Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / cac2135bc8a732c495b52aee85800baca2e23c22 / . / docs / tutorial / LangImpl3.html
blob: 3e629287cc337cb51a71d6fa918ccd3f27771047 [file] [log] [blame]
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3
4<html>
5<head>
6 <title>Kaleidoscope: Implementing code generation to LLVM IR</title>
7 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
8 <meta name="author" content="Chris Lattner">
9 <link rel="stylesheet" href="../llvm.css" type="text/css">
10</head>
11
12<body>
13
14<div class="doc_title">Kaleidoscope: Code generation to LLVM IR</div>
15
Chris Lattner128eb862007-11-05 19:06:59 +0000[diff] [blame]16<ul>
17<li>Chapter 3
18 <ol>
19 <li><a href="#intro">Chapter 3 Introduction</a></li>
20 <li><a href="#basics">Code Generation setup</a></li>
21 <li><a href="#exprs">Expression Code Generation</a></li>
22 <li><a href="#funcs">Function Code Generation</a></li>
23 <li><a href="#driver">Driver Changes and Closing Thoughts</a></li>
24 <li><a href="#code">Full Code Listing</a></li>
25 </ol>
26</li>
27</ul>
28
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]29<div class="doc_author">
30 <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a></p>
31</div>
32
33<!-- *********************************************************************** -->
Chris Lattner128eb862007-11-05 19:06:59 +0000[diff] [blame]34<div class="doc_section"><a name="intro">Chapter 3 Introduction</a></div>
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]35<!-- *********************************************************************** -->
36
37<div class="doc_text">
38
Chris Lattner128eb862007-11-05 19:06:59 +0000[diff] [blame]39<p>Welcome to Chapter 3 of the "<a href="index.html">Implementing a language
40with LLVM</a>" tutorial. This chapter shows you how to transform the <a
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]41href="LangImpl2.html">Abstract Syntax Tree built in Chapter 2</a> into LLVM IR.
42This will teach you a little bit about how LLVM does things, as well as
43demonstrate how easy it is to use. It's much more work to build a lexer and
44parser than it is to generate LLVM IR code.
45</p>
46
47</div>
48
49<!-- *********************************************************************** -->
50<div class="doc_section"><a name="basics">Code Generation setup</a></div>
51<!-- *********************************************************************** -->
52
53<div class="doc_text">
54
55<p>
56In order to generate LLVM IR, we want some simple setup to get started. First,
57we define virtual codegen methods in each AST class:</p>
58
59<div class="doc_code">
60<pre>
61/// ExprAST - Base class for all expression nodes.
62class ExprAST {
63public:
64 virtual ~ExprAST() {}
Chris Lattnercac21352007-11-05 19:25:14 +0000[diff] [blame^]65 <b>virtual Value *Codegen() = 0;</b>
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]66};
67
68/// NumberExprAST - Expression class for numeric literals like "1.0".
69class NumberExprAST : public ExprAST {
70 double Val;
71public:
Chris Lattner28571ed2007-10-23 04:27:44 +0000[diff] [blame]72 explicit NumberExprAST(double val) : Val(val) {}
Chris Lattnercac21352007-11-05 19:25:14 +0000[diff] [blame^]73 <b>virtual Value *Codegen();</b>
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]74};
75...
76</pre>
77</div>
78
Chris Lattner28571ed2007-10-23 04:27:44 +0000[diff] [blame]79<p>The Codegen() method says to emit IR for that AST node and all things it
80depends on, and they all return an LLVM Value object.
81"Value" is the class used to represent a "<a
82href="http://en.wikipedia.org/wiki/Static_single_assignment_form">Static Single
83Assignment (SSA)</a> register" or "SSA value" in LLVM. The most distinct aspect
84of SSA values is that their value is computed as the related instruction
85executes, and it does not get a new value until (and if) the instruction
86re-executes. In order words, there is no way to "change" an SSA value. For
87more information, please read up on <a
88href="http://en.wikipedia.org/wiki/Static_single_assignment_form">Static Single
89Assignment</a> - the concepts are really quite natural once you grok them.</p>
90
Chris Lattnercac21352007-11-05 19:25:14 +0000[diff] [blame^]91<p>Note that instead of adding virtual methods to the ExprAST class hierarchy,
92it could also make sense to use a visitor pattern or some other way to model
93this. Again, this tutorial won't dwell on good software engineering practices:
94for our purposes, adding virtual methods is simplest.</p>
95
Chris Lattner28571ed2007-10-23 04:27:44 +0000[diff] [blame]96<p>The
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]97second thing we want is an "Error" method like we used for parser, which will
98be used to report errors found during code generation (for example, use of an
99undeclared parameter):</p>
100
101<div class="doc_code">
102<pre>
103Value *ErrorV(const char *Str) { Error(Str); return 0; }
104
105static Module *TheModule;
106static LLVMBuilder Builder;
107static std::map&lt;std::string, Value*&gt; NamedValues;
108</pre>
109</div>
110
111<p>The static variables will be used during code generation. <tt>TheModule</tt>
112is the LLVM construct that contains all of the functions and global variables in
113a chunk of code. In many ways, it is the top-level structure that the LLVM IR
114uses to contain code.</p>
115
116<p>The <tt>Builder</tt> object is a helper object that makes it easy to generate
Chris Lattnerf6e53df2007-11-05 18:02:15 +0000[diff] [blame]117LLVM instructions. Instances of the <a
118href="http://llvm.org/doxygen/LLVMBuilder_8h-source.html"><tt>LLVMBuilder</tt>
119class</a> keeps track of the current place to
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]120insert instructions and has methods to create new instructions.</p>
121
122<p>The <tt>NamedValues</tt> map keeps track of which values are defined in the
123current scope and what their LLVM representation is. In this form of
124Kaleidoscope, the only things that can be referenced are function parameters.
125As such, function parameters will be in this map when generating code for their
126function body.</p>
127
128<p>
129With these basics in place, we can start talking about how to generate code for
130each expression. Note that this assumes that the <tt>Builder</tt> has been set
131up to generate code <em>into</em> something. For now, we'll assume that this
132has already been done, and we'll just use it to emit code.
