/usr/share/doc/llvm-3.5-doc/html/ExtendingLLVM.html is in llvm-3.5-doc 1:3.5-10.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 | <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title>Extending LLVM: Adding instructions, intrinsics, types, etc. — LLVM 3.5 documentation</title>
<link rel="stylesheet" href="_static/llvm-theme.css" type="text/css" />
<link rel="stylesheet" href="_static/pygments.css" type="text/css" />
<script type="text/javascript">
var DOCUMENTATION_OPTIONS = {
URL_ROOT: './',
VERSION: '3.5',
COLLAPSE_INDEX: false,
FILE_SUFFIX: '.html',
HAS_SOURCE: true
};
</script>
<script type="text/javascript" src="_static/jquery.js"></script>
<script type="text/javascript" src="_static/underscore.js"></script>
<script type="text/javascript" src="_static/doctools.js"></script>
<link rel="top" title="LLVM 3.5 documentation" href="index.html" />
<link rel="next" title="How to set up LLVM-style RTTI for your class hierarchy" href="HowToSetUpLLVMStyleRTTI.html" />
<link rel="prev" title="Architecture & Platform Information for Compiler Writers" href="CompilerWriterInfo.html" />
<style type="text/css">
table.right { float: right; margin-left: 20px; }
table.right td { border: 1px solid #ccc; }
</style>
</head>
<body>
<div class="logo">
<a href="index.html">
<img src="_static/logo.png"
alt="LLVM Logo" width="250" height="88"/></a>
</div>
<div class="related">
<h3>Navigation</h3>
<ul>
<li class="right" style="margin-right: 10px">
<a href="genindex.html" title="General Index"
accesskey="I">index</a></li>
<li class="right" >
<a href="HowToSetUpLLVMStyleRTTI.html" title="How to set up LLVM-style RTTI for your class hierarchy"
accesskey="N">next</a> |</li>
<li class="right" >
<a href="CompilerWriterInfo.html" title="Architecture & Platform Information for Compiler Writers"
accesskey="P">previous</a> |</li>
<li><a href="http://llvm.org/">LLVM Home</a> | </li>
<li><a href="index.html">Documentation</a>»</li>
</ul>
</div>
<div class="document">
<div class="documentwrapper">
<div class="body">
<div class="section" id="extending-llvm-adding-instructions-intrinsics-types-etc">
<h1>Extending LLVM: Adding instructions, intrinsics, types, etc.<a class="headerlink" href="#extending-llvm-adding-instructions-intrinsics-types-etc" title="Permalink to this headline">¶</a></h1>
<div class="section" id="introduction-and-warning">
<h2>Introduction and Warning<a class="headerlink" href="#introduction-and-warning" title="Permalink to this headline">¶</a></h2>
<p>During the course of using LLVM, you may wish to customize it for your research
project or for experimentation. At this point, you may realize that you need to
add something to LLVM, whether it be a new fundamental type, a new intrinsic
function, or a whole new instruction.</p>
<p>When you come to this realization, stop and think. Do you really need to extend
LLVM? Is it a new fundamental capability that LLVM does not support at its
current incarnation or can it be synthesized from already pre-existing LLVM
elements? If you are not sure, ask on the <a class="reference external" href="http://mail.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVM-dev</a> list. The reason is that
extending LLVM will get involved as you need to update all the different passes
that you intend to use with your extension, and there are <tt class="docutils literal"><span class="pre">many</span></tt> LLVM analyses
and transformations, so it may be quite a bit of work.</p>
<p>Adding an <a class="reference internal" href="#intrinsic-function">intrinsic function</a> is far easier than adding an
instruction, and is transparent to optimization passes. If your added
functionality can be expressed as a function call, an intrinsic function is the
method of choice for LLVM extension.</p>
<p>Before you invest a significant amount of effort into a non-trivial extension,
<strong>ask on the list</strong> if what you are looking to do can be done with
already-existing infrastructure, or if maybe someone else is already working on
it. You will save yourself a lot of time and effort by doing so.</p>
</div>
<div class="section" id="adding-a-new-intrinsic-function">
<span id="intrinsic-function"></span><h2>Adding a new intrinsic function<a class="headerlink" href="#adding-a-new-intrinsic-function" title="Permalink to this headline">¶</a></h2>
<p>Adding a new intrinsic function to LLVM is much easier than adding a new
