2[2] === name) {
+ if (name2[1] === '/') {
+ if (--depth === 0) { break end_tag_loop; }
+ } else {
+ ++depth;
+ }
+ }
+ }
+ if (j < n) {
+ extractedTags.push(
+ sourceBufLen, matches.slice(i, j + 1).join(''));
+ i = j;
+ } else { // Ignore unclosed sections.
+ extractedTags.push(sourceBufLen, match);
+ }
+ } else {
+ extractedTags.push(sourceBufLen, match);
+ }
+ }
+ } else {
+ var literalText = htmlToText(match);
+ sourceBuf.push(literalText);
+ sourceBufLen += literalText.length;
+ }
+ }
+ }
+ return { source: sourceBuf.join(''), tags: extractedTags };
+ }
+
+ /** True if the given tag contains a class attribute with the nocode class. */
+ function isNoCodeTag(tag) {
+ return !!tag
+ // First canonicalize the representation of attributes
+ .replace(/\s(\w+)\s*=\s*(?:\"([^\"]*)\"|'([^\']*)'|(\S+))/g,
+ ' $1="$2$3$4"')
+ // Then look for the attribute we want.
+ .match(/[cC][lL][aA][sS][sS]=\"[^\"]*\bnocode\b/);
+ }
+
+ /** Given triples of [style, pattern, context] returns a lexing function,
+ * The lexing function interprets the patterns to find token boundaries and
+ * returns a decoration list of the form
+ * [index_0, style_0, index_1, style_1, ..., index_n, style_n]
+ * where index_n is an index into the sourceCode, and style_n is a style
+ * constant like PR_PLAIN. index_n-1 <= index_n, and style_n-1 applies to
+ * all characters in sourceCode[index_n-1:index_n].
+ *
+ * The stylePatterns is a list whose elements have the form
+ * [style : string, pattern : RegExp, context : RegExp, shortcut : string].
+ &
+ * Style is a style constant like PR_PLAIN.
+ *
+ * Pattern must only match prefixes, and if it matches a prefix and context
+ * is null or matches the last non-comment token parsed, then that match is
+ * considered a token with the same style.
+ *
+ * Context is applied to the last non-whitespace, non-comment token
+ * recognized.
+ *
+ * Shortcut is an optional string of characters, any of which, if the first
+ * character, gurantee that this pattern and only this pattern matches.
+ *
+ * @param {Array} shortcutStylePatterns patterns that always start with
+ * a known character. Must have a shortcut string.
+ * @param {Array} fallthroughStylePatterns patterns that will be tried in
+ * order if the shortcut ones fail. May have shortcuts.
+ *
+ * @return {function (string, number?) : Array.<number|string>} a
+ * function that takes source code and returns a list of decorations.
+ */
+ function createSimpleLexer(shortcutStylePatterns,
+ fallthroughStylePatterns) {
+ var shortcuts = {};
+ (function () {
+ var allPatterns = shortcutStylePatterns.concat(fallthroughStylePatterns);
+ for (var i = allPatterns.length; --i >= 0;) {
+ var patternParts = allPatterns[i];
+ var shortcutChars = patternParts[3];
+ if (shortcutChars) {
+ for (var c = shortcutChars.length; --c >= 0;) {
+ shortcuts[shortcutChars.charAt(c)] = patternParts;
+ }
+ }
+ }
+ })();
+
+ var nPatterns = fallthroughStylePatterns.length;
+ var notWs = /\S/;
+
+ return function (sourceCode, opt_basePos) {
+ opt_basePos = opt_basePos || 0;
+ var decorations = [opt_basePos, PR_PLAIN];
+ var lastToken = '';
+ var pos = 0; // index into sourceCode
+ var tail = sourceCode;
+
+ while (tail.length) {
+ var style;
+ var token = null;
+ var match;
+
+ var patternParts = shortcuts[tail.charAt(0)];
+ if (patternParts) {
+ match = tail.match(patternParts[1]);
+ token = match[0];
+ style = patternParts[0];
+ } else {
+ for (var i = 0; i < nPatterns; ++i) {
+ patternParts = fallthroughStylePatterns[i];
+ var contextPattern = patternParts[2];
+ if (contextPattern && !contextPattern.test(lastToken)) {
+ // rule can't be used
+ continue;
+ }
+ match = tail.match(patternParts[1]);
+ if (match) {
+ token = match[0];
+ style = patternParts[0];
+ break;
+ }
+ }
+
+ if (!token) { // make sure that we make progress
+ style = PR_PLAIN;
+ token = tail.substring(0, 1);
+ }
+ }
+
+ decorations.push(opt_basePos + pos, style);
+ pos += token.length;
+ tail = tail.substring(token.length);
+ if (style !== PR_COMMENT && notWs.test(token)) { lastToken = token; }
+ }
+ return decorations;
+ };
+ }
+
+ var PR_MARKUP_LEXER = createSimpleLexer([], [
+ [PR_PLAIN, /^[^<]+/, null],
+ [PR_DECLARATION, /^<!\w[^>]*(?:>|$)/, null],
+ [PR_COMMENT, /^<!--[\s\S]*?(?:-->|$)/, null],
+ [PR_SOURCE, /^<\?[\s\S]*?(?:\?>|$)/, null],
+ [PR_SOURCE, /^<%[\s\S]*?(?:%>|$)/, null],
+ [PR_SOURCE,
+ // Tags whose content is not escaped, and which contain source code.
