blob: 97ecdab0fe95010634b827ef59dd0b2cf42e9024 [file] [log] [blame]
Eugene Zelenko2b8e4172016-05-25 01:18:36 +00001//===-- BrainF.cpp - BrainF compiler example ------------------------------===//
Chris Lattner909ef092007-09-12 18:24:00 +00002//
3// The LLVM Compiler Infrastructure
4//
Chris Lattnerbcf65db2007-12-29 20:37:57 +00005// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
Chris Lattner909ef092007-09-12 18:24:00 +00007//
Eugene Zelenko2b8e4172016-05-25 01:18:36 +00008//===----------------------------------------------------------------------===//
Chris Lattner909ef092007-09-12 18:24:00 +00009//
10// This class compiles the BrainF language into LLVM assembly.
11//
12// The BrainF language has 8 commands:
13// Command Equivalent C Action
14// ------- ------------ ------
15// , *h=getchar(); Read a character from stdin, 255 on EOF
16// . putchar(*h); Write a character to stdout
17// - --*h; Decrement tape
18// + ++*h; Increment tape
19// < --h; Move head left
20// > ++h; Move head right
21// [ while(*h) { Start loop
22// ] } End loop
23//
Eugene Zelenko2b8e4172016-05-25 01:18:36 +000024//===----------------------------------------------------------------------===//
Chris Lattner909ef092007-09-12 18:24:00 +000025
26#include "BrainF.h"
Eugene Zelenko2b8e4172016-05-25 01:18:36 +000027#include "llvm/ADT/APInt.h"
28#include "llvm/IR/BasicBlock.h"
29#include "llvm/IR/Constant.h"
Chandler Carruth005f27a2013-01-02 11:56:33 +000030#include "llvm/IR/Constants.h"
Eugene Zelenko2b8e4172016-05-25 01:18:36 +000031#include "llvm/IR/DerivedTypes.h"
32#include "llvm/IR/Function.h"
33#include "llvm/IR/GlobalValue.h"
34#include "llvm/IR/GlobalVariable.h"
35#include "llvm/IR/InstrTypes.h"
36#include "llvm/IR/Instruction.h"
Chandler Carruth005f27a2013-01-02 11:56:33 +000037#include "llvm/IR/Instructions.h"
38#include "llvm/IR/Intrinsics.h"
Eugene Zelenko2b8e4172016-05-25 01:18:36 +000039#include "llvm/IR/Module.h"
40#include "llvm/IR/Type.h"
41#include "llvm/Support/Casting.h"
42#include <cstdlib>
Chris Lattner8e8eae62008-08-23 22:00:15 +000043#include <iostream>
Hans Wennborgcc9deb42015-09-29 18:02:48 +000044
Chris Lattner909ef092007-09-12 18:24:00 +000045using namespace llvm;
46
47//Set the constants for naming
48const char *BrainF::tapereg = "tape";
49const char *BrainF::headreg = "head";
50const char *BrainF::label = "brainf";
51const char *BrainF::testreg = "test";
52
Owen Anderson6773d382009-07-01 16:58:40 +000053Module *BrainF::parse(std::istream *in1, int mem, CompileFlags cf,
Owen Anderson2a154432009-07-01 23:13:44 +000054 LLVMContext& Context) {
Chris Lattner909ef092007-09-12 18:24:00 +000055 in = in1;
56 memtotal = mem;
57 comflag = cf;
58
Owen Anderson6773d382009-07-01 16:58:40 +000059 header(Context);
Hans Wennborgcc9deb42015-09-29 18:02:48 +000060 readloop(nullptr, nullptr, nullptr, Context);
Chris Lattner909ef092007-09-12 18:24:00 +000061 delete builder;
62 return module;
63}
64
Owen Anderson2a154432009-07-01 23:13:44 +000065void BrainF::header(LLVMContext& C) {
