blob: 921fa8908040f96fb6980a70701f155baadef2fd [file] [log] [blame]
Eugene Zelenkof981ec42016-05-19 01:08:04 +00001#include "llvm/ADT/APFloat.h"
Lang Hames09bf4c12015-08-18 18:11:06 +00002#include "llvm/ADT/STLExtras.h"
Eugene Zelenkof981ec42016-05-19 01:08:04 +00003#include "llvm/IR/BasicBlock.h"
4#include "llvm/IR/Constants.h"
5#include "llvm/IR/DerivedTypes.h"
6#include "llvm/IR/Function.h"
Chandler Carruth005f27a2013-01-02 11:56:33 +00007#include "llvm/IR/IRBuilder.h"
8#include "llvm/IR/LLVMContext.h"
Chandler Carruth30d69c22015-02-13 10:01:29 +00009#include "llvm/IR/LegacyPassManager.h"
Chandler Carruth005f27a2013-01-02 11:56:33 +000010#include "llvm/IR/Module.h"
Eugene Zelenkof981ec42016-05-19 01:08:04 +000011#include "llvm/IR/Type.h"
Chandler Carruth20d4e6b2014-01-13 09:58:03 +000012#include "llvm/IR/Verifier.h"
Evan Cheng2bb40352011-08-24 18:08:43 +000013#include "llvm/Support/TargetSelect.h"
Eugene Zelenkof981ec42016-05-19 01:08:04 +000014#include "llvm/Target/TargetMachine.h"
Chandler Carruth605e30e2012-12-04 10:16:57 +000015#include "llvm/Transforms/Scalar.h"
Chandler Carruthec5872b2016-03-11 12:10:15 +000016#include "llvm/Transforms/Scalar/GVN.h"
Eugene Zelenkof981ec42016-05-19 01:08:04 +000017#include "../include/KaleidoscopeJIT.h"
Eugene Zelenkoae7ac952016-11-18 21:57:58 +000018#include <algorithm>
Eugene Zelenkof981ec42016-05-19 01:08:04 +000019#include <cassert>
Will Dietz981af002013-10-12 00:55:57 +000020#include <cctype>
Eugene Zelenkof981ec42016-05-19 01:08:04 +000021#include <cstdint>
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +000022#include <cstdio>
Eugene Zelenkof981ec42016-05-19 01:08:04 +000023#include <cstdlib>
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +000024#include <map>
Eugene Zelenkof981ec42016-05-19 01:08:04 +000025#include <memory>
Chandler Carruth605e30e2012-12-04 10:16:57 +000026#include <string>
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +000027#include <vector>
Lang Hames2d789c32015-08-26 03:07:41 +000028
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +000029using namespace llvm;
Lang Hames2d789c32015-08-26 03:07:41 +000030using namespace llvm::orc;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +000031
32//===----------------------------------------------------------------------===//
33// Lexer
34//===----------------------------------------------------------------------===//
35
36// The lexer returns tokens [0-255] if it is an unknown character, otherwise one
37// of these for known things.
38enum Token {
39 tok_eof = -1,
40
41 // commands
Eric Christopherc0239362014-12-08 18:12:28 +000042 tok_def = -2,
43 tok_extern = -3,
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +000044
45 // primary
Eric Christopherc0239362014-12-08 18:12:28 +000046 tok_identifier = -4,
47 tok_number = -5
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +000048};
49
Eric Christopherc0239362014-12-08 18:12:28 +000050static std::string IdentifierStr; // Filled in if tok_identifier
51static double NumVal; // Filled in if tok_number
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +000052
53/// gettok - Return the next token from standard input.
54static int gettok() {
55 static int LastChar = ' ';
56
57 // Skip any whitespace.
58 while (isspace(LastChar))
59 LastChar = getchar();
60
61 if (isalpha(LastChar)) { // identifier: [a-zA-Z][a-zA-Z0-9]*
62 IdentifierStr = LastChar;
63 while (isalnum((LastChar = getchar())))
64 IdentifierStr += LastChar;
65
Eric Christopherc0239362014-12-08 18:12:28 +000066 if (IdentifierStr == "def")
67 return tok_def;
68 if (IdentifierStr == "extern")
69 return tok_extern;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +000070 return tok_identifier;
71 }
72
Eric Christopherc0239362014-12-08 18:12:28 +000073 if (isdigit(LastChar) || LastChar == '.') { // Number: [0-9.]+
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +000074 std::string NumStr;
75 do {
76 NumStr += LastChar;
77 LastChar = getchar();
78 } while (isdigit(LastChar) || LastChar == '.');
79
Hans Wennborgcc9deb42015-09-29 18:02:48 +000080 NumVal = strtod(NumStr.c_str(), nullptr);
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +000081 return tok_number;
82 }
83
84 if (LastChar == '#') {
85 // Comment until end of line.