+ /^<(script|style|xmp)\b[^>]*>[\s\S]*?<\/\1\b[^>]*>/i, null],
+ [PR_TAG, /^<\/?\w[^<>]*>/, null]
+ ]);
+ // Splits any of the source|style|xmp entries above into a start tag,
+ // source content, and end tag.
+ var PR_SOURCE_CHUNK_PARTS = /^(<[^>]*>)([\s\S]*)(<\/[^>]*>)$/;
+ /** split markup on tags, comments, application directives, and other top
+ * level constructs. Tags are returned as a single token - attributes are
+ * not yet broken out.
+ * @private
+ */
+ function tokenizeMarkup(source) {
+ var decorations = PR_MARKUP_LEXER(source);
+ for (var i = 0; i < decorations.length; i += 2) {
+ if (decorations[i + 1] === PR_SOURCE) {
+ var start, end;
+ start = decorations[i];
+ end = i + 2 < decorations.length ? decorations[i + 2] : source.length;
+ // Split out start and end script tags as actual tags, and leave the
+ // body with style SCRIPT.
+ var sourceChunk = source.substring(start, end);
+ var match = sourceChunk.match(PR_SOURCE_CHUNK_PARTS);
+ if (match) {
+ decorations.splice(
+ i, 2,
+ start, PR_TAG, // the open chunk
+ start + match[1].length, PR_SOURCE,
+ start + match[1].length + (match[2] || '').length, PR_TAG);
+ }
+ }
+ }
+ return decorations;
+ }
+
+ var PR_TAG_LEXER = createSimpleLexer([
+ [PR_ATTRIB_VALUE, /^\'[^\']*(?:\'|$)/, null, "'"],
+ [PR_ATTRIB_VALUE, /^\"[^\"]*(?:\"|$)/, null, '"'],
+ [PR_PUNCTUATION, /^[<>\/=]+/, null, '<>/=']
+ ], [
+ [PR_TAG, /^[\w:\-]+/, /^</],
+ [PR_ATTRIB_VALUE, /^[\w\-]+/, /^=/],
+ [PR_ATTRIB_NAME, /^[\w:\-]+/, null],
+ [PR_PLAIN, /^\s+/, null, ' \t\r\n']
+ ]);
+ /** split tags attributes and their values out from the tag name, and
+ * recursively lex source chunks.
+ * @private
+ */
+ function splitTagAttributes(source, decorations) {
+ for (var i = 0; i < decorations.length; i += 2) {
+ var style = decorations[i + 1];
+ if (style === PR_TAG) {
+ var start, end;
+ start = decorations[i];
+ end = i + 2 < decorations.length ? decorations[i + 2] : source.length;
+ var chunk = source.substring(start, end);
+ var subDecorations = PR_TAG_LEXER(chunk, start);
+ spliceArrayInto(subDecorations, decorations, i, 2);
+ i += subDecorations.length - 2;
+ }
+ }
+ return decorations;
+ }
+
+ /** returns a function that produces a list of decorations from source text.
+ *
+ * This code treats ", ', and ` as string delimiters, and \ as a string
+ * escape. It does not recognize perl's qq() style strings.
+ * It has no special handling for double delimiter escapes as in basic, or
+ * the tripled delimiters used in python, but should work on those regardless
+ * although in those cases a single string literal may be broken up into
+ * multiple adjacent string literals.
+ *
+ * It recognizes C, C++, and shell style comments.
+ *
+ * @param {Object} options a set of optional parameters.
