blob: 419b7c634f282cdd9757fae695ce0c1293645b2f [file] [log] [blame]
Misha Brukmana3ce4292004-04-06 03:53:49 +00001<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<html>
4<head>
5 <title>Extending LLVM: Adding instructions, intrinsics, types, etc.</title>
6 <link rel="stylesheet" href="llvm.css" type="text/css">
7</head>
8
9<body>
10
11<div class="doc_title">
12 Extending LLVM: Adding instructions, intrinsics, types, etc.
13</div>
14
15<ol>
16 <li><a href="#introduction">Introduction and Warning</a></li>
Misha Brukmana3ce4292004-04-06 03:53:49 +000017 <li><a href="#intrinsic">Adding a new intrinsic function</a></li>
Chris Lattner36365402004-04-09 19:24:20 +000018 <li><a href="#instruction">Adding a new instruction</a></li>
Nate Begeman099d76c2006-01-16 07:54:23 +000019 <li><a href="#sdnode">Adding a new SelectionDAG node</a></li>
Misha Brukmana3ce4292004-04-06 03:53:49 +000020 <li><a href="#type">Adding a new type</a>
21 <ol>
22 <li><a href="#fund_type">Adding a new fundamental type</a></li>
23 <li><a href="#derived_type">Adding a new derived type</a></li>
24 </ol></li>
25</ol>
26
Chris Lattner7911ce22004-05-23 21:07:27 +000027<div class="doc_author">
Chris Lattner5eb9f0d2005-05-11 03:53:53 +000028 <p>Written by <a href="http://misha.brukman.net">Misha Brukman</a>,
Chris Lattnerb5f6e252006-01-16 16:31:40 +000029 Brad Jones, Nate Begeman,
30 and <a href="http://nondot.org/sabre">Chris Lattner</a></p>
Misha Brukmana3ce4292004-04-06 03:53:49 +000031</div>
32
33<!-- *********************************************************************** -->
34<div class="doc_section">
35 <a name="introduction">Introduction and Warning</a>
36</div>
37<!-- *********************************************************************** -->
38
39<div class="doc_text">
40
41<p>During the course of using LLVM, you may wish to customize it for your
42research project or for experimentation. At this point, you may realize that
43you need to add something to LLVM, whether it be a new fundamental type, a new
44intrinsic function, or a whole new instruction.</p>
45
46<p>When you come to this realization, stop and think. Do you really need to
47extend LLVM? Is it a new fundamental capability that LLVM does not support at
48its current incarnation or can it be synthesized from already pre-existing LLVM
49elements? If you are not sure, ask on the <a
50href="http://mail.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVM-dev</a> list. The
51reason is that extending LLVM will get involved as you need to update all the
52different passes that you intend to use with your extension, and there are
53<em>many</em> LLVM analyses and transformations, so it may be quite a bit of
54work.</p>
55
Misha Brukmanb3b28272004-04-06 04:17:51 +000056<p>Adding an <a href="#intrinsic">intrinsic function</a> is easier than adding
57an instruction, and is transparent to optimization passes which treat it as an
58unanalyzable function. If your added functionality can be expressed as a
59function call, an intrinsic function is the method of choice for LLVM
60extension.</p>
61
Misha Brukmana3ce4292004-04-06 03:53:49 +000062<p>Before you invest a significant amount of effort into a non-trivial
63extension, <span class="doc_warning">ask on the list</span> if what you are
64looking to do can be done with already-existing infrastructure, or if maybe
65someone else is already working on it. You will save yourself a lot of time and
66effort by doing so.</p>
67
Misha Brukmana3ce4292004-04-06 03:53:49 +000068</div>
69
70<!-- *********************************************************************** -->
71<div class="doc_section">
Chris Lattner36365402004-04-09 19:24:20 +000072 <a name="intrinsic">Adding a new intrinsic function</a>
73</div>