Owen Anderson6773d382009-07-01 16:58:40 +000066 module = new Module("BrainF", C);
Chris Lattner909ef092007-09-12 18:24:00 +000067
68 //Function prototypes
69
Chris Lattner0b6dce42010-08-10 21:45:38 +000070 //declare void @llvm.memset.p0i8.i32(i8 *, i8, i32, i32, i1)
Francois Pichetce206002011-07-12 22:04:11 +000071 Type *Tys[] = { Type::getInt8PtrTy(C), Type::getInt32Ty(C) };
Chris Lattnerdd708342008-11-21 16:42:48 +000072 Function *memset_func = Intrinsic::getDeclaration(module, Intrinsic::memset,
Benjamin Kramere6e19332011-07-14 17:45:39 +000073 Tys);
Chris Lattner909ef092007-09-12 18:24:00 +000074
75 //declare i32 @getchar()
76 getchar_func = cast<Function>(module->
Owen Anderson55f1c092009-08-13 21:58:54 +000077 getOrInsertFunction("getchar", IntegerType::getInt32Ty(C), NULL));
Chris Lattner909ef092007-09-12 18:24:00 +000078
79 //declare i32 @putchar(i32)
80 putchar_func = cast<Function>(module->
Owen Anderson55f1c092009-08-13 21:58:54 +000081 getOrInsertFunction("putchar", IntegerType::getInt32Ty(C),
82 IntegerType::getInt32Ty(C), NULL));
Chris Lattner909ef092007-09-12 18:24:00 +000083
Chris Lattner909ef092007-09-12 18:24:00 +000084 //Function header
85
86 //define void @brainf()
87 brainf_func = cast<Function>(module->
Owen Anderson55f1c092009-08-13 21:58:54 +000088 getOrInsertFunction("brainf", Type::getVoidTy(C), NULL));
Chris Lattner909ef092007-09-12 18:24:00 +000089
Owen Anderson55f1c092009-08-13 21:58:54 +000090 builder = new IRBuilder<>(BasicBlock::Create(C, label, brainf_func));
Chris Lattner909ef092007-09-12 18:24:00 +000091
92 //%arr = malloc i8, i32 %d
Owen Andersonedb4a702009-07-24 23:12:02 +000093 ConstantInt *val_mem = ConstantInt::get(C, APInt(32, memtotal));
Victor Hernandezc7d6a832009-10-17 00:00:19 +000094 BasicBlock* BB = builder->GetInsertBlock();
Chris Lattner805d0942011-07-18 04:52:58 +000095 Type* IntPtrTy = IntegerType::getInt32Ty(C);
96 Type* Int8Ty = IntegerType::getInt8Ty(C);
Victor Hernandezf3db9152009-11-07 00:16:28 +000097 Constant* allocsize = ConstantExpr::getSizeOf(Int8Ty);
98 allocsize = ConstantExpr::getTruncOrBitCast(allocsize, IntPtrTy);
99 ptr_arr = CallInst::CreateMalloc(BB, IntPtrTy, Int8Ty, allocsize, val_mem,
Hans Wennborgcc9deb42015-09-29 18:02:48 +0000100 nullptr, "arr");
Victor Hernandezc7d6a832009-10-17 00:00:19 +0000101 BB->getInstList().push_back(cast<Instruction>(ptr_arr));
Chris Lattner909ef092007-09-12 18:24:00 +0000102
Chris Lattner0b6dce42010-08-10 21:45:38 +0000103 //call void @llvm.memset.p0i8.i32(i8 *%arr, i8 0, i32 %d, i32 1, i1 0)
Chris Lattner909ef092007-09-12 18:24:00 +0000104 {
105 Value *memset_params[] = {
106 ptr_arr,
Owen Andersonedb4a702009-07-24 23:12:02 +0000107 ConstantInt::get(C, APInt(8, 0)),
Chris Lattner909ef092007-09-12 18:24:00 +0000108 val_mem,
Chris Lattner0b6dce42010-08-10 21:45:38 +0000109 ConstantInt::get(C, APInt(32, 1)),
110 ConstantInt::get(C, APInt(1, 0))
Chris Lattner909ef092007-09-12 18:24:00 +0000111 };
112
113 CallInst *memset_call = builder->
Francois Pichetc5d10502011-07-15 10:59:52 +0000114 CreateCall(memset_func, memset_params);