+ * @return {function (string) : Array.<string|number>} a
+ * decorator that takes sourceCode as plain text and that returns a
+ * decoration list
+ */
+ function sourceDecorator(options) {
+ var shortcutStylePatterns = [], fallthroughStylePatterns = [];
+ if (options.tripleQuotedStrings) {
+ // '''multi-line-string''', 'single-line-string', and double-quoted
+ shortcutStylePatterns.push(
+ [PR_STRING, /^(?:\'\'\'(?:[^\'\\]|\\[\s\S]|\'{1,2}(?=[^\']))*(?:\'\'\'|$)|\"\"\"(?:[^\"\\]|\\[\s\S]|\"{1,2}(?=[^\"]))*(?:\"\"\"|$)|\'(?:[^\\\']|\\[\s\S])*(?:\'|$)|\"(?:[^\\\"]|\\[\s\S])*(?:\"|$))/,
+ null, '\'"']);
+ } else if (options.multiLineStrings) {
+ // 'multi-line-string', "multi-line-string"
+ shortcutStylePatterns.push(
+ [PR_STRING, /^(?:\'(?:[^\\\']|\\[\s\S])*(?:\'|$)|\"(?:[^\\\"]|\\[\s\S])*(?:\"|$)|\`(?:[^\\\`]|\\[\s\S])*(?:\`|$))/,
+ null, '\'"`']);
+ } else {
+ // 'single-line-string', "single-line-string"
+ shortcutStylePatterns.push(
+ [PR_STRING,
+ /^(?:\'(?:[^\\\'\r\n]|\\.)*(?:\'|$)|\"(?:[^\\\"\r\n]|\\.)*(?:\"|$))/,
+ null, '"\'']);
+ }
+ fallthroughStylePatterns.push(
+ [PR_PLAIN, /^(?:[^\'\"\`\/\#]+)/, null, ' \r\n']);
+ if (options.hashComments) {
+ shortcutStylePatterns.push([PR_COMMENT, /^#[^\r\n]*/, null, '#']);
+ }
+ if (options.cStyleComments) {
+ fallthroughStylePatterns.push([PR_COMMENT, /^\/\/[^\r\n]*/, null]);
+ fallthroughStylePatterns.push(
+ [PR_COMMENT, /^\/\*[\s\S]*?(?:\*\/|$)/, null]);
+ }
+ if (options.regexLiterals) {
+ var REGEX_LITERAL = (
+ // A regular expression literal starts with a slash that is
+ // not followed by * or / so that it is not confused with
+ // comments.
+ '^/(?=[^/*])'
+ // and then contains any number of raw characters,
+ + '(?:[^/\\x5B\\x5C]'
+ // escape sequences (\x5C),
+ + '|\\x5C[\\s\\S]'
+ // or non-nesting character sets (\x5B\x5D);
+ + '|\\x5B(?:[^\\x5C\\x5D]|\\x5C[\\s\\S])*(?:\\x5D|$))+'
+ // finally closed by a /.
+ + '(?:/|$)');
+ fallthroughStylePatterns.push(
+ [PR_STRING, new RegExp(REGEX_LITERAL), REGEXP_PRECEDER_PATTERN]);
+ }
+
+ var keywords = wordSet(options.keywords);
+
+ options = null;
+
+ /** splits the given string into comment, string, and "other" tokens.
+ * @param {string} sourceCode as plain text
+ * @return {Array.<number|string>} a decoration list.
+ * @private
+ */
+ var splitStringAndCommentTokens = createSimpleLexer(
+ shortcutStylePatterns, fallthroughStylePatterns);
+
+ var styleLiteralIdentifierPuncRecognizer = createSimpleLexer([], [
+ [PR_PLAIN, /^\s+/, null, ' \r\n'],
+ // TODO(mikesamuel): recognize non-latin letters and numerals in idents
+ [PR_PLAIN, /^[a-z_$@][a-z_$@0-9]*/i, null],
+ // A hex number
+ [PR_LITERAL, /^0x[a-f0-9]+[a-z]/i, null],
+ // An octal or decimal number, possibly in scientific notation
+ [PR_LITERAL,
+ /^(?:\d(?:_\d+)*\d*(?:\.\d*)?|\.\d+)(?:e[+\-]?\d+)?[a-z]*/i,
+ null, '123456789'],
+ [PR_PUNCTUATION, /^[^\s\w\.$@]+/, null]
+ // Fallback will handle decimal points not adjacent to a digit
+ ]);
+
+ /** splits plain text tokens into more specific tokens, and then tries to
+ * recognize keywords, and types.
+ * @private
+ */
+ function splitNonStringNonCommentTokens(source, decorations) {
+ for (var i = 0; i < decorations.length; i += 2) {
+ var style = decorations[i + 1];
+ if (style === PR_PLAIN) {
+ var start, end, chunk, subDecs;
+ start = decorations[i];
+ end = i + 2 < decorations.length ? decorations[i + 2] : source.length;
+ chunk = source.substring(start, end);
+ subDecs = styleLiteralIdentifierPuncRecognizer(chunk, start);
+ for (var j = 0, m = subDecs.length; j < m; j += 2) {
+ var subStyle = subDecs[j + 1];
+ if (subStyle === PR_PLAIN) {
+ var subStart = subDecs[j];
+ var subEnd = j + 2 < m ? subDecs[j + 2] : chunk.length;
+ var token = source.substring(subStart, subEnd);
+ if (token === '.') {
+ subDecs[j + 1] = PR_PUNCTUATION;
+ } else if (token in keywords) {
+ subDecs[j + 1] = PR_KEYWORD;
+ } else if (/^@?[A-Z][A-Z$]*[a-z][A-Za-z$]*$/.test(token)) {
+ // classify types and annotations using Java's style conventions
+ subDecs[j + 1] = token.charAt(0) === '@' ? PR_LITERAL : PR_TYPE;
+ }
+ }
+ }
+ spliceArrayInto(subDecs, decorations, i, 2);
+ i += subDecs.length - 2;
+ }
+ }
+ return decorations;
+ }
+
+ return function (sourceCode) {
+ // Split into strings, comments, and other.
