1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
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 <meta name="author" content="Erick Tryzelaar">
10 <link rel="stylesheet" href="../_static/llvm.css" type="text/css">
15 <h1>Kaleidoscope: Code generation to LLVM IR</h1>
18 <li><a href="index.html">Up to Tutorial Index</a></li>
21 <li><a href="#intro">Chapter 3 Introduction</a></li>
22 <li><a href="#basics">Code Generation Setup</a></li>
23 <li><a href="#exprs">Expression Code Generation</a></li>
24 <li><a href="#funcs">Function Code Generation</a></li>
25 <li><a href="#driver">Driver Changes and Closing Thoughts</a></li>
26 <li><a href="#code">Full Code Listing</a></li>
29 <li><a href="OCamlLangImpl4.html">Chapter 4</a>: Adding JIT and Optimizer
33 <div class="doc_author">
35 Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>
36 and <a href="mailto:idadesub@users.sourceforge.net">Erick Tryzelaar</a>
40 <!-- *********************************************************************** -->
41 <h2><a name="intro">Chapter 3 Introduction</a></h2>
42 <!-- *********************************************************************** -->
46 <p>Welcome to Chapter 3 of the "<a href="index.html">Implementing a language
47 with LLVM</a>" tutorial. This chapter shows you how to transform the <a
48 href="OCamlLangImpl2.html">Abstract Syntax Tree</a>, built in Chapter 2, into
49 LLVM IR. This will teach you a little bit about how LLVM does things, as well
50 as demonstrate how easy it is to use. It's much more work to build a lexer and
51 parser than it is to generate LLVM IR code. :)
54 <p><b>Please note</b>: the code in this chapter and later require LLVM 2.3 or
55 LLVM SVN to work. LLVM 2.2 and before will not work with it.</p>
59 <!-- *********************************************************************** -->
60 <h2><a name="basics">Code Generation Setup</a></h2>
61 <!-- *********************************************************************** -->
66 In order to generate LLVM IR, we want some simple setup to get started. First
67 we define virtual code generation (codegen) methods in each AST class:</p>
69 <div class="doc_code">
71 let rec codegen_expr = function
72 | Ast.Number n -> ...
73 | Ast.Variable name -> ...
77 <p>The <tt>Codegen.codegen_expr</tt> function says to emit IR for that AST node
78 along with all the things it depends on, and they all return an LLVM Value
79 object. "Value" is the class used to represent a "<a
80 href="http://en.wikipedia.org/wiki/Static_single_assignment_form">Static Single
81 Assignment (SSA)</a> register" or "SSA value" in LLVM. The most distinct aspect
82 of SSA values is that their value is computed as the related instruction
83 executes, and it does not get a new value until (and if) the instruction
84 re-executes. In other words, there is no way to "change" an SSA value. For
85 more information, please read up on <a
86 href="http://en.wikipedia.org/wiki/Static_single_assignment_form">Static Single
87 Assignment</a> - the concepts are really quite natural once you grok them.</p>
90 second thing we want is an "Error" exception like we used for the parser, which
91 will be used to report errors found during code generation (for example, use of
92 an undeclared parameter):</p>
94 <div class="doc_code">
96 exception Error of string
98 let context = global_context ()
99 let the_module = create_module context "my cool jit"
100 let builder = builder context
101 let named_values:(string, llvalue) Hashtbl.t = Hashtbl.create 10
102 let double_type = double_type context
106 <p>The static variables will be used during code generation.