Eric Christopherc0239362014-12-08 18:12:28 +000086 do
87 LastChar = getchar();
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +000088 while (LastChar != EOF && LastChar != '\n' && LastChar != '\r');
Eric Christopherc0239362014-12-08 18:12:28 +000089
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +000090 if (LastChar != EOF)
91 return gettok();
92 }
Eric Christopherc0239362014-12-08 18:12:28 +000093
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +000094 // Check for end of file. Don't eat the EOF.
95 if (LastChar == EOF)
96 return tok_eof;
97
98 // Otherwise, just return the character as its ascii value.
99 int ThisChar = LastChar;
100 LastChar = getchar();
101 return ThisChar;
102}
103
104//===----------------------------------------------------------------------===//
105// Abstract Syntax Tree (aka Parse Tree)
106//===----------------------------------------------------------------------===//
Eugene Zelenkoae7ac952016-11-18 21:57:58 +0000107
Juergen Ributzka05c5a932013-11-19 03:08:35 +0000108namespace {
Eugene Zelenkoae7ac952016-11-18 21:57:58 +0000109
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000110/// ExprAST - Base class for all expression nodes.
111class ExprAST {
112public:
Eugene Zelenkoae7ac952016-11-18 21:57:58 +0000113 virtual ~ExprAST() = default;
114
Lang Hames2d789c32015-08-26 03:07:41 +0000115 virtual Value *codegen() = 0;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000116};
117
118/// NumberExprAST - Expression class for numeric literals like "1.0".
119class NumberExprAST : public ExprAST {
120 double Val;
Lang Hames59b0da82015-08-19 18:15:58 +0000121
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000122public:
Lang Hames09bf4c12015-08-18 18:11:06 +0000123 NumberExprAST(double Val) : Val(Val) {}
Eugene Zelenkoae7ac952016-11-18 21:57:58 +0000124
Lang Hames2d789c32015-08-26 03:07:41 +0000125 Value *codegen() override;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000126};
127
128/// VariableExprAST - Expression class for referencing a variable, like "a".
129class VariableExprAST : public ExprAST {
130 std::string Name;
Lang Hames59b0da82015-08-19 18:15:58 +0000131
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000132public:
Lang Hames09bf4c12015-08-18 18:11:06 +0000133 VariableExprAST(const std::string &Name) : Name(Name) {}
Eugene Zelenkoae7ac952016-11-18 21:57:58 +0000134
Lang Hames2d789c32015-08-26 03:07:41 +0000135 Value *codegen() override;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000136};
137
138/// BinaryExprAST - Expression class for a binary operator.
139class BinaryExprAST : public ExprAST {
140 char Op;
Lang Hames09bf4c12015-08-18 18:11:06 +0000141 std::unique_ptr<ExprAST> LHS, RHS;
Lang Hames59b0da82015-08-19 18:15:58 +0000142
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000143public:
Lang Hames09bf4c12015-08-18 18:11:06 +0000144 BinaryExprAST(char Op, std::unique_ptr<ExprAST> LHS,
145 std::unique_ptr<ExprAST> RHS)
146 : Op(Op), LHS(std::move(LHS)), RHS(std::move(RHS)) {}
Eugene Zelenkoae7ac952016-11-18 21:57:58 +0000147
Lang Hames2d789c32015-08-26 03:07:41 +0000148 Value *codegen() override;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000149};
150
151/// CallExprAST - Expression class for function calls.
152class CallExprAST : public ExprAST {
153 std::string Callee;
Lang Hames09bf4c12015-08-18 18:11:06 +0000154 std::vector<std::unique_ptr<ExprAST>> Args;
Lang Hames59b0da82015-08-19 18:15:58 +0000155
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000156public:
Lang Hames09bf4c12015-08-18 18:11:06 +0000157 CallExprAST(const std::string &Callee,
158 std::vector<std::unique_ptr<ExprAST>> Args)
159 : Callee(Callee), Args(std::move(Args)) {}
Eugene Zelenkoae7ac952016-11-18 21:57:58 +0000160
Lang Hames2d789c32015-08-26 03:07:41 +0000161 Value *codegen() override;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000162};
163
164/// PrototypeAST - This class represents the "prototype" for a function,
165/// which captures its name, and its argument names (thus implicitly the number
166/// of arguments the function takes).
167class PrototypeAST {
168 std::string Name;
169 std::vector<std::string> Args;
Lang Hames59b0da82015-08-19 18:15:58 +0000170
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000171public:
Lang Hames59b0da82015-08-19 18:15:58 +0000172 PrototypeAST(const std::string &Name, std::vector<std::string> Args)
173 : Name(Name), Args(std::move(Args)) {}
Eugene Zelenkoae7ac952016-11-18 21:57:58 +0000174
Lang Hames2d789c32015-08-26 03:07:41 +0000175 Function *codegen();
176 const std::string &getName() const { return Name; }
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000177};
178
179/// FunctionAST - This class represents a function definition itself.