+ // We do this because strings and comments are easily recognizable and can
+ // contain stuff that looks like other tokens, so we want to mark those
+ // early so we don't recurse into them.
+ var decorations = splitStringAndCommentTokens(sourceCode);
+
+ // Split non comment|string tokens on whitespace and word boundaries
+ decorations = splitNonStringNonCommentTokens(sourceCode, decorations);
+
+ return decorations;
+ };
+ }
+
+ var decorateSource = sourceDecorator({
+ keywords: ALL_KEYWORDS,
+ hashComments: true,
+ cStyleComments: true,
+ multiLineStrings: true,
+ regexLiterals: true
+ });
+
+ /** identify regions of markup that are really source code, and recursivley
+ * lex them.
+ * @private
+ */
+ function splitSourceNodes(source, decorations) {
+ for (var i = 0; i < decorations.length; i += 2) {
+ var style = decorations[i + 1];
+ if (style === PR_SOURCE) {
+ // Recurse using the non-markup lexer
+ var start, end;
+ start = decorations[i];
+ end = i + 2 < decorations.length ? decorations[i + 2] : source.length;
+ var subDecorations = decorateSource(source.substring(start, end));
+ for (var j = 0, m = subDecorations.length; j < m; j += 2) {
+ subDecorations[j] += start;
+ }
+ spliceArrayInto(subDecorations, decorations, i, 2);
+ i += subDecorations.length - 2;
+ }
+ }
+ return decorations;
+ }
+
+ /** identify attribute values that really contain source code and recursively
+ * lex them.
+ * @private
+ */
+ function splitSourceAttributes(source, decorations) {
+ var nextValueIsSource = false;
+ for (var i = 0; i < decorations.length; i += 2) {
+ var style = decorations[i + 1];
+ var start, end;
+ if (style === PR_ATTRIB_NAME) {
+ start = decorations[i];
+ end = i + 2 < decorations.length ? decorations[i + 2] : source.length;
+ nextValueIsSource = /^on|^style$/i.test(source.substring(start, end));
+ } else if (style === PR_ATTRIB_VALUE) {
+ if (nextValueIsSource) {
+ start = decorations[i];
+ end = i + 2 < decorations.length ? decorations[i + 2] : source.length;
+ var attribValue = source.substring(start, end);
+ var attribLen = attribValue.length;
+ var quoted =
+ (attribLen >= 2 && /^[\"\']/.test(attribValue) &&
+ attribValue.charAt(0) === attribValue.charAt(attribLen - 1));
+
+ var attribSource;
+ var attribSourceStart;
+ var attribSourceEnd;
+ if (quoted) {
+ attribSourceStart = start + 1;
+ attribSourceEnd = end - 1;
+ attribSource = attribValue;
+ } else {
+ attribSourceStart = start + 1;
+ attribSourceEnd = end - 1;
+ attribSource = attribValue.substring(1, attribValue.length - 1);
+ }
+
+ var attribSourceDecorations = decorateSource(attribSource);
+ for (var j = 0, m = attribSourceDecorations.length; j < m; j += 2) {
+ attribSourceDecorations[j] += attribSourceStart;
+ }
+
+ if (quoted) {
+ attribSourceDecorations.push(attribSourceEnd, PR_ATTRIB_VALUE);
+ spliceArrayInto(attribSourceDecorations, decorations, i + 2, 0);
+ } else {
+ spliceArrayInto(attribSourceDecorations, decorations, i, 2);
+ }
+ }
+ nextValueIsSource = false;
+ }
+ }
+ return decorations;
+ }
+
+ /** returns a decoration list given a string of markup.
+ *
+ * This code recognizes a number of constructs.
+ * <!-- ... --> comment
+ * <!\w ... > declaration
+ * <\w ... > tag
+ * </\w ... > tag
+ * <?...?> embedded source
+ * <%...%> embedded source
+ * &[#\w]...; entity
+ *
+ * It does not recognizes %foo; doctype entities from .
+ *
+ * It will recurse into any <style>, <script>, and on* attributes using
+ * PR_lexSource.
+ */
+ function decorateMarkup(sourceCode) {
+ // This function works as follows:
+ // 1) Start by splitting the markup into text and tag chunks
+ // Input: string s
+ // Output: List<PR_Token> where style in (PR_PLAIN, null)
+ // 2) Then split the text chunks further into comments, declarations,
+ // tags, etc.
+ // After each split, consider whether the token is the start of an
+ // embedded source section, i.e. is an open <script> tag. If it is, find
+ // the corresponding close token, and don't bother to lex in between.
+ // Input: List<string>
+ // Output: List<PR_Token> with style in
+ // (PR_TAG, PR_PLAIN, PR_SOURCE, null)
+ // 3) Finally go over each tag token and split out attribute names and
+ // values.
