| blob | 98c1124cc12cda7174a058c087b2bf8dfd2f182a |
4 <html>
5 <head>
11 </head>
13 <body>
17 <ul>
20 <ol>
24 </ol>
25 </li>
27 AST</li>
28 </ul>
31 <p>
34 </p>
35 </div>
37 <!-- *********************************************************************** -->
39 <!-- *********************************************************************** -->
44 runs through the implementation of a simple language, showing how fun and
45 easy it can be. This tutorial will get you up and started as well as help to
46 build a framework you can extend to other languages. The code in this tutorial
47 can also be used as a playground to hack on other LLVM specific things.
48 </p>
50 <p>
51 The goal of this tutorial is to progressively unveil our language, describing
52 how it is built up over time. This will let us cover a fairly broad range of
53 language design and LLVM-specific usage issues, showing and explaining the code
54 for it all along the way, without overwhelming you with tons of details up
55 front.</p>
57 <p>It is useful to point out ahead of time that this tutorial is really about
59 modern and sane software engineering principles. In practice, this means that
60 we'll take a number of shortcuts to simplify the exposition. For example, the
61 code leaks memory, uses global variables all over the place, doesn't use nice
62 design patterns like <a
64 is very simple. If you dig in and use the code as a basis for future projects,
65 fixing these deficiencies shouldn't be hard.</p>
67 <p>I've tried to put this tutorial together in a way that makes chapters easy to
68 skip over if you are already familiar with or are uninterested in the various
69 pieces. The structure of the tutorial is:
70 </p>
72 <ul>
74 language, and the definition of its Lexer</b> - This shows where we are going
75 and the basic functionality that we want it to do. In order to make this
76 tutorial maximally understandable and hackable, we choose to implement
77 everything in Objective Caml instead of using lexer and parser generators.
78 LLVM obviously works just fine with such tools, feel free to use one if you
79 prefer.</li>
81 AST</b> - With the lexer in place, we can talk about parsing techniques and
82 basic AST construction. This tutorial describes recursive descent parsing and
84 the code doesn't even link in LLVM at this point. :)</li>
86 IR</b> - With the AST ready, we can show off how easy generation of LLVM IR
87 really is.</li>
89 Support</b> - Because a lot of people are interested in using LLVM as a JIT,
90 we'll dive right into it and show you the 3 lines it takes to add JIT support.
91 LLVM is also useful in many other ways, but this is one simple and "sexy" way
92 to shows off its power. :)</li>
94 Control Flow</b> - With the language up and running, we show how to extend it
95 with control flow operations (if/then/else and a 'for' loop). This gives us a
96 chance to talk about simple SSA construction and control flow.</li>
98 User-defined Operators</b> - This is a silly but fun chapter that talks about
99 extending the language to let the user program define their own arbitrary
100 unary and binary operators (with assignable precedence!). This lets us build a
103 Mutable Variables</b> - This chapter talks about adding user-defined local
104 variables along with an assignment operator. The interesting part about this
105 is how easy and trivial it is to construct SSA form in LLVM: no, LLVM does
108 useful LLVM tidbits</b> - This chapter wraps up the series by talking about
109 potential ways to extend the language, but also includes a bunch of pointers to
110 info about "special topics" like adding garbage collection support, exceptions,
111 debugging, support for "spaghetti stacks", and a bunch of other tips and
112 tricks.</li>
114 </ul>
117 non-comment, non-blank, lines of code. With this small amount of code, we'll
118 have built up a very reasonable compiler for a non-trivial language including
119 a hand-written lexer, parser, AST, as well as code generation support with a JIT
120 compiler. While other systems may have interesting "hello world" tutorials,
121 I think the breadth of this tutorial is a great testament to the strengths of
122 LLVM and why you should consider it if you're interested in language or compiler
123 design.</p>
125 <p>A note about this tutorial: we expect you to extend the language and play
126 with it on your own. Take the code and go crazy hacking away at it, compilers
127 don't need to be scary creatures - it can be a lot of fun to play with
128 languages!</p>
130 </div>
132 <!-- *********************************************************************** -->
134 <!-- *********************************************************************** -->
138 <p>This tutorial will be illustrated with a toy language that we'll call
140 from "meaning beautiful, form, and view").