133</p>
134
135</div>
136
137<!-- *********************************************************************** -->
138<div class="doc_section"><a name="exprs">Expression Code Generation</a></div>
139<!-- *********************************************************************** -->
140
141<div class="doc_text">
142
143<p>Generating LLVM code for expression nodes is very straight-forward: less
144than 45 lines of commented code for all four of our expression nodes. First,
145we'll do numeric literals:</p>
146
147<div class="doc_code">
148<pre>
149Value *NumberExprAST::Codegen() {
150 return ConstantFP::get(Type::DoubleTy, APFloat(Val));
151}
152</pre>
153</div>
154
Chris Lattnerd3f0cdd2007-10-23 04:51:30 +0000[diff] [blame]155<p>In the LLVM IR, numeric constants are represented with the
156<tt>ConstantFP</tt> class, which holds the numeric value in an <tt>APFloat</tt>
157internally (<tt>APFloat</tt> has the capability of holding floating point
158constants of <em>A</em>rbitrary <em>P</em>recision). This code basically just
159creates and returns a <tt>ConstantFP</tt>. Note that in the LLVM IR
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]160that constants are all uniqued together and shared. For this reason, the API
Chris Lattnerd3f0cdd2007-10-23 04:51:30 +0000[diff] [blame]161uses "the foo::get(..)" idiom instead of "new foo(..)" or "foo::create(..).</p>
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]162
163<div class="doc_code">
164<pre>
165Value *VariableExprAST::Codegen() {
166 // Look this variable up in the function.
167 Value *V = NamedValues[Name];
168 return V ? V : ErrorV("Unknown variable name");
169}
170</pre>
171</div>
172
Chris Lattnerd3f0cdd2007-10-23 04:51:30 +0000[diff] [blame]173<p>References to variables is also quite simple here. In the simple version
174of Kaleidoscope, we assume that the variable has already been emited somewhere
175and its value is available. In practice, the only values that can be in the
176<tt>NamedValues</tt> map are function arguments. This
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]177code simply checks to see that the specified name is in the map (if not, an
178unknown variable is being referenced) and returns the value for it.</p>
179
180<div class="doc_code">
181<pre>
182Value *BinaryExprAST::Codegen() {
183 Value *L = LHS-&gt;Codegen();
184 Value *R = RHS-&gt;Codegen();
185 if (L == 0 || R == 0) return 0;
186
187 switch (Op) {
188 case '+': return Builder.CreateAdd(L, R, "addtmp");
189 case '-': return Builder.CreateSub(L, R, "subtmp");
190 case '*': return Builder.CreateMul(L, R, "multmp");
191 case '&lt;':
192 L = Builder.CreateFCmpULT(L, R, "multmp");
193 // Convert bool 0/1 to double 0.0 or 1.0
194 return Builder.CreateUIToFP(L, Type::DoubleTy, "booltmp");
195 default: return ErrorV("invalid binary operator");
196 }
197}
198</pre>
199</div>
200
Chris Lattnerd3f0cdd2007-10-23 04:51:30 +0000[diff] [blame]201<p>Binary operators start to get more interesting. The basic idea here is that
202we recursively emit code for the left-hand side of the expression, then the
203right-hand side, then we compute the result of the binary expression. In this
204code, we do a simple switch on the opcode to create the right LLVM instruction.
205</p>
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]206
Chris Lattnerd3f0cdd2007-10-23 04:51:30 +0000[diff] [blame]207<p>In this example, the LLVM builder class is starting to show its value.
208Because it knows where to insert the newly created instruction, you just have to
209specificy what instruction to create (e.g. with <tt>CreateAdd</tt>), which
210operands to use (<tt>L</tt> and <tt>R</tt> here) and optionally provide a name
211for the generated instruction. One nice thing about LLVM is that the name is
212just a hint: if there are multiple additions in a single function, the first
213will be named "addtmp" and the second will be "autorenamed" by adding a suffix,
214giving it a name like "addtmp42". Local value names for instructions are purely
215optional, but it makes it much easier to read the IR dumps.</p>
216
217<p><a href="../LangRef.html#instref">LLVM instructions</a> are constrained to
218have very strict type properties: for example, the Left and Right operators of
219an <a href="../LangRef.html#i_add">add instruction</a> have to have the same
220type, and that the result of the add matches the operands. Because all values
221in Kaleidoscope are doubles, this makes for very simple code for add, sub and
222mul.</p>
223
224<p>On the other hand, LLVM specifies that the <a
225href="../LangRef.html#i_fcmp">fcmp instruction</a> always returns an 'i1' value
226(a one bit integer). However, Kaleidoscope wants the value to be a 0.0 or 1.0
227value. In order to get these semantics, we combine the fcmp instruction with
228a <a href="../LangRef.html#i_uitofp">uitofp instruction</a>. This instruction
229converts its input integer into a floating point value by treating the input
230as an unsigned value. In contrast, if we used the <a
231href="../LangRef.html#i_sitofp">sitofp instruction</a>, the Kaleidoscope '<'
232operator would return 0.0 and -1.0, depending on the input value.</p>
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]233
234<div class="doc_code">
235<pre>
236Value *CallExprAST::Codegen() {
237 // Look up the name in the global module table.
238 Function *CalleeF = TheModule-&gt;getFunction(Callee);
239 if (CalleeF == 0)
240 return ErrorV("Unknown function referenced");
241
242 // If argument mismatch error.