+ // Input: List<PR_Token>
+ // Output: List<PR_Token> where style in
+ // (PR_TAG, PR_PLAIN, PR_SOURCE, NAME, VALUE, null)
+ var decorations = tokenizeMarkup(sourceCode);
+ decorations = splitTagAttributes(sourceCode, decorations);
+ decorations = splitSourceNodes(sourceCode, decorations);
+ decorations = splitSourceAttributes(sourceCode, decorations);
+ return decorations;
+ }
+
+ /**
+ * @param {string} sourceText plain text
+ * @param {Array.<number|string>} extractedTags chunks of raw html preceded
+ * by their position in sourceText in order.
+ * @param {Array.<number|string>} decorations style classes preceded by their
+ * position in sourceText in order.
+ * @return {string} html
+ * @private
+ */
+ function recombineTagsAndDecorations(sourceText, extractedTags, decorations) {
+ var html = [];
+ // index past the last char in sourceText written to html
+ var outputIdx = 0;
+
+ var openDecoration = null;
+ var currentDecoration = null;
+ var tagPos = 0; // index into extractedTags
+ var decPos = 0; // index into decorations
+ var tabExpander = makeTabExpander(PR_TAB_WIDTH);
+
+ var adjacentSpaceRe = /([\r\n ]) /g;
+ var startOrSpaceRe = /(^| ) /gm;
+ var newlineRe = /\r\n?|\n/g;
+ var trailingSpaceRe = /[ \r\n]$/;
+ var lastWasSpace = true; // the last text chunk emitted ended with a space.
+
+ // A helper function that is responsible for opening sections of decoration
+ // and outputing properly escaped chunks of source
+ function emitTextUpTo(sourceIdx) {
+ if (sourceIdx > outputIdx) {
+ if (openDecoration && openDecoration !== currentDecoration) {
+ // Close the current decoration
+ html.push('</span>');
+ openDecoration = null;
+ }
+ if (!openDecoration && currentDecoration) {
+ openDecoration = currentDecoration;
+ html.push('<span class="', openDecoration, '">');
+ }
+ // This interacts badly with some wikis which introduces paragraph tags
+ // into pre blocks for some strange reason.
+ // It's necessary for IE though which seems to lose the preformattedness
+ // of <pre> tags when their innerHTML is assigned.
+ // http://stud3.tuwien.ac.at/~e0226430/innerHtmlQuirk.html
+ // and it serves to undo the conversion of <br>s to newlines done in
+ // chunkify.
+ var htmlChunk = textToHtml(
+ tabExpander(sourceText.substring(outputIdx, sourceIdx)))
+ .replace(lastWasSpace
+ ? startOrSpaceRe
+ : adjacentSpaceRe, '$1 ');
+ // Keep track of whether we need to escape space at the beginning of the
+ // next chunk.
+ lastWasSpace = trailingSpaceRe.test(htmlChunk);
+ html.push(htmlChunk.replace(newlineRe, '<br />'));
+ outputIdx = sourceIdx;
+ }
+ }
+
+ while (true) {
+ // Determine if we're going to consume a tag this time around. Otherwise
+ // we consume a decoration or exit.
+ var outputTag;
+ if (tagPos < extractedTags.length) {
+ if (decPos < decorations.length) {
+ // Pick one giving preference to extractedTags since we shouldn't open
+ // a new style that we're going to have to immediately close in order
+ // to output a tag.
+ outputTag = extractedTags[tagPos] <= decorations[decPos];
+ } else {
+ outputTag = true;
+ }
+ } else {
+ outputTag = false;
+ }
+ // Consume either a decoration or a tag or exit.
+ if (outputTag) {
+ emitTextUpTo(extractedTags[tagPos]);
+ if (openDecoration) {
+ // Close the current decoration
+ html.push('</span>');
+ openDecoration = null;
+ }
+ html.push(extractedTags[tagPos + 1]);
+ tagPos += 2;
+ } else if (decPos < decorations.length) {
+ emitTextUpTo(decorations[decPos]);
+ currentDecoration = decorations[decPos + 1];
+ decPos += 2;
+ } else {
+ break;
+ }
+ }
+ emitTextUpTo(sourceText.length);
+ if (openDecoration) {
+ html.push('</span>');
+ }
+
+ return html.join('');
+ }
+
+ /** Maps language-specific file extensions to handlers. */
+ var langHandlerRegistry = {};
+ /** Register a language handler for the given file extensions.
+ * @param {function (string) : Array.<number|string>} handler
+ * a function from source code to a list of decorations.