instruction. Almost all extensions to LLVM should start as an intrinsic
function and then be turned into an instruction if warranted.</p>
<ol class="arabic">
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/docs/LangRef.html</span></tt>:</p>
<p>Document the intrinsic. Decide whether it is code generator specific and
what the restrictions are. Talk to other people about it so that you are
sure it’s a good idea.</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/IR/Intrinsics*.td</span></tt>:</p>
<p>Add an entry for your intrinsic. Describe its memory access characteristics
for optimization (this controls whether it will be DCE’d, CSE’d, etc). Note
that any intrinsic using the <tt class="docutils literal"><span class="pre">llvm_int_ty</span></tt> type for an argument will
be deemed by <tt class="docutils literal"><span class="pre">tblgen</span></tt> as overloaded and the corresponding suffix will
be required on the intrinsic’s name.</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/Analysis/ConstantFolding.cpp</span></tt>:</p>
<p>If it is possible to constant fold your intrinsic, add support to it in the
<tt class="docutils literal"><span class="pre">canConstantFoldCallTo</span></tt> and <tt class="docutils literal"><span class="pre">ConstantFoldCall</span></tt> functions.</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/test/Regression/*</span></tt>:</p>
<p>Add test cases for your test cases to the test suite</p>
</li>
</ol>
<p>Once the intrinsic has been added to the system, you must add code generator
support for it. Generally you must do the following steps:</p>
<p>Add support to the .td file for the target(s) of your choice in
<tt class="docutils literal"><span class="pre">lib/Target/*/*.td</span></tt>.</p>
<blockquote>
<div>This is usually a matter of adding a pattern to the .td file that matches the
intrinsic, though it may obviously require adding the instructions you want to
generate as well. There are lots of examples in the PowerPC and X86 backend
to follow.</div></blockquote>
</div>
<div class="section" id="adding-a-new-selectiondag-node">
<h2>Adding a new SelectionDAG node<a class="headerlink" href="#adding-a-new-selectiondag-node" title="Permalink to this headline">¶</a></h2>
<p>As with intrinsics, adding a new SelectionDAG node to LLVM is much easier than
adding a new instruction. New nodes are often added to help represent
instructions common to many targets. These nodes often map to an LLVM
instruction (add, sub) or intrinsic (byteswap, population count). In other
cases, new nodes have been added to allow many targets to perform a common task
(converting between floating point and integer representation) or capture more
complicated behavior in a single node (rotate).</p>
<ol class="arabic">
<li><p class="first"><tt class="docutils literal"><span class="pre">include/llvm/CodeGen/ISDOpcodes.h</span></tt>:</p>
<p>Add an enum value for the new SelectionDAG node.</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">lib/CodeGen/SelectionDAG/SelectionDAG.cpp</span></tt>:</p>
<dl class="docutils">
<dt>Add code to print the node to <tt class="docutils literal"><span class="pre">getOperationName</span></tt>. If your new node can be</dt>
<dd><p class="first last">evaluated at compile time when given constant arguments (such as an add of a
constant with another constant), find the <tt class="docutils literal"><span class="pre">getNode</span></tt> method that takes the
appropriate number of arguments, and add a case for your node to the switch
statement that performs constant folding for nodes that take the same number
of arguments as your new node.</p>
</dd>
</dl>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">lib/CodeGen/SelectionDAG/LegalizeDAG.cpp</span></tt>:</p>
<p>Add code to <a class="reference external" href="CodeGenerator.html#selectiondag_legalize">legalize, promote, and expand</a> the node as necessary. At a
minimum, you will need to add a case statement for your node in
<tt class="docutils literal"><span class="pre">LegalizeOp</span></tt> which calls LegalizeOp on the node’s operands, and returns a
new node if any of the operands changed as a result of being legalized. It
is likely that not all targets supported by the SelectionDAG framework will
natively support the new node. In this case, you must also add code in your