107 <tt>Codgen.the_module</tt> is the LLVM construct that contains all of the
108 functions and global variables in a chunk of code. In many ways, it is the
109 top-level structure that the LLVM IR uses to contain code.</p>
111 <p>The <tt>Codegen.builder</tt> object is a helper object that makes it easy to
112 generate LLVM instructions. Instances of the <a
113 href="http://llvm.org/doxygen/IRBuilder_8h-source.html"><tt>IRBuilder</tt></a>
114 class keep track of the current place to insert instructions and has methods to
115 create new instructions.</p>
117 <p>The <tt>Codegen.named_values</tt> map keeps track of which values are defined
118 in the current scope and what their LLVM representation is. (In other words, it
119 is a symbol table for the code). In this form of Kaleidoscope, the only things
120 that can be referenced are function parameters. As such, function parameters
121 will be in this map when generating code for their function body.</p>
124 With these basics in place, we can start talking about how to generate code for
125 each expression. Note that this assumes that the <tt>Codgen.builder</tt> has
126 been set up to generate code <em>into</em> something. For now, we'll assume
127 that this has already been done, and we'll just use it to emit code.</p>
131 <!-- *********************************************************************** -->
132 <h2><a name="exprs">Expression Code Generation</a></h2>
133 <!-- *********************************************************************** -->
137 <p>Generating LLVM code for expression nodes is very straightforward: less
138 than 30 lines of commented code for all four of our expression nodes. First
139 we'll do numeric literals:</p>
141 <div class="doc_code">
143 | Ast.Number n -> const_float double_type n
147 <p>In the LLVM IR, numeric constants are represented with the
148 <tt>ConstantFP</tt> class, which holds the numeric value in an <tt>APFloat</tt>
149 internally (<tt>APFloat</tt> has the capability of holding floating point
150 constants of <em>A</em>rbitrary <em>P</em>recision). This code basically just
151 creates and returns a <tt>ConstantFP</tt>. Note that in the LLVM IR
152 that constants are all uniqued together and shared. For this reason, the API
153 uses "the foo::get(..)" idiom instead of "new foo(..)" or "foo::Create(..)".</p>
155 <div class="doc_code">
157 | Ast.Variable name ->
158 (try Hashtbl.find named_values name with
159 | Not_found -> raise (Error "unknown variable name"))
163 <p>References to variables are also quite simple using LLVM. In the simple
164 version of Kaleidoscope, we assume that the variable has already been emitted
165 somewhere and its value is available. In practice, the only values that can be
166 in the <tt>Codegen.named_values</tt> map are function arguments. This code
167 simply checks to see that the specified name is in the map (if not, an unknown
168 variable is being referenced) and returns the value for it. In future chapters,
169 we'll add support for <a href="LangImpl5.html#for">loop induction variables</a>
170 in the symbol table, and for <a href="LangImpl7.html#localvars">local
173 <div class="doc_code">
175 | Ast.Binary (op, lhs, rhs) ->
176 let lhs_val = codegen_expr lhs in
177 let rhs_val = codegen_expr rhs in
180 | '+' -> build_fadd lhs_val rhs_val "addtmp" builder
181 | '-' -> build_fsub lhs_val rhs_val "subtmp" builder
182 | '*' -> build_fmul lhs_val rhs_val "multmp" builder
184 (* Convert bool 0/1 to double 0.0 or 1.0 *)
185 let i = build_fcmp Fcmp.Ult lhs_val rhs_val "cmptmp" builder in
186 build_uitofp i double_type "booltmp" builder
187 | _ -> raise (Error "invalid binary operator")
192 <p>Binary operators start to get more interesting. The basic idea here is that
193 we recursively emit code for the left-hand side of the expression, then the
194 right-hand side, then we compute the result of the binary expression. In this
195 code, we do a simple switch on the opcode to create the right LLVM instruction.
198 <p>In the example above, the LLVM builder class is starting to show its value.
199 IRBuilder knows where to insert the newly created instruction, all you have to
200 do is specify what instruction to create (e.g. with <tt>Llvm.create_add</tt>),
201 which operands to use (<tt>lhs</tt> and <tt>rhs</tt> here) and optionally
202 provide a name for the generated instruction.</p>
204 <p>One nice thing about LLVM is that the name is just a hint. For instance, if
205 the code above emits multiple "addtmp" variables, LLVM will automatically
206 provide each one with an increasing, unique numeric suffix. Local value names
207 for instructions are purely optional, but it makes it much easier to read the
210 <p><a href="../LangRef.html#instref">LLVM instructions</a> are constrained by
211 strict rules: for example, the Left and Right operators of
212 an <a href="../LangRef.html#i_add">add instruction</a> must have the same
213 type, and the result type of the add must match the operand types. Because
214 all values in Kaleidoscope are doubles, this makes for very simple code for add,
217 <p>On the other hand, LLVM specifies that the <a
218 href="../LangRef.html#i_fcmp">fcmp instruction</a> always returns an 'i1' value