Chris Lattner909ef092007-09-12 18:24:00 +0000115 memset_call->setTailCall(false);
116 }
117
118 //%arrmax = getelementptr i8 *%arr, i32 %d
119 if (comflag & flag_arraybounds) {
120 ptr_arrmax = builder->
Owen Andersonedb4a702009-07-24 23:12:02 +0000121 CreateGEP(ptr_arr, ConstantInt::get(C, APInt(32, memtotal)), "arrmax");
Chris Lattner909ef092007-09-12 18:24:00 +0000122 }
123
124 //%head.%d = getelementptr i8 *%arr, i32 %d
125 curhead = builder->CreateGEP(ptr_arr,
Owen Andersonedb4a702009-07-24 23:12:02 +0000126 ConstantInt::get(C, APInt(32, memtotal/2)),
Chris Lattner909ef092007-09-12 18:24:00 +0000127 headreg);
128
Chris Lattner909ef092007-09-12 18:24:00 +0000129 //Function footer
130
131 //brainf.end:
Owen Anderson55f1c092009-08-13 21:58:54 +0000132 endbb = BasicBlock::Create(C, label, brainf_func);
Chris Lattner909ef092007-09-12 18:24:00 +0000133
Victor Hernandezde5ad422009-10-26 23:43:48 +0000134 //call free(i8 *%arr)
135 endbb->getInstList().push_back(CallInst::CreateFree(ptr_arr, endbb));
Chris Lattner909ef092007-09-12 18:24:00 +0000136
137 //ret void
Owen Anderson55f1c092009-08-13 21:58:54 +0000138 ReturnInst::Create(C, endbb);
Chris Lattner909ef092007-09-12 18:24:00 +0000139
Chris Lattner909ef092007-09-12 18:24:00 +0000140 //Error block for array out of bounds
141 if (comflag & flag_arraybounds)
142 {
143 //@aberrormsg = internal constant [%d x i8] c"\00"
Owen Andersonb6b25302009-07-14 23:09:55 +0000144 Constant *msg_0 =
Francois Pichet4ce6e6e2012-01-31 09:35:01 +0000145 ConstantDataArray::getString(C, "Error: The head has left the tape.",
146 true);
Chris Lattner909ef092007-09-12 18:24:00 +0000147
148 GlobalVariable *aberrormsg = new GlobalVariable(
Owen Andersonb17f3292009-07-08 19:03:57 +0000149 *module,
Chris Lattner909ef092007-09-12 18:24:00 +0000150 msg_0->getType(),
151 true,
152 GlobalValue::InternalLinkage,
153 msg_0,
Owen Andersonb17f3292009-07-08 19:03:57 +0000154 "aberrormsg");
Chris Lattner909ef092007-09-12 18:24:00 +0000155
156 //declare i32 @puts(i8 *)
157 Function *puts_func = cast<Function>(module->
Owen Anderson55f1c092009-08-13 21:58:54 +0000158 getOrInsertFunction("puts", IntegerType::getInt32Ty(C),
159 PointerType::getUnqual(IntegerType::getInt8Ty(C)), NULL));
Chris Lattner909ef092007-09-12 18:24:00 +0000160
161 //brainf.aberror:
Owen Anderson55f1c092009-08-13 21:58:54 +0000162 aberrorbb = BasicBlock::Create(C, label, brainf_func);
Chris Lattner909ef092007-09-12 18:24:00 +0000163
164 //call i32 @puts(i8 *getelementptr([%d x i8] *@aberrormsg, i32 0, i32 0))
165 {
Owen Anderson55f1c092009-08-13 21:58:54 +0000166 Constant *zero_32 = Constant::getNullValue(IntegerType::getInt32Ty(C));
Chris Lattner909ef092007-09-12 18:24:00 +0000167
168 Constant *gep_params[] = {
169 zero_32,
170 zero_32
171 };
172
173 Constant *msgptr = ConstantExpr::
NAKAMURA Takumi696f2752015-04-02 22:44:00 +0000174 getGetElementPtr(aberrormsg->getValueType(), aberrormsg, gep_params);
Chris Lattner909ef092007-09-12 18:24:00 +0000175
176 Value *puts_params[] = {
177 msgptr
178 };
179