node’s case statement in <tt class="docutils literal"><span class="pre">LegalizeOp</span></tt> to Expand your node into simpler,
legal operations. The case for <tt class="docutils literal"><span class="pre">ISD::UREM</span></tt> for expanding a remainder into
a divide, multiply, and a subtract is a good example.</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">lib/CodeGen/SelectionDAG/LegalizeDAG.cpp</span></tt>:</p>
<dl class="docutils">
<dt>If targets may support the new node being added only at certain sizes, you</dt>
<dd><p class="first last">will also need to add code to your node’s case statement in <tt class="docutils literal"><span class="pre">LegalizeOp</span></tt>
to Promote your node’s operands to a larger size, and perform the correct
operation. You will also need to add code to <tt class="docutils literal"><span class="pre">PromoteOp</span></tt> to do this as
well. For a good example, see <tt class="docutils literal"><span class="pre">ISD::BSWAP</span></tt>, which promotes its operand to
a wider size, performs the byteswap, and then shifts the correct bytes right
to emulate the narrower byteswap in the wider type.</p>
</dd>
</dl>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">lib/CodeGen/SelectionDAG/LegalizeDAG.cpp</span></tt>:</p>
<p>Add a case for your node in <tt class="docutils literal"><span class="pre">ExpandOp</span></tt> to teach the legalizer how to
perform the action represented by the new node on a value that has been split
into high and low halves. This case will be used to support your node with a
64 bit operand on a 32 bit target.</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">lib/CodeGen/SelectionDAG/DAGCombiner.cpp</span></tt>:</p>
<p>If your node can be combined with itself, or other existing nodes in a
peephole-like fashion, add a visit function for it, and call that function
from. There are several good examples for simple combines you can do;
<tt class="docutils literal"><span class="pre">visitFABS</span></tt> and <tt class="docutils literal"><span class="pre">visitSRL</span></tt> are good starting places.</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">lib/Target/PowerPC/PPCISelLowering.cpp</span></tt>:</p>
<p>Each target has an implementation of the <tt class="docutils literal"><span class="pre">TargetLowering</span></tt> class, usually in
its own file (although some targets include it in the same file as the
DAGToDAGISel). The default behavior for a target is to assume that your new
node is legal for all types that are legal for that target. If this target
does not natively support your node, then tell the target to either Promote
it (if it is supported at a larger type) or Expand it. This will cause the
code you wrote in <tt class="docutils literal"><span class="pre">LegalizeOp</span></tt> above to decompose your new node into other
legal nodes for this target.</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">lib/Target/TargetSelectionDAG.td</span></tt>:</p>
<p>Most current targets supported by LLVM generate code using the DAGToDAG
method, where SelectionDAG nodes are pattern matched to target-specific
nodes, which represent individual instructions. In order for the targets to
match an instruction to your new node, you must add a def for that node to
the list in this file, with the appropriate type constraints. Look at
<tt class="docutils literal"><span class="pre">add</span></tt>, <tt class="docutils literal"><span class="pre">bswap</span></tt>, and <tt class="docutils literal"><span class="pre">fadd</span></tt> for examples.</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">lib/Target/PowerPC/PPCInstrInfo.td</span></tt>:</p>
<p>Each target has a tablegen file that describes the target’s instruction set.
For targets that use the DAGToDAG instruction selection framework, add a
pattern for your new node that uses one or more target nodes. Documentation
for this is a bit sparse right now, but there are several decent examples.
See the patterns for <tt class="docutils literal"><span class="pre">rotl</span></tt> in <tt class="docutils literal"><span class="pre">PPCInstrInfo.td</span></tt>.</p>
</li>
<li><p class="first">TODO: document complex patterns.</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/test/Regression/CodeGen/*</span></tt>:</p>
<p>Add test cases for your new node to the test suite.