180class FunctionAST {
Lang Hames09bf4c12015-08-18 18:11:06 +0000181 std::unique_ptr<PrototypeAST> Proto;
182 std::unique_ptr<ExprAST> Body;
Lang Hames59b0da82015-08-19 18:15:58 +0000183
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000184public:
Lang Hames09bf4c12015-08-18 18:11:06 +0000185 FunctionAST(std::unique_ptr<PrototypeAST> Proto,
186 std::unique_ptr<ExprAST> Body)
Lang Hames59b0da82015-08-19 18:15:58 +0000187 : Proto(std::move(Proto)), Body(std::move(Body)) {}
Eugene Zelenkoae7ac952016-11-18 21:57:58 +0000188
Lang Hames2d789c32015-08-26 03:07:41 +0000189 Function *codegen();
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000190};
Eugene Zelenkoae7ac952016-11-18 21:57:58 +0000191
Juergen Ributzka05c5a932013-11-19 03:08:35 +0000192} // end anonymous namespace
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000193
194//===----------------------------------------------------------------------===//
195// Parser
196//===----------------------------------------------------------------------===//
197
198/// CurTok/getNextToken - Provide a simple token buffer. CurTok is the current
199/// token the parser is looking at. getNextToken reads another token from the
200/// lexer and updates CurTok with its results.
201static int CurTok;
Eric Christopherc0239362014-12-08 18:12:28 +0000202static int getNextToken() { return CurTok = gettok(); }
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000203
204/// BinopPrecedence - This holds the precedence for each binary operator that is
205/// defined.
206static std::map<char, int> BinopPrecedence;
207
208/// GetTokPrecedence - Get the precedence of the pending binary operator token.
209static int GetTokPrecedence() {
210 if (!isascii(CurTok))
211 return -1;
Eric Christopherc0239362014-12-08 18:12:28 +0000212
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000213 // Make sure it's a declared binop.
214 int TokPrec = BinopPrecedence[CurTok];
Eric Christopherc0239362014-12-08 18:12:28 +0000215 if (TokPrec <= 0)
216 return -1;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000217 return TokPrec;
218}
219
Lang Hames5d045a92016-03-25 17:41:26 +0000220/// LogError* - These are little helper functions for error handling.
221std::unique_ptr<ExprAST> LogError(const char *Str) {
Eric Christopherc0239362014-12-08 18:12:28 +0000222 fprintf(stderr, "Error: %s\n", Str);
Lang Hames09bf4c12015-08-18 18:11:06 +0000223 return nullptr;
Eric Christopherc0239362014-12-08 18:12:28 +0000224}
Hans Wennborgcc9deb42015-09-29 18:02:48 +0000225
Lang Hames5d045a92016-03-25 17:41:26 +0000226std::unique_ptr<PrototypeAST> LogErrorP(const char *Str) {
227 LogError(Str);
Lang Hames09bf4c12015-08-18 18:11:06 +0000228 return nullptr;
Eric Christopherc0239362014-12-08 18:12:28 +0000229}
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000230
Lang Hames09bf4c12015-08-18 18:11:06 +0000231static std::unique_ptr<ExprAST> ParseExpression();
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000232
Lang Hames59b0da82015-08-19 18:15:58 +0000233/// numberexpr ::= number
234static std::unique_ptr<ExprAST> ParseNumberExpr() {
235 auto Result = llvm::make_unique<NumberExprAST>(NumVal);
236 getNextToken(); // consume the number
237 return std::move(Result);
238}
239
240/// parenexpr ::= '(' expression ')'
241static std::unique_ptr<ExprAST> ParseParenExpr() {
242 getNextToken(); // eat (.
243 auto V = ParseExpression();
244 if (!V)
245 return nullptr;
246
247 if (CurTok != ')')
Lang Hames5d045a92016-03-25 17:41:26 +0000248 return LogError("expected ')'");
Lang Hames59b0da82015-08-19 18:15:58 +0000249 getNextToken(); // eat ).
250 return V;
251}
252
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000253/// identifierexpr
254/// ::= identifier
255/// ::= identifier '(' expression* ')'
Lang Hames09bf4c12015-08-18 18:11:06 +0000256static std::unique_ptr<ExprAST> ParseIdentifierExpr() {
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000257 std::string IdName = IdentifierStr;
Eric Christopherc0239362014-12-08 18:12:28 +0000258
259 getNextToken(); // eat identifier.
260
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000261 if (CurTok != '(') // Simple variable ref.