243 if (CalleeF-&gt;arg_size() != Args.size())
244 return ErrorV("Incorrect # arguments passed");
245
246 std::vector&lt;Value*&gt; ArgsV;
247 for (unsigned i = 0, e = Args.size(); i != e; ++i) {
248 ArgsV.push_back(Args[i]-&gt;Codegen());
249 if (ArgsV.back() == 0) return 0;
250 }
251
252 return Builder.CreateCall(CalleeF, ArgsV.begin(), ArgsV.end(), "calltmp");
253}
254</pre>
255</div>
256
Chris Lattnerd3f0cdd2007-10-23 04:51:30 +0000[diff] [blame]257<p>Code generation for function calls is quite straight-forward with LLVM. The
258code above first looks the name of the function up in the LLVM Module's symbol
259table. Recall that the LLVM Module is the container that holds all of the
260functions we are JIT'ing. By giving each function the same name as what the
261user specifies, we can use the LLVM symbol table to resolve function names for
262us.</p>
263
264<p>Once we have the function to call, we recursively codegen each argument that
265is to be passed in, and create an LLVM <a href="../LangRef.html#i_call">call
266instruction</a>. Note that LLVM uses the native C calling conventions by
267default, allowing these calls to call into standard library functions like
268"sin" and "cos" with no additional effort.</p>
269
270<p>This wraps up our handling of the four basic expressions that we have so far
271in Kaleidoscope. Feel free to go in and add some more. For example, by
272browsing the <a href="../LangRef.html">LLVM language reference</a> you'll find
273several other interesting instructions that are really easy to plug into our
274basic framework.</p>
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]275
276</div>
277
278<!-- *********************************************************************** -->
Chris Lattner35abbf52007-10-23 06:23:57 +0000[diff] [blame]279<div class="doc_section"><a name="funcs">Function Code Generation</a></div>
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]280<!-- *********************************************************************** -->
281
282<div class="doc_text">
283
Chris Lattner35abbf52007-10-23 06:23:57 +0000[diff] [blame]284<p>Code generation for prototypes and functions has to handle a number of
285details, which make their code less beautiful and elegant than expression code
286generation, but they illustrate some important points. First, lets talk about
287code generation for prototypes: this is used both for function bodies as well
288as external function declarations. The code starts with:</p>
289
290<div class="doc_code">
291<pre>
292Function *PrototypeAST::Codegen() {
293 // Make the function type: double(double,double) etc.
294 std::vector&lt;const Type*&gt; Doubles(Args.size(), Type::DoubleTy);
295 FunctionType *FT = FunctionType::get(Type::DoubleTy, Doubles, false);
296
297 Function *F = new Function(FT, Function::ExternalLinkage, Name, TheModule);
298</pre>
299</div>
300
Chris Lattnercf9893d2007-11-05 19:22:50 +0000[diff] [blame]301<p>This code packs a lot of power into a few lines. Note first that this
302function returns a Function* instead of a Value*. Because a "prototype" really
303talks about the external interface for a function (not the value computed by
304an expression), it makes sense for it to return the LLVM Function it corresponds
305to when codegen'd.</p>
306
307<p>The next step is to create
Chris Lattner35abbf52007-10-23 06:23:57 +0000[diff] [blame]308the <tt>FunctionType</tt> that should be used for a given Prototype. Since all
309function arguments in Kaleidoscope are of type double, the first line creates
310a vector of "N" LLVM Double types. It then uses the <tt>FunctionType::get</tt>
311method to create a function type that takes "N" doubles as arguments, returns
312one double as a result, and that is not vararg (the false parameter indicates
313this). Note that Types in LLVM are uniqued just like Constants are, so you
314don't "new" a type, you "get" it.</p>
315
316<p>The final line above actually creates the function that the prototype will
317correspond to. This indicates which type, linkage, and name to use, and which
318module to insert into. "<a href="LangRef.html#linkage">external linkage</a>"
319means that the function may be defined outside the current module and/or that it
320is callable by functions outside the module. The Name passed in is the name the
321user specified: since "<tt>TheModule</tt>" is specified, this name is registered
322in "<tt>TheModule</tt>"s symbol table, which is used by the function call code
323above.</p>
324
325<div class="doc_code">
326<pre>
327 // If F conflicted, there was already something named 'Name'. If it has a
328 // body, don't allow redefinition or reextern.
329 if (F-&gt;getName() != Name) {
330 // Delete the one we just made and get the existing one.
331 F-&gt;eraseFromParent();
332 F = TheModule-&gt;getFunction(Name);
333</pre>
334</div>
335
336<p>The Module symbol table works just like the Function symbol table when it
337comes to name conflicts: if a new function is created with a name was previously
338added to the symbol table, it will get implicitly renamed when added to the
339Module. The code above exploits this fact to tell if there was a previous
340definition of this function.</p>
341
342<p>In Kaleidoscope, I choose to allow redefinitions of functions in two cases:
343first, we want to allow 'extern'ing a function more than once, so long as the
344prototypes for the externs match (since all arguments have the same type, we
345just have to check that the number of arguments match). Second, we want to
346allow 'extern'ing a function and then definining a body for it. This is useful
347when defining mutually recursive functions.</p>
348
349<p>In order to implement this, the code above first checks to see if there is
350a collision on the name of the function. If so, it deletes the function we just
351created (by calling <tt>eraseFromParent</tt>) and then calling
352<tt>getFunction</tt> to get the existing function with the specified name. Note
353that many APIs in LLVM have "erase" forms and "remove" forms. The "remove" form
354unlinks the object from its parent (e.g. a Function from a Module) and returns
355it. The "erase" form unlinks the object and then deletes it.</p>
356
357<div class="doc_code">
358<pre>
359 // If F already has a body, reject this.
360 if (!F-&gt;empty()) {
361 ErrorF("redefinition of function");
362 return 0;
363 }
364
365 // If F took a different number of args, reject.
366 if (F-&gt;arg_size() != Args.size()) {
367 ErrorF("redefinition of function with different # args");
368 return 0;
369 }
370 }
371</pre>
372</div>
373
374<p>In order to verify the logic above, we first check to see if the preexisting
375function is "empty". In this case, empty means that it has no basic blocks in
376it, which means it has no body. If it has no body, this means its a forward
377declaration. Since we don't allow anything after a full definition of the
378function, the code rejects this case. If the previous reference to a function
379was an 'extern', we simply verify that the number of arguments for that
380definition and this one match up. If not, we emit an error.</p>
381
382<div class="doc_code">
383<pre>
384 // Set names for all arguments.
385 unsigned Idx = 0;
386 for (Function::arg_iterator AI = F-&gt;arg_begin(); Idx != Args.size();
387 ++AI, ++Idx) {
388 AI-&gt;setName(Args[Idx]);
389
390 // Add arguments to variable symbol table.