<tt class="docutils literal"><span class="pre">llvm/test/Regression/CodeGen/X86/bswap.ll</span></tt> is a good example.</p>
</li>
</ol>
</div>
<div class="section" id="adding-a-new-instruction">
<h2>Adding a new instruction<a class="headerlink" href="#adding-a-new-instruction" title="Permalink to this headline">¶</a></h2>
<div class="admonition warning">
<p class="first admonition-title">Warning</p>
<p class="last">Adding instructions changes the bitcode format, and it will take some effort
to maintain compatibility with the previous version. Only add an instruction
if it is absolutely necessary.</p>
</div>
<ol class="arabic">
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/Instruction.def</span></tt>:</p>
<p>add a number for your instruction and an enum name</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/Instructions.h</span></tt>:</p>
<p>add a definition for the class that will represent your instruction</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/Support/InstVisitor.h</span></tt>:</p>
<p>add a prototype for a visitor to your new instruction type</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/AsmParser/Lexer.l</span></tt>:</p>
<p>add a new token to parse your instruction from assembly text file</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/AsmParser/llvmAsmParser.y</span></tt>:</p>
<p>add the grammar on how your instruction can be read and what it will
construct as a result</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/Bitcode/Reader/Reader.cpp</span></tt>:</p>
<p>add a case for your instruction and how it will be parsed from bitcode</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/VMCore/Instruction.cpp</span></tt>:</p>
<p>add a case for how your instruction will be printed out to assembly</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/VMCore/Instructions.cpp</span></tt>:</p>
<p>implement the class you defined in <tt class="docutils literal"><span class="pre">llvm/include/llvm/Instructions.h</span></tt></p>
</li>
<li><p class="first">Test your instruction</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/Target/*</span></tt>:</p>
<p>add support for your instruction to code generators, or add a lowering pass.</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/test/Regression/*</span></tt>:</p>
<p>add your test cases to the test suite.</p>
</li>
</ol>
<p>Also, you need to implement (or modify) any analyses or passes that you want to
understand this new instruction.</p>
</div>
<div class="section" id="adding-a-new-type">
<h2>Adding a new type<a class="headerlink" href="#adding-a-new-type" title="Permalink to this headline">¶</a></h2>
<div class="admonition warning">
<p class="first admonition-title">Warning</p>
<p class="last">Adding new types changes the bitcode format, and will break compatibility with
currently-existing LLVM installations. Only add new types if it is absolutely
necessary.</p>
</div>
<div class="section" id="adding-a-fundamental-type">
<h3>Adding a fundamental type<a class="headerlink" href="#adding-a-fundamental-type" title="Permalink to this headline">¶</a></h3>
<ol class="arabic">
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/Type.h</span></tt>:</p>
<p>add enum for the new type; add static <tt class="docutils literal"><span class="pre">Type*</span></tt> for this type</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/VMCore/Type.cpp</span></tt>:</p>
<p>add mapping from <tt class="docutils literal"><span class="pre">TypeID</span></tt> => <tt class="docutils literal"><span class="pre">Type*</span></tt>; initialize the static <tt class="docutils literal"><span class="pre">Type*</span></tt></p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/AsmReader/Lexer.l</span></tt>:</p>
<p>add ability to parse in the type from text assembly</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/AsmReader/llvmAsmParser.y</span></tt>:</p>
<p>add a token for that type</p>
</li>
</ol>
</div>
<div class="section" id="adding-a-derived-type">
<h3>Adding a derived type<a class="headerlink" href="#adding-a-derived-type" title="Permalink to this headline">¶</a></h3>
<ol class="arabic">
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/Type.h</span></tt>:</p>
<p>add enum for the new type; add a forward declaration of the type also</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/include/llvm/DerivedTypes.h</span></tt>:</p>