Lang Hames09bf4c12015-08-18 18:11:06 +0000262 return llvm::make_unique<VariableExprAST>(IdName);
Eric Christopherc0239362014-12-08 18:12:28 +0000263
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000264 // Call.
Eric Christopherc0239362014-12-08 18:12:28 +0000265 getNextToken(); // eat (
Lang Hames09bf4c12015-08-18 18:11:06 +0000266 std::vector<std::unique_ptr<ExprAST>> Args;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000267 if (CurTok != ')') {
Eugene Zelenkof981ec42016-05-19 01:08:04 +0000268 while (true) {
Lang Hames09bf4c12015-08-18 18:11:06 +0000269 if (auto Arg = ParseExpression())
270 Args.push_back(std::move(Arg));
271 else
272 return nullptr;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000273
Eric Christopherc0239362014-12-08 18:12:28 +0000274 if (CurTok == ')')
275 break;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000276
277 if (CurTok != ',')
Lang Hames5d045a92016-03-25 17:41:26 +0000278 return LogError("Expected ')' or ',' in argument list");
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000279 getNextToken();
280 }
281 }
282
283 // Eat the ')'.
284 getNextToken();
Eric Christopherc0239362014-12-08 18:12:28 +0000285
Lang Hames09bf4c12015-08-18 18:11:06 +0000286 return llvm::make_unique<CallExprAST>(IdName, std::move(Args));
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000287}
288
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000289/// primary
290/// ::= identifierexpr
291/// ::= numberexpr
292/// ::= parenexpr
Lang Hames09bf4c12015-08-18 18:11:06 +0000293static std::unique_ptr<ExprAST> ParsePrimary() {
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000294 switch (CurTok) {
Eric Christopherc0239362014-12-08 18:12:28 +0000295 default:
Lang Hames5d045a92016-03-25 17:41:26 +0000296 return LogError("unknown token when expecting an expression");
Eric Christopherc0239362014-12-08 18:12:28 +0000297 case tok_identifier:
298 return ParseIdentifierExpr();
299 case tok_number:
300 return ParseNumberExpr();
301 case '(':
302 return ParseParenExpr();
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000303 }
304}
305
306/// binoprhs
307/// ::= ('+' primary)*
Lang Hames09bf4c12015-08-18 18:11:06 +0000308static std::unique_ptr<ExprAST> ParseBinOpRHS(int ExprPrec,
309 std::unique_ptr<ExprAST> LHS) {
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000310 // If this is a binop, find its precedence.
Eugene Zelenkof981ec42016-05-19 01:08:04 +0000311 while (true) {
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000312 int TokPrec = GetTokPrecedence();
Eric Christopherc0239362014-12-08 18:12:28 +0000313
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000314 // If this is a binop that binds at least as tightly as the current binop,
315 // consume it, otherwise we are done.
316 if (TokPrec < ExprPrec)
317 return LHS;
Eric Christopherc0239362014-12-08 18:12:28 +0000318
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000319 // Okay, we know this is a binop.
320 int BinOp = CurTok;
Eric Christopherc0239362014-12-08 18:12:28 +0000321 getNextToken(); // eat binop
322
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000323 // Parse the primary expression after the binary operator.
Lang Hames09bf4c12015-08-18 18:11:06 +0000324 auto RHS = ParsePrimary();
Eric Christopherc0239362014-12-08 18:12:28 +0000325 if (!RHS)
Lang Hames09bf4c12015-08-18 18:11:06 +0000326 return nullptr;
Eric Christopherc0239362014-12-08 18:12:28 +0000327
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000328 // If BinOp binds less tightly with RHS than the operator after RHS, let
329 // the pending operator take RHS as its LHS.
330 int NextPrec = GetTokPrecedence();
331 if (TokPrec < NextPrec) {
Lang Hames09bf4c12015-08-18 18:11:06 +0000332 RHS = ParseBinOpRHS(TokPrec + 1, std::move(RHS));
333 if (!RHS)
334 return nullptr;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000335 }
Eric Christopherc0239362014-12-08 18:12:28 +0000336
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000337 // Merge LHS/RHS.