219 (a one bit integer). The problem with this is that Kaleidoscope wants the value to be a 0.0 or 1.0 value. In order to get these semantics, we combine the fcmp instruction with
220 a <a href="../LangRef.html#i_uitofp">uitofp instruction</a>. This instruction
221 converts its input integer into a floating point value by treating the input
222 as an unsigned value. In contrast, if we used the <a
223 href="../LangRef.html#i_sitofp">sitofp instruction</a>, the Kaleidoscope '<'
224 operator would return 0.0 and -1.0, depending on the input value.</p>
226 <div class="doc_code">
228 | Ast.Call (callee, args) ->
229 (* Look up the name in the module table. *)
231 match lookup_function callee the_module with
232 | Some callee -> callee
233 | None -> raise (Error "unknown function referenced")
235 let params = params callee in
237 (* If argument mismatch error. *)
238 if Array.length params == Array.length args then () else
239 raise (Error "incorrect # arguments passed");
240 let args = Array.map codegen_expr args in
241 build_call callee args "calltmp" builder
245 <p>Code generation for function calls is quite straightforward with LLVM. The
246 code above initially does a function name lookup in the LLVM Module's symbol
247 table. Recall that the LLVM Module is the container that holds all of the
248 functions we are JIT'ing. By giving each function the same name as what the
249 user specifies, we can use the LLVM symbol table to resolve function names for
252 <p>Once we have the function to call, we recursively codegen each argument that
253 is to be passed in, and create an LLVM <a href="../LangRef.html#i_call">call
254 instruction</a>. Note that LLVM uses the native C calling conventions by
255 default, allowing these calls to also call into standard library functions like
256 "sin" and "cos", with no additional effort.</p>
258 <p>This wraps up our handling of the four basic expressions that we have so far
259 in Kaleidoscope. Feel free to go in and add some more. For example, by
260 browsing the <a href="../LangRef.html">LLVM language reference</a> you'll find
261 several other interesting instructions that are really easy to plug into our
266 <!-- *********************************************************************** -->
267 <h2><a name="funcs">Function Code Generation</a></h2>
268 <!-- *********************************************************************** -->
272 <p>Code generation for prototypes and functions must handle a number of
273 details, which make their code less beautiful than expression code
274 generation, but allows us to illustrate some important points. First, lets
275 talk about code generation for prototypes: they are used both for function
276 bodies and external function declarations. The code starts with:</p>
278 <div class="doc_code">
280 let codegen_proto = function
281 | Ast.Prototype (name, args) ->
282 (* Make the function type: double(double,double) etc. *)
283 let doubles = Array.make (Array.length args) double_type in
284 let ft = function_type double_type doubles in
286 match lookup_function name the_module with
290 <p>This code packs a lot of power into a few lines. Note first that this
291 function returns a "Function*" instead of a "Value*" (although at the moment
292 they both are modeled by <tt>llvalue</tt> in ocaml). Because a "prototype"
293 really talks about the external interface for a function (not the value computed
294 by an expression), it makes sense for it to return the LLVM Function it
295 corresponds to when codegen'd.</p>
297 <p>The call to <tt>Llvm.function_type</tt> creates the <tt>Llvm.llvalue</tt>
298 that should be used for a given Prototype. Since all function arguments in
299 Kaleidoscope are of type double, the first line creates a vector of "N" LLVM
300 double types. It then uses the <tt>Llvm.function_type</tt> method to create a
301 function type that takes "N" doubles as arguments, returns one double as a
302 result, and that is not vararg (that uses the function
303 <tt>Llvm.var_arg_function_type</tt>). Note that Types in LLVM are uniqued just
304 like <tt>Constant</tt>s are, so you don't "new" a type, you "get" it.</p>
306 <p>The final line above checks if the function has already been defined in
307 <tt>Codegen.the_module</tt>. If not, we will create it.</p>
309 <div class="doc_code">
311 | None -> declare_function name ft the_module
315 <p>This indicates the type and name to use, as well as which module to insert
316 into. By default we assume a function has
317 <tt>Llvm.Linkage.ExternalLinkage</tt>. "<a href="LangRef.html#linkage">external
318 linkage</a>" means that the function may be defined outside the current module
319 and/or that it is callable by functions outside the module. The "<tt>name</tt>"
320 passed in is the name the user specified: this name is registered in
321 "<tt>Codegen.the_module</tt>"s symbol table, which is used by the function call
324 <p>In Kaleidoscope, I choose to allow redefinitions of functions in two cases:
325 first, we want to allow 'extern'ing a function more than once, as long as the
326 prototypes for the externs match (since all arguments have the same type, we