+ * @param {Array.<string>} fileExtensions
+ */
+ function registerLangHandler(handler, fileExtensions) {
+ for (var i = fileExtensions.length; --i >= 0;) {
+ var ext = fileExtensions[i];
+ if (!langHandlerRegistry.hasOwnProperty(ext)) {
+ langHandlerRegistry[ext] = handler;
+ } else if ('console' in window) {
+ console.log('cannot override language handler %s', ext);
+ }
+ }
+ }
+ registerLangHandler(decorateSource, ['default-code']);
+ registerLangHandler(decorateMarkup,
+ ['default-markup', 'html', 'htm', 'xhtml', 'xml', 'xsl']);
+ registerLangHandler(sourceDecorator({
+ keywords: CPP_KEYWORDS,
+ hashComments: true,
+ cStyleComments: true
+ }), ['c', 'cc', 'cpp', 'cxx', 'cyc']);
+ registerLangHandler(sourceDecorator({
+ keywords: CSHARP_KEYWORDS,
+ hashComments: true,
+ cStyleComments: true
+ }), ['cs']);
+ registerLangHandler(sourceDecorator({
+ keywords: JAVA_KEYWORDS,
+ cStyleComments: true
+ }), ['java']);
+ registerLangHandler(sourceDecorator({
+ keywords: SH_KEYWORDS,
+ hashComments: true,
+ multiLineStrings: true
+ }), ['bsh', 'csh', 'sh']);
+ registerLangHandler(sourceDecorator({
+ keywords: PYTHON_KEYWORDS,
+ hashComments: true,
+ multiLineStrings: true,
+ tripleQuotedStrings: true
+ }), ['cv', 'py']);
+ registerLangHandler(sourceDecorator({
+ keywords: PERL_KEYWORDS,
+ hashComments: true,
+ multiLineStrings: true,
+ regexLiterals: true
+ }), ['perl', 'pl', 'pm']);
+ registerLangHandler(sourceDecorator({
+ keywords: RUBY_KEYWORDS,
+ hashComments: true,
+ multiLineStrings: true,
+ regexLiterals: true
+ }), ['rb']);
+ registerLangHandler(sourceDecorator({
+ keywords: JSCRIPT_KEYWORDS,
+ cStyleComments: true,
+ regexLiterals: true
+ }), ['js']);
+
+ function prettyPrintOne(sourceCodeHtml, opt_langExtension) {
+ try {
+ // Extract tags, and convert the source code to plain text.
+ var sourceAndExtractedTags = extractTags(sourceCodeHtml);
+ /** Plain text. @type {string} */
+ var source = sourceAndExtractedTags.source;
+
+ /** Even entries are positions in source in ascending order. Odd entries
+ * are tags that were extracted at that position.
+ * @type {Array.<number|string>}
+ */
+ var extractedTags = sourceAndExtractedTags.tags;
+
+ // Pick a lexer and apply it.
+ if (!langHandlerRegistry.hasOwnProperty(opt_langExtension)) {
+ // Treat it as markup if the first non whitespace character is a < and
+ // the last non-whitespace character is a >.
+ opt_langExtension =
+ /^\s*</.test(source) ? 'default-markup' : 'default-code';
+ }
+
+ /** Even entries are positions in source in ascending order. Odd enties
+ * are style markers (e.g., PR_COMMENT) that run from that position until
+ * the end.
+ * @type {Array.<number|string>}
+ */
+ var decorations = langHandlerRegistry[opt_langExtension].call({}, source);
+
+ // Integrate the decorations and tags back into the source code to produce
+ // a decorated html string.
+ return recombineTagsAndDecorations(source, extractedTags, decorations);
+ } catch (e) {
+ if ('console' in window) {
+ console.log(e);
+ console.trace();
+ }
+ return sourceCodeHtml;
+ }
+ }
+
+ function prettyPrint(opt_whenDone) {
+ var isIE6 = _pr_isIE6();
+
+ // fetch a list of nodes to rewrite
+ var codeSegments = [
+ document.getElementsByTagName('pre'),
+ document.getElementsByTagName('code'),
+ document.getElementsByTagName('xmp') ];
+ var elements = [];
+ for (var i = 0; i < codeSegments.length; ++i) {
+ for (var j = 0; j < codeSegments[i].length; ++j) {
+ elements.push(codeSegments[i][j]);
+ }
+ }
+ codeSegments = null;
+
+ // the loop is broken into a series of continuations to make sure that we
+ // don't make the browser unresponsive when rewriting a large page.
+ var k = 0;
+
+ function doWork() {
+ var endTime = (PR_SHOULD_USE_CONTINUATION ?
+ new Date().getTime() + 250 /* ms */ :
+ Infinity);
+ for (; k < elements.length && new Date().getTime() < endTime; k++) {
+ var cs = elements[k];
+ if (cs.className && cs.className.indexOf('prettyprint') >= 0) {
+ // If the classes includes a language extensions, use it.
+ // Language extensions can be specified like
+ // <pre class="prettyprint lang-cpp">
+ // the language extension "cpp" is used to find a language handler as
+ // passed to PR_registerLangHandler.