74<!-- *********************************************************************** -->
75
76<div class="doc_text">
77
78<p>Adding a new intrinsic function to LLVM is much easier than adding a new
79instruction. Almost all extensions to LLVM should start as an intrinsic
80function and then be turned into an instruction if warranted.</p>
81
82<ol>
83<li><tt>llvm/docs/LangRef.html</tt>:
84 Document the intrinsic. Decide whether it is code generator specific and
85 what the restrictions are. Talk to other people about it so that you are
86 sure it's a good idea.</li>
87
88<li><tt>llvm/include/llvm/Intrinsics.h</tt>:
89 add an enum in the <tt>llvm::Intrinsic</tt> namespace</li>
90
Chris Lattner36365402004-04-09 19:24:20 +000091<li><tt>llvm/lib/VMCore/Verifier.cpp</tt>:
92 Add code to check the invariants of the intrinsic are respected.</li>
93
94<li><tt>llvm/lib/VMCore/Function.cpp (<tt>Function::getIntrinsicID()</tt>)</tt>:
95 Identify the new intrinsic function, returning the enum for the intrinsic
96 that you added.</li>
Chris Lattner0190fdb2004-04-10 06:56:53 +000097
98<li><tt>llvm/lib/Analysis/BasicAliasAnalysis.cpp</tt>: If the new intrinsic does
Chris Lattner81519d92004-06-20 07:53:22 +000099 not access memory or does not write to memory, add it to the relevant list
Chris Lattner0190fdb2004-04-10 06:56:53 +0000100 of functions.</li>
101
Nate Begeman2f86c222006-01-14 01:27:10 +0000102<li><tt>llvm/lib/Analysis/ConstantFolding.cpp</tt>: If it is possible to
103 constant fold your intrinsic, add support to it in the
104 <tt>canConstantFoldCallTo</tt> and <tt>ConstantFoldCall</tt> functions.</li>
105
106<li><tt>llvm/lib/Transforms/Utils/Local.cpp</tt>: If your intrinsic has no side-
107 effects, add it to the list of intrinsics in the
108 <tt>isInstructionTriviallyDead</tt> function.</li>
Chris Lattnerd828bc62004-04-13 19:48:55 +0000109
Nate Begeman099d76c2006-01-16 07:54:23 +0000110<li><tt>llvm/test/Regression/*</tt>: Add test cases for your test cases to the
111 test suite</li>
Chris Lattner36365402004-04-09 19:24:20 +0000112</ol>
113
Chris Lattner5eb9f0d2005-05-11 03:53:53 +0000114<p>Once the intrinsic has been added to the system, you must add code generator
115support for it. Generally you must do the following steps:</p>
116
117<dl>
118<dt>Add support to the C backend in <tt>lib/Target/CBackend/</tt></dt>
119
Nate Begeman099d76c2006-01-16 07:54:23 +0000120<dd>Depending on the intrinsic, there are a few ways to implement this. For
121most intrinsics, it makes sense to add code to lower your intrinsic in
122<tt>LowerIntrinsicCall</tt> in <tt>lib/CodeGen/IntrinsicLowering.cpp</tt>.
123Second, if it makes sense to lower the intrinsic to an expanded sequence of C
124code in all cases, just emit the expansion in <tt>visitCallInst</tt> in
125<tt>Writer.cpp</tt>. If the intrinsic has some way to express it with GCC
126(or any other compiler) extensions, it can be conditionally supported based on
Chris Lattnerb5f6e252006-01-16 16:31:40 +0000127the compiler compiling the CBE output (see <tt>llvm.prefetch</tt> for an
128example).
Nate Begeman099d76c2006-01-16 07:54:23 +0000129Third, if the intrinsic really has no way to be lowered, just have the code
130generator emit code that prints an error message and calls abort if executed.
Chris Lattner5eb9f0d2005-05-11 03:53:53 +0000131</dd>
132
Nate Begeman099d76c2006-01-16 07:54:23 +0000133<dl>
134<dt>Add support to the SelectionDAG Instruction Selector in
135<tt>lib/CodeGen/SelectionDAG/</tt></dt>
Chris Lattner5eb9f0d2005-05-11 03:53:53 +0000136
Nate Begeman099d76c2006-01-16 07:54:23 +0000137<dd>Since most targets in LLVM use the SelectionDAG framework for generating
138code, you will likely need to add support for your intrinsic there as well.