391 NamedValues[Args[Idx]] = AI;
392 }
393 return F;
394}
395</pre>
396</div>
397
398<p>The last bit of code for prototypes loops over all of the arguments in the
399function, setting the name of the LLVM Argument objects to match and registering
400the arguments in the <tt>NamedValues</tt> map for future use by the
401<tt>VariableExprAST</tt> AST node. Once this is set up, it returns the Function
402object to the caller. Note that we don't check for conflicting
403argument names here (e.g. "extern foo(a b a)"). Doing so would be very
404straight-forward.</p>
405
406<div class="doc_code">
407<pre>
408Function *FunctionAST::Codegen() {
409 NamedValues.clear();
410
411 Function *TheFunction = Proto-&gt;Codegen();
412 if (TheFunction == 0)
413 return 0;
414</pre>
415</div>
416
417<p>Code generation for function definitions starts out simply enough: first we
418codegen the prototype and verify that it is ok. We also clear out the
419<tt>NamedValues</tt> map to make sure that there isn't anything in it from the
420last function we compiled.</p>
421
422<div class="doc_code">
423<pre>
424 // Create a new basic block to start insertion into.
425 BasicBlock *BB = new BasicBlock("entry", TheFunction);
426 Builder.SetInsertPoint(BB);
427
428 if (Value *RetVal = Body-&gt;Codegen()) {
Chris Lattner35abbf52007-10-23 06:23:57 +0000[diff] [blame]429</pre>
430</div>
431
432<p>Now we get to the point where the <tt>Builder</tt> is set up. The first
433line creates a new <a href="http://en.wikipedia.org/wiki/Basic_block">basic
434block</a> (named "entry"), which is inserted into <tt>TheFunction</tt>. The
435second line then tells the builder that new instructions should be inserted into
436the end of the new basic block. Basic blocks in LLVM are an important part
437of functions that define the <a
438href="http://en.wikipedia.org/wiki/Control_flow_graph">Control Flow Graph</a>.
439Since we don't have any control flow, our functions will only contain one
440block so far. We'll fix this in a future installment :).</p>
441
Chris Lattnerd9b86162007-10-25 04:30:35 +0000[diff] [blame]442<div class="doc_code">
443<pre>
444 if (Value *RetVal = Body-&gt;Codegen()) {
445 // Finish off the function.
446 Builder.CreateRet(RetVal);
447
448 // Validate the generated code, checking for consistency.
449 verifyFunction(*TheFunction);
450 return TheFunction;
451 }
452</pre>
453</div>
454
Chris Lattner35abbf52007-10-23 06:23:57 +0000[diff] [blame]455<p>Once the insertion point is set up, we call the <tt>CodeGen()</tt> method for
456the root expression of the function. If no error happens, this emits code to
457compute the expression into the entry block and returns the value that was
458computed. Assuming no error, we then create an LLVM <a
Chris Lattnerd9b86162007-10-25 04:30:35 +0000[diff] [blame]459href="../LangRef.html#i_ret">ret instruction</a>, which completes the function.
460Once the function is built, we call the <tt>verifyFunction</tt> function, which
461is provided by LLVM. This function does a variety of consistency checks on the
462generated code, to determine if our compiler is doing everything right. Using
463this is important: it can catch a lot of bugs. Once the function is finished
464and validated, we return it.</p>
Chris Lattner35abbf52007-10-23 06:23:57 +0000[diff] [blame]465
466<div class="doc_code">
467<pre>
468 // Error reading body, remove function.
469 TheFunction-&gt;eraseFromParent();
470 return 0;
471}
472</pre>
473</div>
474
475<p>The only piece left here is handling of the error case. For simplicity, we
476simply handle this by deleting the function we produced with the
477<tt>eraseFromParent</tt> method. This allows the user to redefine a function
478that they incorrectly typed in before: if we didn't delete it, it would live in
479the symbol table, with a body, preventing future redefinition.</p>
480
481<p>This code does have a bug though. Since the <tt>PrototypeAST::Codegen</tt>
482can return a previously defined forward declaration, this can actually delete
483a forward declaration. There are a number of ways to fix this bug, see what you
484can come up with! Here is a testcase:</p>
485
486<div class="doc_code">
487<pre>
488extern foo(a b); # ok, defines foo.
489def foo(a b) c; # error, 'c' is invalid.