327 just have to check that the number of arguments match). Second, we want to
328 allow 'extern'ing a function and then defining a body for it. This is useful
329 when defining mutually recursive functions.</p>
331 <div class="doc_code">
333 (* If 'f' conflicted, there was already something named 'name'. If it
334 * has a body, don't allow redefinition or reextern. *)
336 (* If 'f' already has a body, reject this. *)
337 if Array.length (basic_blocks f) == 0 then () else
338 raise (Error "redefinition of function");
340 (* If 'f' took a different number of arguments, reject. *)
341 if Array.length (params f) == Array.length args then () else
342 raise (Error "redefinition of function with different # args");
348 <p>In order to verify the logic above, we first check to see if the pre-existing
349 function is "empty". In this case, empty means that it has no basic blocks in
350 it, which means it has no body. If it has no body, it is a forward
351 declaration. Since we don't allow anything after a full definition of the
352 function, the code rejects this case. If the previous reference to a function
353 was an 'extern', we simply verify that the number of arguments for that
354 definition and this one match up. If not, we emit an error.</p>
356 <div class="doc_code">
358 (* Set names for all arguments. *)
359 Array.iteri (fun i a ->
362 Hashtbl.add named_values n a;
368 <p>The last bit of code for prototypes loops over all of the arguments in the
369 function, setting the name of the LLVM Argument objects to match, and registering
370 the arguments in the <tt>Codegen.named_values</tt> map for future use by the
371 <tt>Ast.Variable</tt> variant. Once this is set up, it returns the Function
372 object to the caller. Note that we don't check for conflicting
373 argument names here (e.g. "extern foo(a b a)"). Doing so would be very
374 straight-forward with the mechanics we have already used above.</p>
376 <div class="doc_code">
378 let codegen_func = function
379 | Ast.Function (proto, body) ->
380 Hashtbl.clear named_values;
381 let the_function = codegen_proto proto in
385 <p>Code generation for function definitions starts out simply enough: we just
386 codegen the prototype (Proto) and verify that it is ok. We then clear out the
387 <tt>Codegen.named_values</tt> map to make sure that there isn't anything in it
388 from the last function we compiled. Code generation of the prototype ensures
389 that there is an LLVM Function object that is ready to go for us.</p>
391 <div class="doc_code">
393 (* Create a new basic block to start insertion into. *)
394 let bb = append_block context "entry" the_function in
395 position_at_end bb builder;
398 let ret_val = codegen_expr body in
402 <p>Now we get to the point where the <tt>Codegen.builder</tt> is set up. The
403 first line creates a new
404 <a href="http://en.wikipedia.org/wiki/Basic_block">basic block</a> (named
405 "entry"), which is inserted into <tt>the_function</tt>. The second line then
406 tells the builder that new instructions should be inserted into the end of the
407 new basic block. Basic blocks in LLVM are an important part of functions that
409 href="http://en.wikipedia.org/wiki/Control_flow_graph">Control Flow Graph</a>.
410 Since we don't have any control flow, our functions will only contain one
411 block at this point. We'll fix this in <a href="OCamlLangImpl5.html">Chapter
414 <div class="doc_code">
416 let ret_val = codegen_expr body in
418 (* Finish off the function. *)
419 let _ = build_ret ret_val builder in
421 (* Validate the generated code, checking for consistency. *)
422 Llvm_analysis.assert_valid_function the_function;
428 <p>Once the insertion point is set up, we call the <tt>Codegen.codegen_func</tt>
429 method for the root expression of the function. If no error happens, this emits
430 code to compute the expression into the entry block and returns the value that
431 was computed. Assuming no error, we then create an LLVM <a
432 href="../LangRef.html#i_ret">ret instruction</a>, which completes the function.
433 Once the function is built, we call
434 <tt>Llvm_analysis.assert_valid_function</tt>, which is provided by LLVM. This
435 function does a variety of consistency checks on the generated code, to
436 determine if our compiler is doing everything right. Using this is important:
437 it can catch a lot of bugs. Once the function is finished and validated, we
440 <div class="doc_code">
443 delete_function the_function;
448 <p>The only piece left here is handling of the error case. For simplicity, we
449 handle this by merely deleting the function we produced with the
450 <tt>Llvm.delete_function</tt> method. This allows the user to redefine a
451 function that they incorrectly typed in before: if we didn't delete it, it
452 would live in the symbol table, with a body, preventing future redefinition.</p>
454 <p>This code does have a bug, though. Since the <tt>Codegen.codegen_proto</tt>
455 can return a previously defined forward declaration, our code can actually delete
456 a forward declaration. There are a number of ways to fix this bug, see what you
457 can come up with! Here is a testcase:</p>
459 <div class="doc_code">
461 extern foo(a b); # ok, defines foo.