180 CallInst *puts_call =
Gabor Greife9ecc682008-04-06 20:25:17 +0000181 CallInst::Create(puts_func,
Francois Pichetc5d10502011-07-15 10:59:52 +0000182 puts_params,
Gabor Greife9ecc682008-04-06 20:25:17 +0000183 "", aberrorbb);
Chris Lattner909ef092007-09-12 18:24:00 +0000184 puts_call->setTailCall(false);
185 }
186
187 //br label %brainf.end
Gabor Greife9ecc682008-04-06 20:25:17 +0000188 BranchInst::Create(endbb, aberrorbb);
Chris Lattner909ef092007-09-12 18:24:00 +0000189 }
190}
191
Owen Andersonb6b25302009-07-14 23:09:55 +0000192void BrainF::readloop(PHINode *phi, BasicBlock *oldbb, BasicBlock *testbb,
193 LLVMContext &C) {
Chris Lattner909ef092007-09-12 18:24:00 +0000194 Symbol cursym = SYM_NONE;
195 int curvalue = 0;
196 Symbol nextsym = SYM_NONE;
197 int nextvalue = 0;
198 char c;
199 int loop;
200 int direction;
201
202 while(cursym != SYM_EOF && cursym != SYM_ENDLOOP) {
203 // Write out commands
204 switch(cursym) {
205 case SYM_NONE:
206 // Do nothing
207 break;
208
209 case SYM_READ:
210 {
211 //%tape.%d = call i32 @getchar()
NAKAMURA Takumi5b9bc2f2015-05-19 06:50:19 +0000212 CallInst *getchar_call =
213 builder->CreateCall(getchar_func, {}, tapereg);
Chris Lattner909ef092007-09-12 18:24:00 +0000214 getchar_call->setTailCall(false);
215 Value *tape_0 = getchar_call;
216
217 //%tape.%d = trunc i32 %tape.%d to i8
Duncan Sandsa07136e2008-04-13 06:22:09 +0000218 Value *tape_1 = builder->
Owen Anderson55f1c092009-08-13 21:58:54 +0000219 CreateTrunc(tape_0, IntegerType::getInt8Ty(C), tapereg);
Chris Lattner909ef092007-09-12 18:24:00 +0000220
221 //store i8 %tape.%d, i8 *%head.%d
222 builder->CreateStore(tape_1, curhead);
223 }
224 break;
225
226 case SYM_WRITE:
227 {
228 //%tape.%d = load i8 *%head.%d
229 LoadInst *tape_0 = builder->CreateLoad(curhead, tapereg);
230
231 //%tape.%d = sext i8 %tape.%d to i32
Duncan Sandsa07136e2008-04-13 06:22:09 +0000232 Value *tape_1 = builder->
Owen Anderson55f1c092009-08-13 21:58:54 +0000233 CreateSExt(tape_0, IntegerType::getInt32Ty(C), tapereg);
Chris Lattner909ef092007-09-12 18:24:00 +0000234
235 //call i32 @putchar(i32 %tape.%d)
236 Value *putchar_params[] = {
237 tape_1
238 };
239 CallInst *putchar_call = builder->
240 CreateCall(putchar_func,
Francois Pichetc5d10502011-07-15 10:59:52 +0000241 putchar_params);
Chris Lattner909ef092007-09-12 18:24:00 +0000242 putchar_call->setTailCall(false);
243 }
244 break;
245
246 case SYM_MOVE:
247 {
248 //%head.%d = getelementptr i8 *%head.%d, i32 %d
249 curhead = builder->
Owen Andersonedb4a702009-07-24 23:12:02 +0000250 CreateGEP(curhead, ConstantInt::get(C, APInt(32, curvalue)),
Chris Lattner909ef092007-09-12 18:24:00 +0000251 headreg);
252
253 //Error block for array out of bounds
254 if (comflag & flag_arraybounds)
255 {
256 //%test.%d = icmp uge i8 *%head.%d, %arrmax
Duncan Sandsa07136e2008-04-13 06:22:09 +0000257 Value *test_0 = builder->
Chris Lattner909ef092007-09-12 18:24:00 +0000258 CreateICmpUGE(curhead, ptr_arrmax, testreg);
259
260 //%test.%d = icmp ult i8 *%head.%d, %arr