490def bar() foo(1, 2); # error, unknown function "foo"
491</pre>
492</div>
493
494</div>
495
496<!-- *********************************************************************** -->
497<div class="doc_section"><a name="driver">Driver Changes and
498Closing Thoughts</a></div>
499<!-- *********************************************************************** -->
500
501<div class="doc_text">
502
503<p>
504For now, code generation to LLVM doesn't really get us much, except that we can
505look at the pretty IR calls. The sample code inserts calls to Codegen into the
506"<tt>HandleDefinition</tt>", "<tt>HandleExtern</tt>" etc functions, and then
507dumps out the LLVM IR. This gives a nice way to look at the LLVM IR for simple
508functions. For example:
509</p>
510
511<div class="doc_code">
512<pre>
513ready> <b>4+5</b>;
514ready> Read top-level expression:
515define double @""() {
516entry:
517 %addtmp = add double 4.000000e+00, 5.000000e+00
518 ret double %addtmp
519}
520</pre>
521</div>
522
523<p>Note how the parser turns the top-level expression into anonymous functions
524for us. This will be handy when we add JIT support in the next chapter. Also
525note that the code is very literally transcribed, no optimizations are being
526performed. We will add optimizations explicitly in the next chapter.</p>
527
528<div class="doc_code">
529<pre>
530ready&gt; <b>def foo(a b) a*a + 2*a*b + b*b;</b>
531ready&gt; Read function definition:
532define double @foo(double %a, double %b) {
533entry:
534 %multmp = mul double %a, %a
535 %multmp1 = mul double 2.000000e+00, %a
536 %multmp2 = mul double %multmp1, %b
537 %addtmp = add double %multmp, %multmp2
538 %multmp3 = mul double %b, %b
539 %addtmp4 = add double %addtmp, %multmp3
540 ret double %addtmp4
541}
542</pre>
543</div>
544
545<p>This shows some simple arithmetic. Notice the striking similarity to the
546LLVM builder calls that we use to create the instructions.</p>
547
548<div class="doc_code">
549<pre>
550ready&gt; <b>def bar(a) foo(a, 4.0) + bar(31337);</b>
551ready&gt; Read function definition:
552define double @bar(double %a) {
553entry:
554 %calltmp = call double @foo( double %a, double 4.000000e+00 )
555 %calltmp1 = call double @bar( double 3.133700e+04 )
556 %addtmp = add double %calltmp, %calltmp1
557 ret double %addtmp
558}
559</pre>
560</div>
561
Chris Lattnerfc60ab02007-11-05 17:39:26 +0000[diff] [blame]562<p>This shows some function calls. Note that this function will take a long
563time to execute if you call it. In the future we'll add conditional control
564flow to make recursion actually be useful :).</p>
Chris Lattner35abbf52007-10-23 06:23:57 +0000[diff] [blame]565
566<div class="doc_code">
567<pre>
568ready&gt; <b>extern cos(x);</b>
569ready&gt; Read extern:
570declare double @cos(double)
571
572ready&gt; <b>cos(1.234);</b>
573ready&gt; Read top-level expression:
574define double @""() {
575entry:
Chris Lattner8eef4b22007-10-23 06:30:50 +0000[diff] [blame]576 %calltmp = call double @cos( double 1.234000e+00 )
Chris Lattner35abbf52007-10-23 06:23:57 +0000[diff] [blame]577 ret double %calltmp
578}
579</pre>
580</div>
581
582<p>This shows an extern for the libm "cos" function, and a call to it.</p>
583
584
585<div class="doc_code">
586<pre>
587ready&gt; <b>^D</b>
588; ModuleID = 'my cool jit'
589
590define double @""() {
591entry:
592 %addtmp = add double 4.000000e+00, 5.000000e+00
593 ret double %addtmp
594}
595
596define double @foo(double %a, double %b) {
597entry:
598 %multmp = mul double %a, %a
599 %multmp1 = mul double 2.000000e+00, %a
600 %multmp2 = mul double %multmp1, %b
601 %addtmp = add double %multmp, %multmp2
602 %multmp3 = mul double %b, %b
603 %addtmp4 = add double %addtmp, %multmp3
604 ret double %addtmp4
605}
606
607define double @bar(double %a) {
608entry:
609 %calltmp = call double @foo( double %a, double 4.000000e+00 )
610 %calltmp1 = call double @bar( double 3.133700e+04 )
611 %addtmp = add double %calltmp, %calltmp1
612 ret double %addtmp
613}
614
615declare double @cos(double)
616
617define double @""() {
618entry:
619 %calltmp = call double @cos( double 1.234000e+00 )
620 ret double %calltmp
621}
622</pre>
623</div>
624
625<p>When you quit the current demo, it dumps out the IR for the entire module
626generated. Here you can see the big picture with all the functions referencing
627each other.</p>
628
629<p>This wraps up this chapter of the Kaleidoscope tutorial. Up next we'll
630describe how to <a href="LangImpl4.html">add JIT codegen and optimizer
631support</a> to this so we can actually start running code!</p>
632
633</div>
634
635
636<!-- *********************************************************************** -->
637<div class="doc_section"><a name="code">Full Code Listing</a></div>
638<!-- *********************************************************************** -->
639
640<div class="doc_text">
641
642<p>
643Here is the complete code listing for our running example, enhanced with the
644LLVM code generator. Because this uses the LLVM libraries, we need to link
645them in. To do this, we use the <a
646href="http://llvm.org/cmds/llvm-config.html">llvm-config</a> tool to inform
647our makefile/command line about which options to use:</p>
648
649<div class="doc_code">
650<pre>
651 # Compile
Chris Lattner9ac0ca02007-10-24 05:09:48 +0000[diff] [blame]652 g++ -g toy.cpp `llvm-config --cppflags --ldflags --libs core` -o toy
Chris Lattner35abbf52007-10-23 06:23:57 +0000[diff] [blame]653 # Run
654 ./toy
655</pre>
656</div>
657
658<p>Here is the code:</p>
659
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]660<div class="doc_code">
661<pre>
662// To build this:
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]663// See example below.
664
665#include "llvm/DerivedTypes.h"
666#include "llvm/Module.h"
Chris Lattnerd9b86162007-10-25 04:30:35 +0000[diff] [blame]667#include "llvm/Analysis/Verifier.h"
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]668#include "llvm/Support/LLVMBuilder.h"
669#include &lt;cstdio&gt;
670#include &lt;string&gt;
671#include &lt;map&gt;
672#include &lt;vector&gt;
673using namespace llvm;
674
675//===----------------------------------------------------------------------===//
676// Lexer
677//===----------------------------------------------------------------------===//
678
679// The lexer returns tokens [0-255] if it is an unknown character, otherwise one
680// of these for known things.
681enum Token {
682 tok_eof = -1,
683
684 // commands
685 tok_def = -2, tok_extern = -3,
686
687 // primary
688 tok_identifier = -4, tok_number = -5,
689};
690
691static std::string IdentifierStr; // Filled in if tok_identifier
692static double NumVal; // Filled in if tok_number
693
694/// gettok - Return the next token from standard input.
695static int gettok() {
696 static int LastChar = ' ';
697
698 // Skip any whitespace.
699 while (isspace(LastChar))
700 LastChar = getchar();
701
702 if (isalpha(LastChar)) { // identifier: [a-zA-Z][a-zA-Z0-9]*
703 IdentifierStr = LastChar;
704 while (isalnum((LastChar = getchar())))
705 IdentifierStr += LastChar;
706
707 if (IdentifierStr == "def") return tok_def;
708 if (IdentifierStr == "extern") return tok_extern;
709 return tok_identifier;
710 }
711
712 if (isdigit(LastChar) || LastChar == '.') { // Number: [0-9.]+
713 std::string NumStr;
714 do {
715 NumStr += LastChar;
716 LastChar = getchar();
717 } while (isdigit(LastChar) || LastChar == '.');
718
719 NumVal = strtod(NumStr.c_str(), 0);
720 return tok_number;
721 }
722
723 if (LastChar == '#') {
724 // Comment until end of line.