462 def foo(a b) c; # error, 'c' is invalid.
463 def bar() foo(1, 2); # error, unknown function "foo"
469 <!-- *********************************************************************** -->
470 <h2><a name="driver">Driver Changes and Closing Thoughts</a></h2>
471 <!-- *********************************************************************** -->
476 For now, code generation to LLVM doesn't really get us much, except that we can
477 look at the pretty IR calls. The sample code inserts calls to Codegen into the
478 "<tt>Toplevel.main_loop</tt>", and then dumps out the LLVM IR. This gives a
479 nice way to look at the LLVM IR for simple functions. For example:
482 <div class="doc_code">
484 ready> <b>4+5</b>;
485 Read top-level expression:
486 define double @""() {
488 %addtmp = fadd double 4.000000e+00, 5.000000e+00
494 <p>Note how the parser turns the top-level expression into anonymous functions
495 for us. This will be handy when we add <a href="OCamlLangImpl4.html#jit">JIT
496 support</a> in the next chapter. Also note that the code is very literally
497 transcribed, no optimizations are being performed. We will
498 <a href="OCamlLangImpl4.html#trivialconstfold">add optimizations</a> explicitly
499 in the next chapter.</p>
501 <div class="doc_code">
503 ready> <b>def foo(a b) a*a + 2*a*b + b*b;</b>
504 Read function definition:
505 define double @foo(double %a, double %b) {
507 %multmp = fmul double %a, %a
508 %multmp1 = fmul double 2.000000e+00, %a
509 %multmp2 = fmul double %multmp1, %b
510 %addtmp = fadd double %multmp, %multmp2
511 %multmp3 = fmul double %b, %b
512 %addtmp4 = fadd double %addtmp, %multmp3
518 <p>This shows some simple arithmetic. Notice the striking similarity to the
519 LLVM builder calls that we use to create the instructions.</p>
521 <div class="doc_code">
523 ready> <b>def bar(a) foo(a, 4.0) + bar(31337);</b>
524 Read function definition:
525 define double @bar(double %a) {
527 %calltmp = call double @foo(double %a, double 4.000000e+00)
528 %calltmp1 = call double @bar(double 3.133700e+04)
529 %addtmp = fadd double %calltmp, %calltmp1
535 <p>This shows some function calls. Note that this function will take a long
536 time to execute if you call it. In the future we'll add conditional control
537 flow to actually make recursion useful :).</p>
539 <div class="doc_code">
541 ready> <b>extern cos(x);</b>
543 declare double @cos(double)
545 ready> <b>cos(1.234);</b>
546 Read top-level expression:
547 define double @""() {
549 %calltmp = call double @cos(double 1.234000e+00)
555 <p>This shows an extern for the libm "cos" function, and a call to it.</p>
558 <div class="doc_code">
561 ; ModuleID = 'my cool jit'
563 define double @""() {
565 %addtmp = fadd double 4.000000e+00, 5.000000e+00
569 define double @foo(double %a, double %b) {
571 %multmp = fmul double %a, %a
572 %multmp1 = fmul double 2.000000e+00, %a
573 %multmp2 = fmul double %multmp1, %b
574 %addtmp = fadd double %multmp, %multmp2
575 %multmp3 = fmul double %b, %b
576 %addtmp4 = fadd double %addtmp, %multmp3
580 define double @bar(double %a) {
582 %calltmp = call double @foo(double %a, double 4.000000e+00)
583 %calltmp1 = call double @bar(double 3.133700e+04)
584 %addtmp = fadd double %calltmp, %calltmp1
588 declare double @cos(double)
590 define double @""() {
592 %calltmp = call double @cos(double 1.234000e+00)
598 <p>When you quit the current demo, it dumps out the IR for the entire module
599 generated. Here you can see the big picture with all the functions referencing
602 <p>This wraps up the third chapter of the Kaleidoscope tutorial. Up next, we'll
603 describe how to <a href="OCamlLangImpl4.html">add JIT codegen and optimizer
604 support</a> to this so we can actually start running code!</p>
609 <!-- *********************************************************************** -->
610 <h2><a name="code">Full Code Listing</a></h2>
611 <!-- *********************************************************************** -->
616 Here is the complete code listing for our running example, enhanced with the
617 LLVM code generator. Because this uses the LLVM libraries, we need to link
618 them in. To do this, we use the <a
619 href="http://llvm.org/cmds/llvm-config.html">llvm-config</a> tool to inform
620 our makefile/command line about which options to use:</p>
622 <div class="doc_code">
631 <p>Here is the code:</p>
635 <dd class="doc_code">
637 <{lexer,parser}.ml>: use_camlp4, pp(camlp4of)