139This is usually accomplished by adding a new node, and then teaching the
140SelectionDAG code how to handle that node. To do this, follow the steps in
Chris Lattnerb5f6e252006-01-16 16:31:40 +0000141the <a href="#sdnode">Adding a new SelectionDAG node</a> section.</dd>
Chris Lattner5eb9f0d2005-05-11 03:53:53 +0000142
Nate Begeman099d76c2006-01-16 07:54:23 +0000143<dl>
144<dt>Once you have added the new node, add code to
145<tt>SelectionDAG/SelectionDAGISel.cpp</tt> to recognize the intrinsic. In most
146cases, the intrinsic will just be turned into the node you just added. For an
Chris Lattnerb5f6e252006-01-16 16:31:40 +0000147example of this, see how <tt>visitIntrinsicCall</tt> handles
148<tt>Intrinsic::ctpop_*</tt>.
Nate Begeman099d76c2006-01-16 07:54:23 +0000149</dt>
Chris Lattner5eb9f0d2005-05-11 03:53:53 +0000150
Nate Begeman099d76c2006-01-16 07:54:23 +0000151</div>
Chris Lattner5eb9f0d2005-05-11 03:53:53 +0000152
Nate Begeman099d76c2006-01-16 07:54:23 +0000153<!-- *********************************************************************** -->
154<div class="doc_section">
155 <a name="sdnode">Adding a new SelectionDAG node</a>
156</div>
157<!-- *********************************************************************** -->
Chris Lattner5eb9f0d2005-05-11 03:53:53 +0000158
Nate Begeman099d76c2006-01-16 07:54:23 +0000159<div class="doc_text">
Chris Lattner5eb9f0d2005-05-11 03:53:53 +0000160
Nate Begeman099d76c2006-01-16 07:54:23 +0000161<p>As with intrinsics, adding a new SelectionDAG node to LLVM is much easier
162than adding a new instruction. New nodes are often added to help represent
163instructions common to many targets. These nodes often map to an LLVM
164instruction (add, sub) or intrinsic (byteswap, population count). In other
165cases, new nodes have been added to allow many targets to perform a common task
166(converting between floating point and integer representation) or capture more
167complicated behavior in a single node (rotate).</p>
Chris Lattner5eb9f0d2005-05-11 03:53:53 +0000168
Nate Begeman099d76c2006-01-16 07:54:23 +0000169<ol>
170<li><tt>include/llvm/CodeGen/SelectionDAGNodes.h</tt>:
171 Add an enum value for the new SelectionDAG node.</li>
172<li><tt>lib/CodeGen/SelectionDAG/SelectionDAG.cpp</tt>:
173 Add code to print the node to <tt>getOperationName</tt>. If your new node
174 can be evaluated at compile time when given constant arguments (such as an
175 add of a constant with another constant), find the <tt>getNode</tt> method
176 that takes the appropriate number of arguments, and add a case for your node
177 to the switch statement that performs constant folding for nodes that take
178 the same number of arguments as your new node.</li>
179<li><tt>lib/CodeGen/SelectionDAG/LegalizeDAG.cpp</tt>:
180 Add code to <a href="CodeGenerator.html#selectiondag_legalize">legalize,
181 promote, and expand</a> the node as necessary. At a minimum, you will need
182 to add a case statement for your node in <tt>LegalizeOp</tt> which calls
183 LegalizeOp on the node's operands, and returns a new node if any of the
184 operands changed as a result of being legalized. It is likely that not all
185 targets supported by the SelectionDAG framework will natively support the