Lang Hames59b0da82015-08-19 18:15:58 +0000338 LHS =
339 llvm::make_unique<BinaryExprAST>(BinOp, std::move(LHS), std::move(RHS));
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000340 }
341}
342
343/// expression
344/// ::= primary binoprhs
345///
Lang Hames09bf4c12015-08-18 18:11:06 +0000346static std::unique_ptr<ExprAST> ParseExpression() {
347 auto LHS = ParsePrimary();
Eric Christopherc0239362014-12-08 18:12:28 +0000348 if (!LHS)
Lang Hames09bf4c12015-08-18 18:11:06 +0000349 return nullptr;
Eric Christopherc0239362014-12-08 18:12:28 +0000350
Lang Hames09bf4c12015-08-18 18:11:06 +0000351 return ParseBinOpRHS(0, std::move(LHS));
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000352}
353
354/// prototype
355/// ::= id '(' id* ')'
Lang Hames09bf4c12015-08-18 18:11:06 +0000356static std::unique_ptr<PrototypeAST> ParsePrototype() {
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000357 if (CurTok != tok_identifier)
Lang Hames5d045a92016-03-25 17:41:26 +0000358 return LogErrorP("Expected function name in prototype");
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000359
360 std::string FnName = IdentifierStr;
361 getNextToken();
Eric Christopherc0239362014-12-08 18:12:28 +0000362
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000363 if (CurTok != '(')
Lang Hames5d045a92016-03-25 17:41:26 +0000364 return LogErrorP("Expected '(' in prototype");
Eric Christopherc0239362014-12-08 18:12:28 +0000365
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000366 std::vector<std::string> ArgNames;
367 while (getNextToken() == tok_identifier)
368 ArgNames.push_back(IdentifierStr);
369 if (CurTok != ')')
Lang Hames5d045a92016-03-25 17:41:26 +0000370 return LogErrorP("Expected ')' in prototype");
Eric Christopherc0239362014-12-08 18:12:28 +0000371
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000372 // success.
Eric Christopherc0239362014-12-08 18:12:28 +0000373 getNextToken(); // eat ')'.
374
Lang Hames09bf4c12015-08-18 18:11:06 +0000375 return llvm::make_unique<PrototypeAST>(FnName, std::move(ArgNames));
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000376}
377
378/// definition ::= 'def' prototype expression
Lang Hames09bf4c12015-08-18 18:11:06 +0000379static std::unique_ptr<FunctionAST> ParseDefinition() {
Eric Christopherc0239362014-12-08 18:12:28 +0000380 getNextToken(); // eat def.
Lang Hames09bf4c12015-08-18 18:11:06 +0000381 auto Proto = ParsePrototype();
382 if (!Proto)
383 return nullptr;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000384
Lang Hames09bf4c12015-08-18 18:11:06 +0000385 if (auto E = ParseExpression())
386 return llvm::make_unique<FunctionAST>(std::move(Proto), std::move(E));
387 return nullptr;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000388}
389
390/// toplevelexpr ::= expression
Lang Hames09bf4c12015-08-18 18:11:06 +0000391static std::unique_ptr<FunctionAST> ParseTopLevelExpr() {
392 if (auto E = ParseExpression()) {
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000393 // Make an anonymous proto.
Lang Hames2d789c32015-08-26 03:07:41 +0000394 auto Proto = llvm::make_unique<PrototypeAST>("__anon_expr",
395 std::vector<std::string>());
Lang Hames09bf4c12015-08-18 18:11:06 +0000396 return llvm::make_unique<FunctionAST>(std::move(Proto), std::move(E));
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000397 }
Lang Hames09bf4c12015-08-18 18:11:06 +0000398 return nullptr;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000399}
400
401/// external ::= 'extern' prototype
Lang Hames09bf4c12015-08-18 18:11:06 +0000402static std::unique_ptr<PrototypeAST> ParseExtern() {
Eric Christopherc0239362014-12-08 18:12:28 +0000403 getNextToken(); // eat extern.
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000404 return ParsePrototype();
405}
406
407//===----------------------------------------------------------------------===//
408// Code Generation
409//===----------------------------------------------------------------------===//
410
Mehdi Amini03b42e42016-04-14 21:59:01 +0000411static LLVMContext TheContext;
412static IRBuilder<> Builder(TheContext);
Lang Hames24796802016-05-22 22:48:36 +0000413static std::unique_ptr<Module> TheModule;
Eric Christopherc0239362014-12-08 18:12:28 +0000414static std::map<std::string, Value *> NamedValues;
Lang Hames2d789c32015-08-26 03:07:41 +0000415static std::unique_ptr<legacy::FunctionPassManager> TheFPM;
416static std::unique_ptr<KaleidoscopeJIT> TheJIT;
417static std::map<std::string, std::unique_ptr<PrototypeAST>> FunctionProtos;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000418
Lang Hames5d045a92016-03-25 17:41:26 +0000419Value *LogErrorV(const char *Str) {
420 LogError(Str);
Lang Hames09bf4c12015-08-18 18:11:06 +0000421 return nullptr;
Eric Christopherc0239362014-12-08 18:12:28 +0000422}
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000423
Lang Hames2d789c32015-08-26 03:07:41 +0000424Function *getFunction(std::string Name) {
425 // First, see if the function has already been added to the current module.
426 if (auto *F = TheModule->getFunction(Name))
427 return F;
428
429 // If not, check whether we can codegen the declaration from some existing
430 // prototype.