638 <*.{byte,native}>: g++, use_llvm, use_llvm_analysis
642 <dt>myocamlbuild.ml:</dt>
643 <dd class="doc_code">
645 open Ocamlbuild_plugin;;
647 ocaml_lib ~extern:true "llvm";;
648 ocaml_lib ~extern:true "llvm_analysis";;
650 flag ["link"; "ocaml"; "g++"] (S[A"-cc"; A"g++"]);;
655 <dd class="doc_code">
657 (*===----------------------------------------------------------------------===
659 *===----------------------------------------------------------------------===*)
661 (* The lexer returns these 'Kwd' if it is an unknown character, otherwise one of
662 * these others for known things. *)
668 | Ident of string | Number of float
676 <dd class="doc_code">
678 (*===----------------------------------------------------------------------===
680 *===----------------------------------------------------------------------===*)
683 (* Skip any whitespace. *)
684 | [< ' (' ' | '\n' | '\r' | '\t'); stream >] -> lex stream
686 (* identifier: [a-zA-Z][a-zA-Z0-9] *)
687 | [< ' ('A' .. 'Z' | 'a' .. 'z' as c); stream >] ->
688 let buffer = Buffer.create 1 in
689 Buffer.add_char buffer c;
690 lex_ident buffer stream
692 (* number: [0-9.]+ *)
693 | [< ' ('0' .. '9' as c); stream >] ->
694 let buffer = Buffer.create 1 in
695 Buffer.add_char buffer c;
696 lex_number buffer stream
698 (* Comment until end of line. *)
699 | [< ' ('#'); stream >] ->
702 (* Otherwise, just return the character as its ascii value. *)
703 | [< 'c; stream >] ->
704 [< 'Token.Kwd c; lex stream >]
707 | [< >] -> [< >]
709 and lex_number buffer = parser
710 | [< ' ('0' .. '9' | '.' as c); stream >] ->
711 Buffer.add_char buffer c;
712 lex_number buffer stream
713 | [< stream=lex >] ->
714 [< 'Token.Number (float_of_string (Buffer.contents buffer)); stream >]
716 and lex_ident buffer = parser
717 | [< ' ('A' .. 'Z' | 'a' .. 'z' | '0' .. '9' as c); stream >] ->
718 Buffer.add_char buffer c;
719 lex_ident buffer stream
720 | [< stream=lex >] ->
721 match Buffer.contents buffer with
722 | "def" -> [< 'Token.Def; stream >]
723 | "extern" -> [< 'Token.Extern; stream >]
724 | id -> [< 'Token.Ident id; stream >]
726 and lex_comment = parser
727 | [< ' ('\n'); stream=lex >] -> stream
728 | [< 'c; e=lex_comment >] -> e
729 | [< >] -> [< >]
734 <dd class="doc_code">
736 (*===----------------------------------------------------------------------===
737 * Abstract Syntax Tree (aka Parse Tree)
738 *===----------------------------------------------------------------------===*)
740 (* expr - Base type for all expression nodes. *)
742 (* variant for numeric literals like "1.0". *)
745 (* variant for referencing a variable, like "a". *)
748 (* variant for a binary operator. *)
749 | Binary of char * expr * expr
751 (* variant for function calls. *)
752 | Call of string * expr array
754 (* proto - This type represents the "prototype" for a function, which captures
755 * its name, and its argument names (thus implicitly the number of arguments the
756 * function takes). *)
757 type proto = Prototype of string * string array
759 (* func - This type represents a function definition itself. *)
760 type func = Function of proto * expr
765 <dd class="doc_code">
767 (*===---------------------------------------------------------------------===
769 *===---------------------------------------------------------------------===*)
771 (* binop_precedence - This holds the precedence for each binary operator that is
773 let binop_precedence:(char, int) Hashtbl.t = Hashtbl.create 10
775 (* precedence - Get the precedence of the pending binary operator token. *)
776 let precedence c = try Hashtbl.find binop_precedence c with Not_found -> -1
782 let rec parse_primary = parser
783 (* numberexpr ::= number *)
784 | [< 'Token.Number n >] -> Ast.Number n
786 (* parenexpr ::= '(' expression ')' *)
787 | [< 'Token.Kwd '('; e=parse_expr; 'Token.Kwd ')' ?? "expected ')'" >] -> e
791 * ::= identifier '(' argumentexpr ')' *)
792 | [< 'Token.Ident id; stream >] ->
793 let rec parse_args accumulator = parser
794 | [< e=parse_expr; stream >] ->
796 | [< 'Token.Kwd ','; e=parse_args (e :: accumulator) >] -> e
797 | [< >] -> e :: accumulator
799 | [< >] -> accumulator
801 let rec parse_ident id = parser
803 | [< 'Token.Kwd '(';
805 'Token.Kwd ')' ?? "expected ')'">] ->
806 Ast.Call (id, Array.of_list (List.rev args))
808 (* Simple variable ref. *)
809 | [< >] -> Ast.Variable id
811 parse_ident id stream
813 | [< >] -> raise (Stream.Error "unknown token when expecting an expression.")