Duncan Sandsa07136e2008-04-13 06:22:09 +0000261 Value *test_1 = builder->
Chris Lattner909ef092007-09-12 18:24:00 +0000262 CreateICmpULT(curhead, ptr_arr, testreg);
263
264 //%test.%d = or i1 %test.%d, %test.%d
Duncan Sandsa07136e2008-04-13 06:22:09 +0000265 Value *test_2 = builder->
Chris Lattner909ef092007-09-12 18:24:00 +0000266 CreateOr(test_0, test_1, testreg);
267
268 //br i1 %test.%d, label %main.%d, label %main.%d
Owen Anderson55f1c092009-08-13 21:58:54 +0000269 BasicBlock *nextbb = BasicBlock::Create(C, label, brainf_func);
Chris Lattner909ef092007-09-12 18:24:00 +0000270 builder->CreateCondBr(test_2, aberrorbb, nextbb);
271
272 //main.%d:
273 builder->SetInsertPoint(nextbb);
274 }
275 }
276 break;
277
278 case SYM_CHANGE:
279 {
280 //%tape.%d = load i8 *%head.%d
281 LoadInst *tape_0 = builder->CreateLoad(curhead, tapereg);
282
283 //%tape.%d = add i8 %tape.%d, %d
Duncan Sandsa07136e2008-04-13 06:22:09 +0000284 Value *tape_1 = builder->
Owen Andersonedb4a702009-07-24 23:12:02 +0000285 CreateAdd(tape_0, ConstantInt::get(C, APInt(8, curvalue)), tapereg);
Chris Lattner909ef092007-09-12 18:24:00 +0000286
287 //store i8 %tape.%d, i8 *%head.%d\n"
288 builder->CreateStore(tape_1, curhead);
289 }
290 break;
291
292 case SYM_LOOP:
293 {
294 //br label %main.%d
Owen Anderson55f1c092009-08-13 21:58:54 +0000295 BasicBlock *testbb = BasicBlock::Create(C, label, brainf_func);
Chris Lattner909ef092007-09-12 18:24:00 +0000296 builder->CreateBr(testbb);
297
298 //main.%d:
299 BasicBlock *bb_0 = builder->GetInsertBlock();
Owen Anderson55f1c092009-08-13 21:58:54 +0000300 BasicBlock *bb_1 = BasicBlock::Create(C, label, brainf_func);
Chris Lattner909ef092007-09-12 18:24:00 +0000301 builder->SetInsertPoint(bb_1);
302
Gabor Greif697e94c2008-05-15 10:04:30 +0000303 // Make part of PHI instruction now, wait until end of loop to finish
304 PHINode *phi_0 =
Owen Anderson55f1c092009-08-13 21:58:54 +0000305 PHINode::Create(PointerType::getUnqual(IntegerType::getInt8Ty(C)),
Jay Foad52131342011-03-30 11:28:46 +0000306 2, headreg, testbb);
Chris Lattner909ef092007-09-12 18:24:00 +0000307 phi_0->addIncoming(curhead, bb_0);
308 curhead = phi_0;
309
Owen Andersonb6b25302009-07-14 23:09:55 +0000310 readloop(phi_0, bb_1, testbb, C);
Chris Lattner909ef092007-09-12 18:24:00 +0000311 }
312 break;
313
314 default:
Chris Lattner8e8eae62008-08-23 22:00:15 +0000315 std::cerr << "Error: Unknown symbol.\n";
Chris Lattner909ef092007-09-12 18:24:00 +0000316 abort();
317 break;
318 }
319
320 cursym = nextsym;
321 curvalue = nextvalue;
322 nextsym = SYM_NONE;
323
324 // Reading stdin loop
325 loop = (cursym == SYM_NONE)
326 || (cursym == SYM_MOVE)
327 || (cursym == SYM_CHANGE);
328 while(loop) {
329 *in>>c;
330 if (in->eof()) {
331 if (cursym == SYM_NONE) {
332 cursym = SYM_EOF;
333 } else {
334 nextsym = SYM_EOF;
335 }
336 loop = 0;
337 } else {
338 direction = 1;
339 switch(c) {
340 case '-':
341 direction = -1;
342 // Fall through
343
344 case '+':
345 if (cursym == SYM_CHANGE) {
346 curvalue += direction;
347 // loop = 1
348 } else {
349 if (cursym == SYM_NONE) {