431 auto FI = FunctionProtos.find(Name);
432 if (FI != FunctionProtos.end())
433 return FI->second->codegen();
434
435 // If no existing prototype exists, return null.
436 return nullptr;
437}
438
439Value *NumberExprAST::codegen() {
Mehdi Amini03b42e42016-04-14 21:59:01 +0000440 return ConstantFP::get(TheContext, APFloat(Val));
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000441}
442
Lang Hames2d789c32015-08-26 03:07:41 +0000443Value *VariableExprAST::codegen() {
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000444 // Look this variable up in the function.
445 Value *V = NamedValues[Name];
Lang Hames596aec92015-08-19 18:32:58 +0000446 if (!V)
Lang Hames5d045a92016-03-25 17:41:26 +0000447 return LogErrorV("Unknown variable name");
Lang Hames596aec92015-08-19 18:32:58 +0000448 return V;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000449}
450
Lang Hames2d789c32015-08-26 03:07:41 +0000451Value *BinaryExprAST::codegen() {
452 Value *L = LHS->codegen();
453 Value *R = RHS->codegen();
Lang Hames09bf4c12015-08-18 18:11:06 +0000454 if (!L || !R)
455 return nullptr;
Eric Christopherc0239362014-12-08 18:12:28 +0000456
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000457 switch (Op) {
Eric Christopherc0239362014-12-08 18:12:28 +0000458 case '+':
459 return Builder.CreateFAdd(L, R, "addtmp");
460 case '-':
461 return Builder.CreateFSub(L, R, "subtmp");
462 case '*':
463 return Builder.CreateFMul(L, R, "multmp");
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000464 case '<':
465 L = Builder.CreateFCmpULT(L, R, "cmptmp");
466 // Convert bool 0/1 to double 0.0 or 1.0
Mehdi Amini03b42e42016-04-14 21:59:01 +0000467 return Builder.CreateUIToFP(L, Type::getDoubleTy(TheContext), "booltmp");
Eric Christopherc0239362014-12-08 18:12:28 +0000468 default:
Lang Hames5d045a92016-03-25 17:41:26 +0000469 return LogErrorV("invalid binary operator");
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000470 }
471}
472
Lang Hames2d789c32015-08-26 03:07:41 +0000473Value *CallExprAST::codegen() {
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000474 // Look up the name in the global module table.
Lang Hames2d789c32015-08-26 03:07:41 +0000475 Function *CalleeF = getFunction(Callee);
Lang Hames09bf4c12015-08-18 18:11:06 +0000476 if (!CalleeF)
Lang Hames5d045a92016-03-25 17:41:26 +0000477 return LogErrorV("Unknown function referenced");
Eric Christopherc0239362014-12-08 18:12:28 +0000478
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000479 // If argument mismatch error.
480 if (CalleeF->arg_size() != Args.size())
Lang Hames5d045a92016-03-25 17:41:26 +0000481 return LogErrorV("Incorrect # arguments passed");
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000482
Eric Christopherc0239362014-12-08 18:12:28 +0000483 std::vector<Value *> ArgsV;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000484 for (unsigned i = 0, e = Args.size(); i != e; ++i) {
Lang Hames2d789c32015-08-26 03:07:41 +0000485 ArgsV.push_back(Args[i]->codegen());
Lang Hames09bf4c12015-08-18 18:11:06 +0000486 if (!ArgsV.back())
487 return nullptr;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000488 }
Eric Christopherc0239362014-12-08 18:12:28 +0000489
Francois Pichetc5d10502011-07-15 10:59:52 +0000490 return Builder.CreateCall(CalleeF, ArgsV, "calltmp");
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000491}
492
Lang Hames2d789c32015-08-26 03:07:41 +0000493Function *PrototypeAST::codegen() {
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000494 // Make the function type: double(double,double) etc.
Mehdi Amini03b42e42016-04-14 21:59:01 +0000495 std::vector<Type *> Doubles(Args.size(), Type::getDoubleTy(TheContext));
Eric Christopherc0239362014-12-08 18:12:28 +0000496 FunctionType *FT =
Mehdi Amini03b42e42016-04-14 21:59:01 +0000497 FunctionType::get(Type::getDoubleTy(TheContext), Doubles, false);
Eric Christopherc0239362014-12-08 18:12:28 +0000498
Lang Hames2d789c32015-08-26 03:07:41 +0000499 Function *F =
500 Function::Create(FT, Function::ExternalLinkage, Name, TheModule.get());
Eric Christopherc0239362014-12-08 18:12:28 +0000501
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000502 // Set names for all arguments.