725 do LastChar = getchar();
726 while (LastChar != EOF &amp;&amp; LastChar != '\n' &amp; LastChar != '\r');
727
728 if (LastChar != EOF)
729 return gettok();
730 }
731
732 // Check for end of file. Don't eat the EOF.
733 if (LastChar == EOF)
734 return tok_eof;
735
736 // Otherwise, just return the character as its ascii value.
737 int ThisChar = LastChar;
738 LastChar = getchar();
739 return ThisChar;
740}
741
742//===----------------------------------------------------------------------===//
743// Abstract Syntax Tree (aka Parse Tree)
744//===----------------------------------------------------------------------===//
745
746/// ExprAST - Base class for all expression nodes.
747class ExprAST {
748public:
749 virtual ~ExprAST() {}
750 virtual Value *Codegen() = 0;
751};
752
753/// NumberExprAST - Expression class for numeric literals like "1.0".
754class NumberExprAST : public ExprAST {
755 double Val;
756public:
Chris Lattner28571ed2007-10-23 04:27:44 +0000[diff] [blame]757 explicit NumberExprAST(double val) : Val(val) {}
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]758 virtual Value *Codegen();
759};
760
761/// VariableExprAST - Expression class for referencing a variable, like "a".
762class VariableExprAST : public ExprAST {
763 std::string Name;
764public:
Chris Lattner28571ed2007-10-23 04:27:44 +0000[diff] [blame]765 explicit VariableExprAST(const std::string &amp;name) : Name(name) {}
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]766 virtual Value *Codegen();
767};
768
769/// BinaryExprAST - Expression class for a binary operator.
770class BinaryExprAST : public ExprAST {
771 char Op;
772 ExprAST *LHS, *RHS;
773public:
774 BinaryExprAST(char op, ExprAST *lhs, ExprAST *rhs)
775 : Op(op), LHS(lhs), RHS(rhs) {}
776 virtual Value *Codegen();
777};
778
779/// CallExprAST - Expression class for function calls.
780class CallExprAST : public ExprAST {
781 std::string Callee;
782 std::vector&lt;ExprAST*&gt; Args;
783public:
784 CallExprAST(const std::string &amp;callee, std::vector&lt;ExprAST*&gt; &amp;args)
785 : Callee(callee), Args(args) {}
786 virtual Value *Codegen();
787};
788
789/// PrototypeAST - This class represents the "prototype" for a function,
790/// which captures its argument names as well as if it is an operator.
791class PrototypeAST {
792 std::string Name;
793 std::vector&lt;std::string&gt; Args;
794public:
795 PrototypeAST(const std::string &amp;name, const std::vector&lt;std::string&gt; &amp;args)
796 : Name(name), Args(args) {}
797
798 Function *Codegen();
799};
800
801/// FunctionAST - This class represents a function definition itself.
802class FunctionAST {
803 PrototypeAST *Proto;
804 ExprAST *Body;
805public:
806 FunctionAST(PrototypeAST *proto, ExprAST *body)
807 : Proto(proto), Body(body) {}
808
809 Function *Codegen();
810};
811
812//===----------------------------------------------------------------------===//
813// Parser
814//===----------------------------------------------------------------------===//
815
816/// CurTok/getNextToken - Provide a simple token buffer. CurTok is the current
817/// token the parser it looking at. getNextToken reads another token from the
818/// lexer and updates CurTok with its results.
819static int CurTok;
820static int getNextToken() {
821 return CurTok = gettok();
822}
823
824/// BinopPrecedence - This holds the precedence for each binary operator that is
825/// defined.
826static std::map&lt;char, int&gt; BinopPrecedence;
827
828/// GetTokPrecedence - Get the precedence of the pending binary operator token.
829static int GetTokPrecedence() {
830 if (!isascii(CurTok))
831 return -1;
832
833 // Make sure it's a declared binop.
834 int TokPrec = BinopPrecedence[CurTok];
835 if (TokPrec &lt;= 0) return -1;
836 return TokPrec;
837}
838
839/// Error* - These are little helper functions for error handling.
840ExprAST *Error(const char *Str) { fprintf(stderr, "Error: %s\n", Str);return 0;}
841PrototypeAST *ErrorP(const char *Str) { Error(Str); return 0; }
842FunctionAST *ErrorF(const char *Str) { Error(Str); return 0; }
843
844static ExprAST *ParseExpression();
845
846/// identifierexpr
Chris Lattner20a0c802007-11-05 17:54:34 +0000[diff] [blame]847/// ::= identifier
848/// ::= identifier '(' expression* ')'
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]849static ExprAST *ParseIdentifierExpr() {
850 std::string IdName = IdentifierStr;
851
Chris Lattner20a0c802007-11-05 17:54:34 +0000[diff] [blame]852 getNextToken(); // eat identifier.
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]853
854 if (CurTok != '(') // Simple variable ref.
855 return new VariableExprAST(IdName);
856
857 // Call.
858 getNextToken(); // eat (
859 std::vector&lt;ExprAST*&gt; Args;
860 while (1) {
861 ExprAST *Arg = ParseExpression();
862 if (!Arg) return 0;
863 Args.push_back(Arg);
864
865 if (CurTok == ')') break;
866
867 if (CurTok != ',')
868 return Error("Expected ')'");
869 getNextToken();
870 }
871
872 // Eat the ')'.
873 getNextToken();
874
875 return new CallExprAST(IdName, Args);
876}
877
878/// numberexpr ::= number
879static ExprAST *ParseNumberExpr() {
880 ExprAST *Result = new NumberExprAST(NumVal);
881 getNextToken(); // consume the number
882 return Result;
883}
884
885/// parenexpr ::= '(' expression ')'
886static ExprAST *ParseParenExpr() {
887 getNextToken(); // eat (.