816 * ::= ('+' primary)* *)
817 and parse_bin_rhs expr_prec lhs stream =
818 match Stream.peek stream with
819 (* If this is a binop, find its precedence. *)
820 | Some (Token.Kwd c) when Hashtbl.mem binop_precedence c ->
821 let token_prec = precedence c in
823 (* If this is a binop that binds at least as tightly as the current binop,
824 * consume it, otherwise we are done. *)
825 if token_prec < expr_prec then lhs else begin
829 (* Parse the primary expression after the binary operator. *)
830 let rhs = parse_primary stream in
832 (* Okay, we know this is a binop. *)
834 match Stream.peek stream with
835 | Some (Token.Kwd c2) ->
836 (* If BinOp binds less tightly with rhs than the operator after
837 * rhs, let the pending operator take rhs as its lhs. *)
838 let next_prec = precedence c2 in
839 if token_prec < next_prec
840 then parse_bin_rhs (token_prec + 1) rhs stream
846 let lhs = Ast.Binary (c, lhs, rhs) in
847 parse_bin_rhs expr_prec lhs stream
852 * ::= primary binoprhs *)
853 and parse_expr = parser
854 | [< lhs=parse_primary; stream >] -> parse_bin_rhs 0 lhs stream
857 * ::= id '(' id* ')' *)
858 let parse_prototype =
859 let rec parse_args accumulator = parser
860 | [< 'Token.Ident id; e=parse_args (id::accumulator) >] -> e
861 | [< >] -> accumulator
865 | [< 'Token.Ident id;
866 'Token.Kwd '(' ?? "expected '(' in prototype";
868 'Token.Kwd ')' ?? "expected ')' in prototype" >] ->
870 Ast.Prototype (id, Array.of_list (List.rev args))
873 raise (Stream.Error "expected function name in prototype")
875 (* definition ::= 'def' prototype expression *)
876 let parse_definition = parser
877 | [< 'Token.Def; p=parse_prototype; e=parse_expr >] ->
880 (* toplevelexpr ::= expression *)
881 let parse_toplevel = parser
882 | [< e=parse_expr >] ->
883 (* Make an anonymous proto. *)
884 Ast.Function (Ast.Prototype ("", [||]), e)
886 (* external ::= 'extern' prototype *)
887 let parse_extern = parser
888 | [< 'Token.Extern; e=parse_prototype >] -> e
893 <dd class="doc_code">
895 (*===----------------------------------------------------------------------===
897 *===----------------------------------------------------------------------===*)
901 exception Error of string
903 let context = global_context ()
904 let the_module = create_module context "my cool jit"
905 let builder = builder context
906 let named_values:(string, llvalue) Hashtbl.t = Hashtbl.create 10
907 let double_type = double_type context
909 let rec codegen_expr = function
910 | Ast.Number n -> const_float double_type n
911 | Ast.Variable name ->
912 (try Hashtbl.find named_values name with
913 | Not_found -> raise (Error "unknown variable name"))
914 | Ast.Binary (op, lhs, rhs) ->
915 let lhs_val = codegen_expr lhs in
916 let rhs_val = codegen_expr rhs in
919 | '+' -> build_add lhs_val rhs_val "addtmp" builder
920 | '-' -> build_sub lhs_val rhs_val "subtmp" builder
921 | '*' -> build_mul lhs_val rhs_val "multmp" builder
923 (* Convert bool 0/1 to double 0.0 or 1.0 *)
924 let i = build_fcmp Fcmp.Ult lhs_val rhs_val "cmptmp" builder in
925 build_uitofp i double_type "booltmp" builder
926 | _ -> raise (Error "invalid binary operator")
928 | Ast.Call (callee, args) ->
929 (* Look up the name in the module table. *)
931 match lookup_function callee the_module with
932 | Some callee -> callee
933 | None -> raise (Error "unknown function referenced")
935 let params = params callee in
937 (* If argument mismatch error. *)
938 if Array.length params == Array.length args then () else
939 raise (Error "incorrect # arguments passed");
940 let args = Array.map codegen_expr args in
941 build_call callee args "calltmp" builder
943 let codegen_proto = function
944 | Ast.Prototype (name, args) ->
945 (* Make the function type: double(double,double) etc. *)
946 let doubles = Array.make (Array.length args) double_type in