503 unsigned Idx = 0;
Lang Hames2d789c32015-08-26 03:07:41 +0000504 for (auto &Arg : F->args())
505 Arg.setName(Args[Idx++]);
Eric Christopherc0239362014-12-08 18:12:28 +0000506
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000507 return F;
508}
509
Lang Hames2d789c32015-08-26 03:07:41 +0000510Function *FunctionAST::codegen() {
511 // Transfer ownership of the prototype to the FunctionProtos map, but keep a
512 // reference to it for use below.
513 auto &P = *Proto;
514 FunctionProtos[Proto->getName()] = std::move(Proto);
515 Function *TheFunction = getFunction(P.getName());
Lang Hames09bf4c12015-08-18 18:11:06 +0000516 if (!TheFunction)
517 return nullptr;
Eric Christopherc0239362014-12-08 18:12:28 +0000518
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000519 // Create a new basic block to start insertion into.
Mehdi Amini03b42e42016-04-14 21:59:01 +0000520 BasicBlock *BB = BasicBlock::Create(TheContext, "entry", TheFunction);
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000521 Builder.SetInsertPoint(BB);
Eric Christopherc0239362014-12-08 18:12:28 +0000522
Lang Hames2d789c32015-08-26 03:07:41 +0000523 // Record the function arguments in the NamedValues map.
524 NamedValues.clear();
525 for (auto &Arg : TheFunction->args())
526 NamedValues[Arg.getName()] = &Arg;
527
528 if (Value *RetVal = Body->codegen()) {
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000529 // Finish off the function.
530 Builder.CreateRet(RetVal);
531
532 // Validate the generated code, checking for consistency.
533 verifyFunction(*TheFunction);
534
Lang Hames2d789c32015-08-26 03:07:41 +0000535 // Run the optimizer on the function.
536 TheFPM->run(*TheFunction);
537
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000538 return TheFunction;
539 }
Eric Christopherc0239362014-12-08 18:12:28 +0000540
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000541 // Error reading body, remove function.
542 TheFunction->eraseFromParent();
Lang Hames09bf4c12015-08-18 18:11:06 +0000543 return nullptr;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000544}
545
546//===----------------------------------------------------------------------===//
547// Top-Level parsing and JIT Driver
548//===----------------------------------------------------------------------===//
549
Lang Hames2d789c32015-08-26 03:07:41 +0000550static void InitializeModuleAndPassManager() {
551 // Open a new module.
Mehdi Amini03b42e42016-04-14 21:59:01 +0000552 TheModule = llvm::make_unique<Module>("my cool jit", TheContext);
Lang Hames2d789c32015-08-26 03:07:41 +0000553 TheModule->setDataLayout(TheJIT->getTargetMachine().createDataLayout());
554
555 // Create a new pass manager attached to it.
556 TheFPM = llvm::make_unique<legacy::FunctionPassManager>(TheModule.get());
557
Lang Hames2d789c32015-08-26 03:07:41 +0000558 // Do simple "peephole" optimizations and bit-twiddling optzns.
559 TheFPM->add(createInstructionCombiningPass());
560 // Reassociate expressions.
561 TheFPM->add(createReassociatePass());
562 // Eliminate Common SubExpressions.
563 TheFPM->add(createGVNPass());
564 // Simplify the control flow graph (deleting unreachable blocks, etc).
565 TheFPM->add(createCFGSimplificationPass());
566
567 TheFPM->doInitialization();
568}
569
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000570static void HandleDefinition() {
Lang Hames09bf4c12015-08-18 18:11:06 +0000571 if (auto FnAST = ParseDefinition()) {
Lang Hames2d789c32015-08-26 03:07:41 +0000572 if (auto *FnIR = FnAST->codegen()) {
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000573 fprintf(stderr, "Read function definition:");
Matthias Braun25bcaba2017-01-28 02:47:46 +0000574 FnIR->print(errs());
575 fprintf(stderr, "\n");
Lang Hames2d789c32015-08-26 03:07:41 +0000576 TheJIT->addModule(std::move(TheModule));
577 InitializeModuleAndPassManager();
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000578 }
579 } else {
580 // Skip token for error recovery.
581 getNextToken();
582 }
583}
584
585static void HandleExtern() {
Lang Hames09bf4c12015-08-18 18:11:06 +0000586 if (auto ProtoAST = ParseExtern()) {
Lang Hames2d789c32015-08-26 03:07:41 +0000587 if (auto *FnIR = ProtoAST->codegen()) {
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000588 fprintf(stderr, "Read extern: ");
Matthias Braun25bcaba2017-01-28 02:47:46 +0000589 FnIR->print(errs());
590 fprintf(stderr, "\n");
Lang Hames2d789c32015-08-26 03:07:41 +0000591 FunctionProtos[ProtoAST->getName()] = std::move(ProtoAST);
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000592 }
593 } else {
594 // Skip token for error recovery.