888 ExprAST *V = ParseExpression();
889 if (!V) return 0;
890
891 if (CurTok != ')')
892 return Error("expected ')'");
893 getNextToken(); // eat ).
894 return V;
895}
896
897/// primary
898/// ::= identifierexpr
899/// ::= numberexpr
900/// ::= parenexpr
901static ExprAST *ParsePrimary() {
902 switch (CurTok) {
903 default: return Error("unknown token when expecting an expression");
904 case tok_identifier: return ParseIdentifierExpr();
905 case tok_number: return ParseNumberExpr();
906 case '(': return ParseParenExpr();
907 }
908}
909
910/// binoprhs
911/// ::= ('+' primary)*
912static ExprAST *ParseBinOpRHS(int ExprPrec, ExprAST *LHS) {
913 // If this is a binop, find its precedence.
914 while (1) {
915 int TokPrec = GetTokPrecedence();
916
917 // If this is a binop that binds at least as tightly as the current binop,
918 // consume it, otherwise we are done.
919 if (TokPrec &lt; ExprPrec)
920 return LHS;
921
922 // Okay, we know this is a binop.
923 int BinOp = CurTok;
924 getNextToken(); // eat binop
925
926 // Parse the primary expression after the binary operator.
927 ExprAST *RHS = ParsePrimary();
928 if (!RHS) return 0;
929
930 // If BinOp binds less tightly with RHS than the operator after RHS, let
931 // the pending operator take RHS as its LHS.
932 int NextPrec = GetTokPrecedence();
933 if (TokPrec &lt; NextPrec) {
934 RHS = ParseBinOpRHS(TokPrec+1, RHS);
935 if (RHS == 0) return 0;
936 }
937
938 // Merge LHS/RHS.
939 LHS = new BinaryExprAST(BinOp, LHS, RHS);
940 }
941}
942
943/// expression
944/// ::= primary binoprhs
945///
946static ExprAST *ParseExpression() {
947 ExprAST *LHS = ParsePrimary();
948 if (!LHS) return 0;
949
950 return ParseBinOpRHS(0, LHS);
951}
952
953/// prototype
954/// ::= id '(' id* ')'
955static PrototypeAST *ParsePrototype() {
956 if (CurTok != tok_identifier)
957 return ErrorP("Expected function name in prototype");
958
959 std::string FnName = IdentifierStr;
960 getNextToken();
961
962 if (CurTok != '(')
963 return ErrorP("Expected '(' in prototype");
964
965 std::vector&lt;std::string&gt; ArgNames;
966 while (getNextToken() == tok_identifier)
967 ArgNames.push_back(IdentifierStr);
968 if (CurTok != ')')
969 return ErrorP("Expected ')' in prototype");
970
971 // success.
972 getNextToken(); // eat ')'.
973
974 return new PrototypeAST(FnName, ArgNames);
975}
976
977/// definition ::= 'def' prototype expression
978static FunctionAST *ParseDefinition() {
979 getNextToken(); // eat def.
980 PrototypeAST *Proto = ParsePrototype();
981 if (Proto == 0) return 0;
982
983 if (ExprAST *E = ParseExpression())
984 return new FunctionAST(Proto, E);
985 return 0;
986}
987
988/// toplevelexpr ::= expression
989static FunctionAST *ParseTopLevelExpr() {
990 if (ExprAST *E = ParseExpression()) {
991 // Make an anonymous proto.
992 PrototypeAST *Proto = new PrototypeAST("", std::vector&lt;std::string&gt;());
993 return new FunctionAST(Proto, E);
994 }
995 return 0;
996}
997
998/// external ::= 'extern' prototype
999static PrototypeAST *ParseExtern() {
1000 getNextToken(); // eat extern.
1001 return ParsePrototype();
1002}
1003
1004//===----------------------------------------------------------------------===//
1005// Code Generation
1006//===----------------------------------------------------------------------===//
1007
1008static Module *TheModule;
1009static LLVMBuilder Builder;
1010static std::map&lt;std::string, Value*&gt; NamedValues;
1011
1012Value *ErrorV(const char *Str) { Error(Str); return 0; }
1013
1014Value *NumberExprAST::Codegen() {
1015 return ConstantFP::get(Type::DoubleTy, APFloat(Val));
1016}
1017
1018Value *VariableExprAST::Codegen() {
1019 // Look this variable up in the function.
1020 Value *V = NamedValues[Name];
1021 return V ? V : ErrorV("Unknown variable name");
1022}
1023
1024Value *BinaryExprAST::Codegen() {
1025 Value *L = LHS-&gt;Codegen();
1026 Value *R = RHS-&gt;Codegen();
1027 if (L == 0 || R == 0) return 0;
1028
1029 switch (Op) {
1030 case '+': return Builder.CreateAdd(L, R, "addtmp");
1031 case '-': return Builder.CreateSub(L, R, "subtmp");
1032 case '*': return Builder.CreateMul(L, R, "multmp");
1033 case '&lt;':
1034 L = Builder.CreateFCmpULT(L, R, "multmp");
1035 // Convert bool 0/1 to double 0.0 or 1.0
1036 return Builder.CreateUIToFP(L, Type::DoubleTy, "booltmp");
1037 default: return ErrorV("invalid binary operator");
1038 }
1039}
1040
1041Value *CallExprAST::Codegen() {
1042 // Look up the name in the global module table.
1043 Function *CalleeF = TheModule-&gt;getFunction(Callee);
1044 if (CalleeF == 0)
1045 return ErrorV("Unknown function referenced");
1046
1047 // If argument mismatch error.
1048 if (CalleeF-&gt;arg_size() != Args.size())
1049 return ErrorV("Incorrect # arguments passed");
1050
1051 std::vector&lt;Value*&gt; ArgsV;
1052 for (unsigned i = 0, e = Args.size(); i != e; ++i) {
1053 ArgsV.push_back(Args[i]-&gt;Codegen());
1054 if (ArgsV.back() == 0) return 0;
1055 }
1056
1057 return Builder.CreateCall(CalleeF, ArgsV.begin(), ArgsV.end(), "calltmp");
1058}
1059
1060Function *PrototypeAST::Codegen() {
1061 // Make the function type: double(double,double) etc.