141 Kaleidoscope is a procedural language that allows you to define functions, use
142 conditionals, math, etc. Over the course of the tutorial, we'll extend
143 Kaleidoscope to support the if/then/else construct, a for loop, user defined
144 operators, JIT compilation with a simple command line interface, etc.</p>
146 <p>Because we want to keep things simple, the only datatype in Kaleidoscope is a
147 64-bit floating point type (aka 'float' in O'Caml parlance). As such, all
148 values are implicitly double precision and the language doesn't require type
149 declarations. This gives the language a very nice and simple syntax. For
150 example, the following simple example computes <a
154 <pre>
155 # Compute the x'th fibonacci number.
156 def fib(x)
158 1
159 else
162 # This expression will compute the 40th number.
163 fib(40)
164 </pre>
165 </div>
167 <p>We also allow Kaleidoscope to call into standard library functions (the LLVM
168 JIT makes this completely trivial). This means that you can use the 'extern'
169 keyword to define a function before you use it (this is also useful for mutually
170 recursive functions). For example:</p>
173 <pre>
174 extern sin(arg);
175 extern cos(arg);
176 extern atan2(arg1 arg2);
179 </pre>
180 </div>
188 </div>
190 <!-- *********************************************************************** -->
192 <!-- *********************************************************************** -->
196 <p>When it comes to implementing a language, the first thing needed is
197 the ability to process a text file and recognize what it says. The traditional
198 way to do this is to use a "<a
200 to break the input up into "tokens". Each token returned by the lexer includes
201 a token code and potentially some metadata (e.g. the numeric value of a number).
202 First, we define the possibilities:
203 </p>
206 <pre>
207 (* The lexer returns these 'Kwd' if it is an unknown character, otherwise one of
208 * these others for known things. *)
209 type token =
210 (* commands *)
211 | Def | Extern
213 (* primary *)
214 | Ident of string | Number of float
216 (* unknown *)
217 | Kwd of char
218 </pre>
219 </div>
221 <p>Each token returned by our lexer will be one of the token variant values.
224 the current token is a numeric literal (like 1.0), the value will be
226 </p>
228 <p>The actual implementation of the lexer is a collection of functions driven
230 called to return the next token from standard input. We will use
232 to simplify the tokenization of the standard input. Its definition starts
233 as:</p>
236 <pre>
237 (*===----------------------------------------------------------------------===
238 * Lexer
239 *===----------------------------------------------------------------------===*)
241 let rec lex = parser
242 (* Skip any whitespace. *)
244 </pre>
245 </div>
247 <p>
249 characters one at a time from the standard input. It eats them as it recognizes
251 it has to do is ignore whitespace between tokens. This is accomplished with the
252 recursive call above.</p>
255 specific keywords like "def". Kaleidoscope does this with a pattern match
256 and a helper function.<p>
259 <pre>
260 (* identifier: [a-zA-Z][a-zA-Z0-9] *)
262 let buffer = Buffer.create 1 in
263 Buffer.add_char buffer c;
264 lex_ident buffer stream
266 ...
268 and lex_ident buffer = parser
270 Buffer.add_char buffer c;
271 lex_ident buffer stream
273 match Buffer.contents buffer with
277 </pre>
278 </div>
283 <pre>
286 let buffer = Buffer.create 1 in
287 Buffer.add_char buffer c;
288 lex_number buffer stream
290 ...
292 and lex_number buffer = parser
294 Buffer.add_char buffer c;
295 lex_number buffer stream
298 </pre>
299 </div>
301 <p>This is all pretty straight-forward code for processing input. When reading
305 if the string "1.23.45.67". Feel free to extend it :). Next we handle
306 comments:
307 </p>
310 <pre>
311 (* Comment until end of line. *)
313 lex_comment stream
315 ...
317 and lex_comment = parser
321 </pre>
322 </div>
324 <p>We handle comments by skipping to the end of the line and then return the
325 next token. Finally, if the input doesn't match one of the above cases, it is
326 either an operator character like '+' or the end of the file. These are handled
327 with this code:</p>
330 <pre>
331 (* Otherwise, just return the character as its ascii value. *)
335 (* end of stream. *)
337 </pre>
338 </div>
340 <p>With this, we have the complete lexer for the basic Kaleidoscope language
344 uses this to build an Abstract Syntax Tree</a>. When we have that, we'll
345 include a driver so that you can use the lexer and parser together.
346 </p>
349 </div>
351 <!-- *********************************************************************** -->
352 <hr>
353 <address>
362 Last modified: $Date$
363 </address>
364 </body>
365 </html>