<p>add new class to represent new class in the hierarchy; add forward
declaration to the TypeMap value type</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/VMCore/Type.cpp</span></tt>:</p>
<p>add support for derived type to:</p>
<div class="highlight-c++"><div class="highlight"><pre><span class="n">std</span><span class="o">::</span><span class="n">string</span> <span class="n">getTypeDescription</span><span class="p">(</span><span class="k">const</span> <span class="n">Type</span> <span class="o">&</span><span class="n">Ty</span><span class="p">,</span>
<span class="n">std</span><span class="o">::</span><span class="n">vector</span><span class="o"><</span><span class="k">const</span> <span class="n">Type</span><span class="o">*></span> <span class="o">&</span><span class="n">TypeStack</span><span class="p">)</span>
<span class="kt">bool</span> <span class="n">TypesEqual</span><span class="p">(</span><span class="k">const</span> <span class="n">Type</span> <span class="o">*</span><span class="n">Ty</span><span class="p">,</span> <span class="k">const</span> <span class="n">Type</span> <span class="o">*</span><span class="n">Ty2</span><span class="p">,</span>
<span class="n">std</span><span class="o">::</span><span class="n">map</span><span class="o"><</span><span class="k">const</span> <span class="n">Type</span><span class="o">*</span><span class="p">,</span> <span class="k">const</span> <span class="n">Type</span><span class="o">*></span> <span class="o">&</span><span class="n">EqTypes</span><span class="p">)</span>
</pre></div>
</div>
<p>add necessary member functions for type, and factory methods</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/AsmReader/Lexer.l</span></tt>:</p>
<p>add ability to parse in the type from text assembly</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/Bitcode/Writer/Writer.cpp</span></tt>:</p>
<p>modify <tt class="docutils literal"><span class="pre">void</span> <span class="pre">BitcodeWriter::outputType(const</span> <span class="pre">Type</span> <span class="pre">*T)</span></tt> to serialize your
type</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/Bitcode/Reader/Reader.cpp</span></tt>:</p>
<p>modify <tt class="docutils literal"><span class="pre">const</span> <span class="pre">Type</span> <span class="pre">*BitcodeReader::ParseType()</span></tt> to read your data type</p>
</li>
<li><p class="first"><tt class="docutils literal"><span class="pre">llvm/lib/VMCore/AsmWriter.cpp</span></tt>:</p>
<p>modify</p>
<div class="highlight-c++"><div class="highlight"><pre><span class="kt">void</span> <span class="n">calcTypeName</span><span class="p">(</span><span class="k">const</span> <span class="n">Type</span> <span class="o">*</span><span class="n">Ty</span><span class="p">,</span>
<span class="n">std</span><span class="o">::</span><span class="n">vector</span><span class="o"><</span><span class="k">const</span> <span class="n">Type</span><span class="o">*></span> <span class="o">&</span><span class="n">TypeStack</span><span class="p">,</span>
<span class="n">std</span><span class="o">::</span><span class="n">map</span><span class="o"><</span><span class="k">const</span> <span class="n">Type</span><span class="o">*</span><span class="p">,</span><span class="n">std</span><span class="o">::</span><span class="n">string</span><span class="o">></span> <span class="o">&</span><span class="n">TypeNames</span><span class="p">,</span>
<span class="n">std</span><span class="o">::</span><span class="n">string</span> <span class="o">&</span><span class="n">Result</span><span class="p">)</span>
</pre></div>
</div>
<p>to output the new derived type</p>
</li>
</ol>
</div>
</div>
</div>
</div>
</div>
<div class="clearer"></div>
</div>
<div class="related">
<h3>Navigation</h3>
<ul>
<li class="right" style="margin-right: 10px">
<a href="genindex.html" title="General Index"
>index</a></li>
<li class="right" >
<a href="HowToSetUpLLVMStyleRTTI.html" title="How to set up LLVM-style RTTI for your class hierarchy"
>next</a> |</li>
<li class="right" >
<a href="CompilerWriterInfo.html" title="Architecture & Platform Information for Compiler Writers"
>previous</a> |</li>
<li><a href="http://llvm.org/">LLVM Home</a> | </li>
<li><a href="index.html">Documentation</a>»</li>
</ul>
</div>
<div class="footer">
© Copyright 2003-2014, LLVM Project.
Last updated on 2015-02-27.
Created using <a href="http://sphinx-doc.org/">Sphinx</a> 1.2.3.
</div>
</body>
</html>
|