595 getNextToken();
596 }
597}
598
599static void HandleTopLevelExpression() {
600 // Evaluate a top-level expression into an anonymous function.
Lang Hames09bf4c12015-08-18 18:11:06 +0000601 if (auto FnAST = ParseTopLevelExpr()) {
Lang Hames2d789c32015-08-26 03:07:41 +0000602 if (FnAST->codegen()) {
Lang Hames2d789c32015-08-26 03:07:41 +0000603 // JIT the module containing the anonymous expression, keeping a handle so
604 // we can free it later.
605 auto H = TheJIT->addModule(std::move(TheModule));
606 InitializeModuleAndPassManager();
607
608 // Search the JIT for the __anon_expr symbol.
609 auto ExprSymbol = TheJIT->findSymbol("__anon_expr");
610 assert(ExprSymbol && "Function not found");
611
612 // Get the symbol's address and cast it to the right type (takes no
613 // arguments, returns a double) so we can call it as a native function.
Lang Hames4ce98662017-07-07 02:59:13 +0000614 double (*FP)() = (double (*)())(intptr_t)cantFail(ExprSymbol.getAddress());
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000615 fprintf(stderr, "Evaluated to %f\n", FP());
Lang Hames2d789c32015-08-26 03:07:41 +0000616
617 // Delete the anonymous expression module from the JIT.
618 TheJIT->removeModule(H);
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000619 }
620 } else {
621 // Skip token for error recovery.
622 getNextToken();
623 }
624}
625
626/// top ::= definition | external | expression | ';'
627static void MainLoop() {
Eugene Zelenkof981ec42016-05-19 01:08:04 +0000628 while (true) {
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000629 fprintf(stderr, "ready> ");
630 switch (CurTok) {
Eric Christopherc0239362014-12-08 18:12:28 +0000631 case tok_eof:
632 return;
Lang Hames59b0da82015-08-19 18:15:58 +0000633 case ';': // ignore top-level semicolons.
Eric Christopherc0239362014-12-08 18:12:28 +0000634 getNextToken();
Lang Hames59b0da82015-08-19 18:15:58 +0000635 break;
Eric Christopherc0239362014-12-08 18:12:28 +0000636 case tok_def:
637 HandleDefinition();
638 break;
639 case tok_extern:
640 HandleExtern();
641 break;
642 default:
643 HandleTopLevelExpression();
644 break;
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000645 }
646 }
647}
648
649//===----------------------------------------------------------------------===//
650// "Library" functions that can be "extern'd" from user code.
651//===----------------------------------------------------------------------===//
652
Mehdi Aminibb6805d2017-02-11 21:26:52 +0000653#ifdef LLVM_ON_WIN32
654#define DLLEXPORT __declspec(dllexport)
655#else
656#define DLLEXPORT
657#endif
658
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000659/// putchard - putchar that takes a double and returns 0.
Mehdi Aminibb6805d2017-02-11 21:26:52 +0000660extern "C" DLLEXPORT double putchard(double X) {
Lang Hamesd76e0672015-08-27 20:31:44 +0000661 fputc((char)X, stderr);
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000662 return 0;
663}
664
Lang Hames59b0da82015-08-19 18:15:58 +0000665/// printd - printf that takes a double prints it as "%f\n", returning 0.
Mehdi Aminibb6805d2017-02-11 21:26:52 +0000666extern "C" DLLEXPORT double printd(double X) {
Lang Hamesd76e0672015-08-27 20:31:44 +0000667 fprintf(stderr, "%f\n", X);
Lang Hames59b0da82015-08-19 18:15:58 +0000668 return 0;
669}
670
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000671//===----------------------------------------------------------------------===//
672// Main driver code.
673//===----------------------------------------------------------------------===//
674
675int main() {
676 InitializeNativeTarget();
Eric Christopher1b74b652014-12-08 18:00:38 +0000677 InitializeNativeTargetAsmPrinter();
678 InitializeNativeTargetAsmParser();
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000679
680 // Install standard binary operators.
681 // 1 is lowest precedence.
682 BinopPrecedence['<'] = 10;
683 BinopPrecedence['+'] = 20;
684 BinopPrecedence['-'] = 20;
Eric Christopherc0239362014-12-08 18:12:28 +0000685 BinopPrecedence['*'] = 40; // highest.
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000686
687 // Prime the first token.
688 fprintf(stderr, "ready> ");
689 getNextToken();
690
Lang Hames2d789c32015-08-26 03:07:41 +0000691 TheJIT = llvm::make_unique<KaleidoscopeJIT>();
692
693 InitializeModuleAndPassManager();
694
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000695 // Run the main "interpreter loop" now.
696 MainLoop();
697
Erick Tryzelaar21e83ea2009-09-22 21:15:19 +0000698 return 0;
699}