+ var langExtension = cs.className.match(/\blang-(\w+)\b/);
+ if (langExtension) { langExtension = langExtension[1]; }
+
+ // make sure this is not nested in an already prettified element
+ var nested = false;
+ for (var p = cs.parentNode; p; p = p.parentNode) {
+ if ((p.tagName === 'pre' || p.tagName === 'code' ||
+ p.tagName === 'xmp') &&
+ p.className && p.className.indexOf('prettyprint') >= 0) {
+ nested = true;
+ break;
+ }
+ }
+ if (!nested) {
+ // fetch the content as a snippet of properly escaped HTML.
+ // Firefox adds newlines at the end.
+ var content = getInnerHtml(cs);
+ content = content.replace(/(?:\r\n?|\n)$/, '');
+
+ // do the pretty printing
+ var newContent = prettyPrintOne(content, langExtension);
+
+ // push the prettified html back into the tag.
+ if (!isRawContent(cs)) {
+ // just replace the old html with the new
+ cs.innerHTML = newContent;
+ } else {
+ // we need to change the tag to a <pre> since <xmp>s do not allow
+ // embedded tags such as the span tags used to attach styles to
+ // sections of source code.
+ var pre = document.createElement('PRE');
+ for (var i = 0; i < cs.attributes.length; ++i) {
+ var a = cs.attributes[i];
+ if (a.specified) {
+ var aname = a.name.toLowerCase();
+ if (aname === 'class') {
+ pre.className = a.value; // For IE 6
+ } else {
+ pre.setAttribute(a.name, a.value);
+ }
+ }
+ }
+ pre.innerHTML = newContent;
+
+ // remove the old
+ cs.parentNode.replaceChild(pre, cs);
+ cs = pre;
+ }
+
+ // Replace <br>s with line-feeds so that copying and pasting works
+ // on IE 6.
+ // Doing this on other browsers breaks lots of stuff since \r\n is
+ // treated as two newlines on Firefox, and doing this also slows
+ // down rendering.
+ if (isIE6 && cs.tagName === 'PRE') {
+ var lineBreaks = cs.getElementsByTagName('br');
+ for (var j = lineBreaks.length; --j >= 0;) {
+ var lineBreak = lineBreaks[j];
+ lineBreak.parentNode.replaceChild(
+ document.createTextNode('\r\n'), lineBreak);
+ }
+ }
+ }
+ }
+ }
+ if (k < elements.length) {
+ // finish up in a continuation
+ setTimeout(doWork, 250);
+ } else if (opt_whenDone) {
+ opt_whenDone();
+ }
+ }
+
+ doWork();
+ }
+
+ window['PR_normalizedHtml'] = normalizedHtml;
+ window['prettyPrintOne'] = prettyPrintOne;
+ window['prettyPrint'] = prettyPrint;
+ window['PR'] = {
+ 'createSimpleLexer': createSimpleLexer,
+ 'registerLangHandler': registerLangHandler,
+ 'sourceDecorator': sourceDecorator,
+ 'PR_ATTRIB_NAME': PR_ATTRIB_NAME,
+ 'PR_ATTRIB_VALUE': PR_ATTRIB_VALUE,
+ 'PR_COMMENT': PR_COMMENT,
+ 'PR_DECLARATION': PR_DECLARATION,
+ 'PR_KEYWORD': PR_KEYWORD,
+ 'PR_LITERAL': PR_LITERAL,
+ 'PR_NOCODE': PR_NOCODE,
+ 'PR_PLAIN': PR_PLAIN,
+ 'PR_PUNCTUATION': PR_PUNCTUATION,
+ 'PR_SOURCE': PR_SOURCE,
+ 'PR_STRING': PR_STRING,
+ 'PR_TAG': PR_TAG,
+ 'PR_TYPE': PR_TYPE
+ };
+})();
diff --git a/docs/verifier.html b/docs/verifier.html
new file mode 100644
index 0000000..656b832
--- /dev/null
+++ b/docs/verifier.html
@@ -0,0 +1,159 @@
+<html>
+<head>
+<title>Dalvik Bytecode Verifier Notes</title>
+</head>
+
+<body>
+<h1>Dalvik Bytecode Verifier Notes</h1>
+
+<p>
+The bytecode verifier in the Dalvik VM attempts to provide the same sorts
+of checks and guarantees that other popular virtual machines do. We
+perform generally the same set of checks as are described in _The Java
+Virtual Machine Specification, Second Edition_, including the updates
+planned for the Third Edition.
+
+<p>
+Verification can be enabled for all classes, disabled for all, or enabled
+only for "remote" (non-bootstrap) classes. It should be performed for any
+class that will be processed with the DEX optimizer, and in fact the
+default VM behavior is to only optimize verified classes.
+
+
+<h2>Why Verify?</h2>
+
+<p>
+The verification process adds additional time to the build and to
+the installation of new applications. It's fairly quick for app-sized
+DEX files, but rather slow for the big "core" and "framework" files.
+Why do it all, when our system relies on UNIX processes for security?
+<p>
+<ol>
+ <li>Optimizations. The interpreter can ignore a lot of potential
+ error cases because the verifier guarantees that they are impossible.