186 new node. In this case, you must also add code in your node's case
187 statement in <tt>LegalizeOp</tt> to Expand your node into simpler, legal
Chris Lattnerb5f6e252006-01-16 16:31:40 +0000188 operations. The case for <tt>ISD::UREM</tt> for expanding a remainder into
189 a divide, multiply, and a subtract is a good example.</li>
Nate Begeman099d76c2006-01-16 07:54:23 +0000190<li><tt>lib/CodeGen/SelectionDAG/LegalizeDAG.cpp</tt>:
191 If targets may support the new node being added only at certain sizes, you
192 will also need to add code to your node's case statement in
193 <tt>LegalizeOp</tt> to Promote your node's operands to a larger size, and
194 perform the correct operation. You will also need to add code to
Chris Lattnerb5f6e252006-01-16 16:31:40 +0000195 <tt>PromoteOp</tt> to do this as well. For a good example, see
196 <tt>ISD::BSWAP</tt>,
Nate Begeman099d76c2006-01-16 07:54:23 +0000197 which promotes its operand to a wider size, performs the byteswap, and then
198 shifts the correct bytes right to emulate the narrower byteswap in the
199 wider type.</li>
200<li><tt>lib/CodeGen/SelectionDAG/LegalizeDAG.cpp</tt>:
201 Add a case for your node in <tt>ExpandOp</tt> to teach the legalizer how to
202 perform the action represented by the new node on a value that has been
203 split into high and low halves. This case will be used to support your
204 node with a 64 bit operand on a 32 bit target.</li>
205<li><tt>lib/CodeGen/SelectionDAG/DAGCombiner.cpp</tt>:
206 If your node can be combined with itself, or other existing nodes in a
207 peephole-like fashion, add a visit function for it, and call that function
208 from <tt></tt>. There are several good examples for simple combines you
209 can do; <tt>visitFABS</tt> and <tt>visitSRL</tt> are good starting places.
210 </li>
211<li><tt>lib/Target/PowerPC/PPCISelLowering.cpp</tt>:
212 Each target has an implementation of the <tt>TargetLowering</tt> class,
213 usually in its own file (although some targets include it in the same
214 file as the DAGToDAGISel). The default behavior for a target is to
215 assume that your new node is legal for all types that are legal for
216 that target. If this target does not natively support your node, then
217 tell the target to either Promote it (if it is supported at a larger
218 type) or Expand it. This will cause the code you wrote in
219 <tt>LegalizeOp</tt> above to decompose your new node into other legal
220 nodes for this target.</li>
221<li><tt>lib/Target/TargetSelectionDAG.td</tt>:
222 Most current targets supported by LLVM generate code using the DAGToDAG
223 method, where SelectionDAG nodes are pattern matched to target-specific
224 nodes, which represent individual instructions. In order for the targets
225 to match an instruction to your new node, you must add a def for that node
226 to the list in this file, with the appropriate type constraints. Look at
227 <tt>add</tt>, <tt>bswap</tt>, and <tt>fadd</tt> for examples.</li>
228<li><tt>lib/Target/PowerPC/PPCInstrInfo.td</tt>:
229 Each target has a tablegen file that describes the target's instruction
230 set. For targets that use the DAGToDAG instruction selection framework,
231 add a pattern for your new node that uses one or more target nodes.