947 let ft = function_type double_type doubles in
949 match lookup_function name the_module with
950 | None -> declare_function name ft the_module
952 (* If 'f' conflicted, there was already something named 'name'. If it
953 * has a body, don't allow redefinition or reextern. *)
955 (* If 'f' already has a body, reject this. *)
956 if block_begin f <> At_end f then
957 raise (Error "redefinition of function");
959 (* If 'f' took a different number of arguments, reject. *)
960 if element_type (type_of f) <> ft then
961 raise (Error "redefinition of function with different # args");
965 (* Set names for all arguments. *)
966 Array.iteri (fun i a ->
969 Hashtbl.add named_values n a;
973 let codegen_func = function
974 | Ast.Function (proto, body) ->
975 Hashtbl.clear named_values;
976 let the_function = codegen_proto proto in
978 (* Create a new basic block to start insertion into. *)
979 let bb = append_block context "entry" the_function in
980 position_at_end bb builder;
983 let ret_val = codegen_expr body in
985 (* Finish off the function. *)
986 let _ = build_ret ret_val builder in
988 (* Validate the generated code, checking for consistency. *)
989 Llvm_analysis.assert_valid_function the_function;
993 delete_function the_function;
998 <dt>toplevel.ml:</dt>
999 <dd class="doc_code">
1001 (*===----------------------------------------------------------------------===
1002 * Top-Level parsing and JIT Driver
1003 *===----------------------------------------------------------------------===*)
1007 (* top ::= definition | external | expression | ';' *)
1008 let rec main_loop stream =
1009 match Stream.peek stream with
1012 (* ignore top-level semicolons. *)
1013 | Some (Token.Kwd ';') ->
1019 try match token with
1021 let e = Parser.parse_definition stream in
1022 print_endline "parsed a function definition.";
1023 dump_value (Codegen.codegen_func e);
1024 | Token.Extern ->
1025 let e = Parser.parse_extern stream in
1026 print_endline "parsed an extern.";
1027 dump_value (Codegen.codegen_proto e);
1029 (* Evaluate a top-level expression into an anonymous function. *)
1030 let e = Parser.parse_toplevel stream in
1031 print_endline "parsed a top-level expr";
1032 dump_value (Codegen.codegen_func e);
1033 with Stream.Error s | Codegen.Error s ->
1034 (* Skip token for error recovery. *)
1038 print_string "ready> "; flush stdout;
1044 <dd class="doc_code">
1046 (*===----------------------------------------------------------------------===
1048 *===----------------------------------------------------------------------===*)
1053 (* Install standard binary operators.
1054 * 1 is the lowest precedence. *)
1055 Hashtbl.add Parser.binop_precedence '<' 10;
1056 Hashtbl.add Parser.binop_precedence '+' 20;
1057 Hashtbl.add Parser.binop_precedence '-' 20;
1058 Hashtbl.add Parser.binop_precedence '*' 40; (* highest. *)
1060 (* Prime the first token. *)
1061 print_string "ready> "; flush stdout;
1062 let stream = Lexer.lex (Stream.of_channel stdin) in
1064 (* Run the main "interpreter loop" now. *)
1065 Toplevel.main_loop stream;
1067 (* Print out all the generated code. *)
1068 dump_module Codegen.the_module
1076 <a href="OCamlLangImpl4.html">Next: Adding JIT and Optimizer Support</a>
1079 <!-- *********************************************************************** -->
1082 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
1083 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
1084 <a href="http://validator.w3.org/check/referer"><img
1085 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
1087 <a href="mailto:sabre@nondot.org">Chris Lattner</a><br>
1088 <a href="mailto:idadesub@users.sourceforge.net">Erick Tryzelaar</a><br>
1089 <a href="http://llvm.org/">The LLVM Compiler Infrastructure</a><br>
1090 Last modified: $Date$