+ Also, we can optimize the DEX file more aggressively if we start
+ with a stronger set of assumptions about the bytecode.
+ <li>"Exact" GC. The work peformed during verification has significant
+ overlap with the work required to compute register use maps for exact
+ GC. Improper register use, caught by the verifier, could lead to
+ subtle problems with an "exact" GC.
+ <li>Intra-application security. If an app wants to download bits
+ of interpreted code over the network and execute them, it can safely
+ do so using well-established security mechanisms.
+ <li>3rd party app failure analysis. We have no way to control the
+ tools and post-processing utilities that external developers employ,
+ so when we get bug reports with a weird exception or native crash
+ it's very helpful to start with the assumption that the bytecode
+ is valid.
+</ol>
+
+
+<h2>Verifier Differences</h2>
+
+<p>
+There are a few checks that the Dalvik bytecode verifier does not perform,
+because they're not relevant. For example:
+<ul>
+ <li>Type restrictions on constant pool references are not enforced,
+ because Dalvik does not have a pool of typed constants. (Dalvik
+ uses a simple index into type-specific pools.)
+ <li>Verification of the operand stack size is not performed, because
+ Dalvik does not have an operand stack.
+ <li>Limitations on <code>jsr</code> and <code>ret</code> do not apply,
+ because Dalvik doesn't support subroutines.
+</ul>
+
+In some cases they are implemented differently, e.g.:
+<ul>
+ <li>In a conventional VM, backward branches and exceptions are
+ forbidden when a local variable holds an uninitialized reference. The
+ restriction was changed to mark registers as invalid when they hold
+ references to the uninitialized result of a previous invocation of the
+ same <code>new-instance</code> instruction.
+ This solves the same problem -- trickery potentially allowing
+ uninitialized objects to slip past the verifier -- without unduly
+ limiting branches.
+</ul>
+
+There are also some new ones, such as:
+<ul>
+ <li>The <code>move-exception</code> instruction can only appear as
+ the first instruction in an exception handler.
+ <li>The <code>move-result*</code> instructions can only appear
+ immediately after an appropriate <code>invoke-*</code>
+ or <code>filled-new-array</code> instruction.
+</ul>
+
+<p>
+The Dalvik verifier is more restrictive than other VMs in one area:
+type safety on sub-32-bit integer widths. These additional restrictions
+should make it impossible to, say, pass a value outside the range
+[-128, 127] to a function that takes a <code>byte</code> as an argument.
+
+
+<h2>Verification Failures</h2>
+
+<p>
+When the verifier rejects a class, it always throws a VerifyError.
+This is different in some cases from other implementations. For example,
+if a class attempts to perform an illegal access on a field, the expected
+behavior is to receive an IllegalAccessError at runtime the first time
+the field is actually accessed. The Dalvik verifier will reject the
+entire class immediately.
+
+<p>
+It's difficult to throw the error on first use in Dalvik. Possible ways
+to implement this behavior include:
+
+<ol>
+<li>We could replace the invalid field access instruction with a special
+instruction that generates an illegal access error, and allow class
+verification to complete successfully. This type of verification must
+often be deferred to first class load, rather than be performed ahead of time
+during DEX optimization, which means the bytecode instructions will be
+mapped read-only during verification. So this won't work.
+</li>
+
+<li>We can perform the access checks when the field/method/class is
+resolved. In a typical VM implementation we would do the check when the
+entry is resolved in the context of the current classfile, but our DEX
+files combine multiple classfiles together, merging the field/method/class
+resolution results into a single large table. Once one class successfully
+resolves the field, every other class in the same DEX file would be able
+to access the field. This is bad.
+</li>
+
+<li>Perform the access checks on every field/method/class access.
+This adds significant overhead. This is mitigated somewhat by the DEX
+optimizer, which will convert many field/method/class accesses into a
+simpler form after performing the access check. However, not all accesses
+can be optimized (e.g. accesses to classes unknown at dexopt time),
+and we don't currently have an optimized form of certain instructions
+(notably static field operations).
+</li>
+</ol>
+
+<p>
+Other implementations are possible, but they all involve allocating
+some amount of additional memory or spending additional cycles
+on non-DEX-optimized instructions. We don't want to throw an
+IllegalAccessError at verification time, since that would indicate that
+access to the class being verified was illegal.
+<p>
+One approach that might be worth pursuing: for situations like illegal
+accesses, the verifier makes an in-RAM private copy of the method, and
+alters the instructions there. The class object is altered to point at
+the new copy of the instructions. This requires minimal memory overhead
+and provides a better experience for developers.
+
+<p>
+The VerifyError is accompanied by detailed, if somewhat cryptic,
+information in the log file. From this it's possible to determine the
+exact instruction that failed, and the reason for the failure. We can
+also constructor the VerifyError with an IllegalAccessError passed in as
+the cause.
+
+<address>Copyright © 2008 The Android Open Source Project</address>
+
+</body>
+</html>