232 Documentation for this is a bit sparse right now, but there are several
233 decent examples. See the patterns for <tt>rotl</tt> in
234 <tt>PPCInstrInfo.td</tt>.</li>
235<li>TODO: document complex patterns.</li>
236<li><tt>llvm/test/Regression/CodeGen/*</tt>: Add test cases for your new node
237 to the test suite. <tt>llvm/test/Regression/CodeGen/X86/bswap.ll</tt> is
238 a good example.</li>
239</ol>
Chris Lattner36365402004-04-09 19:24:20 +0000240
241</div>
242
243<!-- *********************************************************************** -->
244<div class="doc_section">
Misha Brukmana3ce4292004-04-06 03:53:49 +0000245 <a name="instruction">Adding a new instruction</a>
246</div>
247<!-- *********************************************************************** -->
248
249<div class="doc_text">
250
251<p><span class="doc_warning">WARNING: adding instructions changes the bytecode
Misha Brukmanb3b28272004-04-06 04:17:51 +0000252format, and it will take some effort to maintain compatibility with
253the previous version.</span> Only add an instruction if it is absolutely
Misha Brukmana3ce4292004-04-06 03:53:49 +0000254necessary.</p>
255
256<ol>
Misha Brukmanb3b28272004-04-06 04:17:51 +0000257
Misha Brukmana3ce4292004-04-06 03:53:49 +0000258<li><tt>llvm/include/llvm/Instruction.def</tt>:
259 add a number for your instruction and an enum name</li>
260
Misha Brukman47b14a42004-07-29 17:30:56 +0000261<li><tt>llvm/include/llvm/Instructions.h</tt>:
Misha Brukmana3ce4292004-04-06 03:53:49 +0000262 add a definition for the class that will represent your instruction</li>
263
264<li><tt>llvm/include/llvm/Support/InstVisitor.h</tt>:
265 add a prototype for a visitor to your new instruction type</li>
266
267<li><tt>llvm/lib/AsmParser/Lexer.l</tt>:
268 add a new token to parse your instruction from assembly text file</li>
269
270<li><tt>llvm/lib/AsmParser/llvmAsmParser.y</tt>:
271 add the grammar on how your instruction can be read and what it will
272 construct as a result</li>
273
Misha Brukmane39cd632004-09-28 16:58:12 +0000274<li><tt>llvm/lib/Bytecode/Reader/Reader.cpp</tt>:
Misha Brukmana3ce4292004-04-06 03:53:49 +0000275 add a case for your instruction and how it will be parsed from bytecode</li>
276
277<li><tt>llvm/lib/VMCore/Instruction.cpp</tt>:
278 add a case for how your instruction will be printed out to assembly</li>
279
Chris Lattner8f363212004-07-29 17:31:57 +0000280<li><tt>llvm/lib/VMCore/Instructions.cpp</tt>:
Misha Brukmane39cd632004-09-28 16:58:12 +0000281 implement the class you defined in
282 <tt>llvm/include/llvm/Instructions.h</tt></li>
Misha Brukmana3ce4292004-04-06 03:53:49 +0000283
Misha Brukmana4242282004-12-01 20:58:54 +0000284<li>Test your instruction</li>
285
286<li><tt>llvm/lib/Target/*</tt>:
287 Add support for your instruction to code generators, or add a lowering
288 pass.</li>
289
290<li><tt>llvm/test/Regression/*</tt>: add your test cases to the test suite.</li>
291
Misha Brukmana3ce4292004-04-06 03:53:49 +0000292</ol>
293
294<p>Also, you need to implement (or modify) any analyses or passes that you want
295to understand this new instruction.</p>
296
297</div>
298
Misha Brukmana3ce4292004-04-06 03:53:49 +0000299
300<!-- *********************************************************************** -->
301<div class="doc_section">
302 <a name="type">Adding a new type</a>
303</div>
304<!-- *********************************************************************** -->
305
306<div class="doc_text">
307
308<p><span class="doc_warning">WARNING: adding new types changes the bytecode
309format, and will break compatibility with currently-existing LLVM
310installations.</span> Only add new types if it is absolutely necessary.</p>
311
312</div>
313
314<!-- ======================================================================= -->