Chris Lattner35abbf52007-10-23 06:23:57 +0000[diff] [blame]1062 std::vector&lt;const Type*&gt; Doubles(Args.size(), Type::DoubleTy);
1063 FunctionType *FT = FunctionType::get(Type::DoubleTy, Doubles, false);
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]1064
1065 Function *F = new Function(FT, Function::ExternalLinkage, Name, TheModule);
1066
1067 // If F conflicted, there was already something named 'Name'. If it has a
1068 // body, don't allow redefinition or reextern.
1069 if (F-&gt;getName() != Name) {
1070 // Delete the one we just made and get the existing one.
1071 F-&gt;eraseFromParent();
1072 F = TheModule-&gt;getFunction(Name);
1073
1074 // If F already has a body, reject this.
1075 if (!F-&gt;empty()) {
1076 ErrorF("redefinition of function");
1077 return 0;
1078 }
1079
1080 // If F took a different number of args, reject.
1081 if (F-&gt;arg_size() != Args.size()) {
1082 ErrorF("redefinition of function with different # args");
1083 return 0;
1084 }
1085 }
1086
1087 // Set names for all arguments.
1088 unsigned Idx = 0;
1089 for (Function::arg_iterator AI = F-&gt;arg_begin(); Idx != Args.size();
1090 ++AI, ++Idx) {
1091 AI-&gt;setName(Args[Idx]);
1092
1093 // Add arguments to variable symbol table.
1094 NamedValues[Args[Idx]] = AI;
1095 }
1096
1097 return F;
1098}
1099
1100Function *FunctionAST::Codegen() {
1101 NamedValues.clear();
1102
1103 Function *TheFunction = Proto-&gt;Codegen();
1104 if (TheFunction == 0)
1105 return 0;
1106
1107 // Create a new basic block to start insertion into.
Chris Lattner35abbf52007-10-23 06:23:57 +0000[diff] [blame]1108 BasicBlock *BB = new BasicBlock("entry", TheFunction);
1109 Builder.SetInsertPoint(BB);
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]1110
1111 if (Value *RetVal = Body-&gt;Codegen()) {
1112 // Finish off the function.
1113 Builder.CreateRet(RetVal);
Chris Lattnerd9b86162007-10-25 04:30:35 +0000[diff] [blame]1114
1115 // Validate the generated code, checking for consistency.
1116 verifyFunction(*TheFunction);
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]1117 return TheFunction;
1118 }
1119
1120 // Error reading body, remove function.
1121 TheFunction-&gt;eraseFromParent();
1122 return 0;
1123}
1124
1125//===----------------------------------------------------------------------===//
1126// Top-Level parsing and JIT Driver
1127//===----------------------------------------------------------------------===//
1128
1129static void HandleDefinition() {
1130 if (FunctionAST *F = ParseDefinition()) {
1131 if (Function *LF = F-&gt;Codegen()) {
1132 fprintf(stderr, "Read function definition:");
1133 LF-&gt;dump();
1134 }
1135 } else {
1136 // Skip token for error recovery.
1137 getNextToken();
1138 }
1139}
1140
1141static void HandleExtern() {
1142 if (PrototypeAST *P = ParseExtern()) {
1143 if (Function *F = P-&gt;Codegen()) {
1144 fprintf(stderr, "Read extern: ");
1145 F-&gt;dump();
1146 }
1147 } else {
1148 // Skip token for error recovery.
1149 getNextToken();
1150 }
1151}
1152
1153static void HandleTopLevelExpression() {
1154 // Evaluate a top level expression into an anonymous function.
1155 if (FunctionAST *F = ParseTopLevelExpr()) {
1156 if (Function *LF = F-&gt;Codegen()) {
1157 fprintf(stderr, "Read top-level expression:");
1158 LF-&gt;dump();
1159 }
1160 } else {
1161 // Skip token for error recovery.
1162 getNextToken();
1163 }
1164}
1165
1166/// top ::= definition | external | expression | ';'
1167static void MainLoop() {
1168 while (1) {
1169 fprintf(stderr, "ready&gt; ");
1170 switch (CurTok) {
1171 case tok_eof: return;
1172 case ';': getNextToken(); break; // ignore top level semicolons.
1173 case tok_def: HandleDefinition(); break;
1174 case tok_extern: HandleExtern(); break;
1175 default: HandleTopLevelExpression(); break;
1176 }
1177 }
1178}
1179
1180
1181
1182//===----------------------------------------------------------------------===//
1183// "Library" functions that can be "extern'd" from user code.
1184//===----------------------------------------------------------------------===//
1185
1186/// putchard - putchar that takes a double and returns 0.
1187extern "C"
1188double putchard(double X) {
1189 putchar((char)X);
1190 return 0;
1191}
1192
1193//===----------------------------------------------------------------------===//
1194// Main driver code.
1195//===----------------------------------------------------------------------===//
1196
1197int main() {
1198 TheModule = new Module("my cool jit");
1199
1200 // Install standard binary operators.
1201 // 1 is lowest precedence.
1202 BinopPrecedence['&lt;'] = 10;
1203 BinopPrecedence['+'] = 20;
1204 BinopPrecedence['-'] = 20;
1205 BinopPrecedence['*'] = 40; // highest.
1206
1207 // Prime the first token.
1208 fprintf(stderr, "ready&gt; ");
1209 getNextToken();
1210
1211 MainLoop();
1212 TheModule-&gt;dump();
1213 return 0;
1214}
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]1215</pre>
1216</div>
1217</div>
1218
1219<!-- *********************************************************************** -->
1220<hr>
1221<address>
1222 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
1223 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
1224 <a href="http://validator.w3.org/check/referer"><img
Chris Lattner8eef4b22007-10-23 06:30:50 +0000[diff] [blame]1225 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
Chris Lattner2e902042007-10-22 07:01:42 +0000[diff] [blame]1226
1227 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
1228 <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br>
1229 Last modified: $Date: 2007-10-17 11:05:13 -0700 (Wed, 17 Oct 2007) $
1230</address>
1231</body>
1232</html>