350 cursym = SYM_CHANGE;
351 curvalue = direction;
352 // loop = 1
353 } else {
354 nextsym = SYM_CHANGE;
355 nextvalue = direction;
356 loop = 0;
357 }
358 }
359 break;
360
361 case '<':
362 direction = -1;
363 // Fall through
364
365 case '>':
366 if (cursym == SYM_MOVE) {
367 curvalue += direction;
368 // loop = 1
369 } else {
370 if (cursym == SYM_NONE) {
371 cursym = SYM_MOVE;
372 curvalue = direction;
373 // loop = 1
374 } else {
375 nextsym = SYM_MOVE;
376 nextvalue = direction;
377 loop = 0;
378 }
379 }
380 break;
381
382 case ',':
383 if (cursym == SYM_NONE) {
384 cursym = SYM_READ;
385 } else {
386 nextsym = SYM_READ;
387 }
388 loop = 0;
389 break;
390
391 case '.':
392 if (cursym == SYM_NONE) {
393 cursym = SYM_WRITE;
394 } else {
395 nextsym = SYM_WRITE;
396 }
397 loop = 0;
398 break;
399
400 case '[':
401 if (cursym == SYM_NONE) {
402 cursym = SYM_LOOP;
403 } else {
404 nextsym = SYM_LOOP;
405 }
406 loop = 0;
407 break;
408
409 case ']':
410 if (cursym == SYM_NONE) {
411 cursym = SYM_ENDLOOP;
412 } else {
413 nextsym = SYM_ENDLOOP;
414 }
415 loop = 0;
416 break;
417
418 // Ignore other characters
419 default:
420 break;
421 }
422 }
423 }
424 }
425
426 if (cursym == SYM_ENDLOOP) {
427 if (!phi) {
Chris Lattner8e8eae62008-08-23 22:00:15 +0000428 std::cerr << "Error: Extra ']'\n";
Chris Lattner909ef092007-09-12 18:24:00 +0000429 abort();
430 }
431
432 // Write loop test
433 {
434 //br label %main.%d
435 builder->CreateBr(testbb);
436
437 //main.%d:
438
439 //%head.%d = phi i8 *[%head.%d, %main.%d], [%head.%d, %main.%d]
440 //Finish phi made at beginning of loop
441 phi->addIncoming(curhead, builder->GetInsertBlock());
442 Value *head_0 = phi;
443
444 //%tape.%d = load i8 *%head.%d
445 LoadInst *tape_0 = new LoadInst(head_0, tapereg, testbb);
446
447 //%test.%d = icmp eq i8 %tape.%d, 0
Owen Anderson1e5f00e2009-07-09 23:48:35 +0000448 ICmpInst *test_0 = new ICmpInst(*testbb, ICmpInst::ICMP_EQ, tape_0,
Owen Andersonedb4a702009-07-24 23:12:02 +0000449 ConstantInt::get(C, APInt(8, 0)), testreg);
Chris Lattner909ef092007-09-12 18:24:00 +0000450
451 //br i1 %test.%d, label %main.%d, label %main.%d
Owen Anderson55f1c092009-08-13 21:58:54 +0000452 BasicBlock *bb_0 = BasicBlock::Create(C, label, brainf_func);
Gabor Greife9ecc682008-04-06 20:25:17 +0000453 BranchInst::Create(bb_0, oldbb, test_0, testbb);
Chris Lattner909ef092007-09-12 18:24:00 +0000454
455 //main.%d:
456 builder->SetInsertPoint(bb_0);
457
458 //%head.%d = phi i8 *[%head.%d, %main.%d]
459 PHINode *phi_1 = builder->
Jay Foad52131342011-03-30 11:28:46 +0000460 CreatePHI(PointerType::getUnqual(IntegerType::getInt8Ty(C)), 1,
461 headreg);
Chris Lattner909ef092007-09-12 18:24:00 +0000462 phi_1->addIncoming(head_0, testbb);
463 curhead = phi_1;
464 }
465
466 return;
467 }
468
469 //End of the program, so go to return block
470 builder->CreateBr(endbb);
471
472 if (phi) {
Chris Lattner8e8eae62008-08-23 22:00:15 +0000473 std::cerr << "Error: Missing ']'\n";
Chris Lattner909ef092007-09-12 18:24:00 +0000474 abort();
475 }
476}