315<div class="doc_subsection">
316 <a name="fund_type">Adding a fundamental type</a>
317</div>
318
319<div class="doc_text">
320
321<ol>
322
Misha Brukmana3ce4292004-04-06 03:53:49 +0000323<li><tt>llvm/include/llvm/Type.h</tt>:
Chris Lattner55f95012005-04-23 21:59:11 +0000324 add enum for the new type; add static <tt>Type*</tt> for this type</li>
Misha Brukmana3ce4292004-04-06 03:53:49 +0000325
326<li><tt>llvm/lib/VMCore/Type.cpp</tt>:
327 add mapping from <tt>TypeID</tt> =&gt; <tt>Type*</tt>;
328 initialize the static <tt>Type*</tt></li>
329
330<li><tt>llvm/lib/AsmReader/Lexer.l</tt>:
331 add ability to parse in the type from text assembly</li>
332
333<li><tt>llvm/lib/AsmReader/llvmAsmParser.y</tt>:
334 add a token for that type</li>
335
336</ol>
337
338</div>
339
340<!-- ======================================================================= -->
341<div class="doc_subsection">
342 <a name="derived_type">Adding a derived type</a>
343</div>
344
345<div class="doc_text">
346
Chris Lattner8dad40c2004-08-12 19:06:24 +0000347<ol>
Chris Lattner8dad40c2004-08-12 19:06:24 +0000348<li><tt>llvm/include/llvm/Type.h</tt>:
Chris Lattner55f95012005-04-23 21:59:11 +0000349 add enum for the new type; add a forward declaration of the type
Misha Brukman7cc8a892004-08-12 19:58:43 +0000350 also</li>
Chris Lattner8dad40c2004-08-12 19:06:24 +0000351
Chris Lattner47746aa2005-11-13 02:09:55 +0000352<li><tt>llvm/include/llvm/DerivedTypes.h</tt>:
Chris Lattner8dad40c2004-08-12 19:06:24 +0000353 add new class to represent new class in the hierarchy; add forward
354 declaration to the TypeMap value type</li>
355
356<li><tt>llvm/lib/VMCore/Type.cpp</tt>:
357 add support for derived type to:
Misha Brukman7cc8a892004-08-12 19:58:43 +0000358<div class="doc_code">
359<pre>
360std::string getTypeDescription(const Type &amp;Ty,
361 std::vector&lt;const Type*&gt; &amp;TypeStack)
362bool TypesEqual(const Type *Ty, const Type *Ty2,
363 std::map&lt;const Type*, const Type*&gt; &amp; EqTypes)
364</pre>
365</div>
366 add necessary member functions for type, and factory methods</li>
Chris Lattner8dad40c2004-08-12 19:06:24 +0000367
368<li><tt>llvm/lib/AsmReader/Lexer.l</tt>:
369 add ability to parse in the type from text assembly</li>
370
371<li><tt>llvm/lib/ByteCode/Writer/Writer.cpp</tt>:
Misha Brukman7cc8a892004-08-12 19:58:43 +0000372 modify <tt>void BytecodeWriter::outputType(const Type *T)</tt> to serialize
373 your type</li>
Chris Lattner8dad40c2004-08-12 19:06:24 +0000374
375<li><tt>llvm/lib/ByteCode/Reader/Reader.cpp</tt>:
Misha Brukman7cc8a892004-08-12 19:58:43 +0000376 modify <tt>const Type *BytecodeReader::ParseType()</tt> to read your data
377 type</li>
Chris Lattner8dad40c2004-08-12 19:06:24 +0000378
379<li><tt>llvm/lib/VMCore/AsmWriter.cpp</tt>:
Misha Brukman7cc8a892004-08-12 19:58:43 +0000380 modify
381<div class="doc_code">
382<pre>
383void calcTypeName(const Type *Ty,
384 std::vector&lt;const Type*&gt; &amp;TypeStack,
385 std::map&lt;const Type*,std::string&gt; &amp;TypeNames,
386 std::string &amp; Result)
387</pre>
388</div>
Chris Lattner8dad40c2004-08-12 19:06:24 +0000389 to output the new derived type
390</li>
391
392
393</ol>
Misha Brukmana3ce4292004-04-06 03:53:49 +0000394
395</div>
396
397<!-- *********************************************************************** -->
398
399<hr>
400<address>
401 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
402 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
403 <a href="http://validator.w3.org/check/referer"><img
404 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
405
Misha Brukmana3ce4292004-04-06 03:53:49 +0000406 <a href="http://llvm.cs.uiuc.edu">The LLVM Compiler Infrastructure</a>
407 <br>
408 Last modified: $Date$
409</address>
410
411</body>
412</html>