blob: 5448b5c6b25742986b1f51ca6450900c20de6cbd [file] [log] [blame]
Sean Callananee5dfd42010-02-01 08:49:35 +00001//===-EDDisassembler.cpp - LLVM Enhanced Disassembler ---------------------===//
2//
3// The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10// This file implements the Enhanced Disassembly library's disassembler class.
11// The disassembler is responsible for vending individual instructions according
12// to a given architecture and disassembly syntax.
13//
14//===----------------------------------------------------------------------===//
15
Sean Callanan8f993b82010-04-08 00:48:21 +000016#include "EDDisassembler.h"
17#include "EDInst.h"
Sean Callanan9899f702010-04-13 21:21:57 +000018#include "llvm/MC/EDInstInfo.h"
Sean Callananee5dfd42010-02-01 08:49:35 +000019#include "llvm/MC/MCAsmInfo.h"
20#include "llvm/MC/MCContext.h"
21#include "llvm/MC/MCDisassembler.h"
22#include "llvm/MC/MCExpr.h"
23#include "llvm/MC/MCInst.h"
24#include "llvm/MC/MCInstPrinter.h"
Evan Cheng0e6a0522011-07-18 20:57:22 +000025#include "llvm/MC/MCRegisterInfo.h"
Sean Callananee5dfd42010-02-01 08:49:35 +000026#include "llvm/MC/MCStreamer.h"
Evan Chengffc0e732011-07-09 05:47:46 +000027#include "llvm/MC/MCSubtargetInfo.h"
Sean Callananee5dfd42010-02-01 08:49:35 +000028#include "llvm/MC/MCParser/AsmLexer.h"
Sean Callananee5dfd42010-02-01 08:49:35 +000029#include "llvm/MC/MCParser/MCAsmParser.h"
30#include "llvm/MC/MCParser/MCParsedAsmOperand.h"
31#include "llvm/Support/MemoryBuffer.h"
32#include "llvm/Support/MemoryObject.h"
33#include "llvm/Support/SourceMgr.h"
34#include "llvm/Target/TargetAsmLexer.h"
35#include "llvm/Target/TargetAsmParser.h"
36#include "llvm/Target/TargetRegistry.h"
37#include "llvm/Target/TargetMachine.h"
38#include "llvm/Target/TargetRegisterInfo.h"
39#include "llvm/Target/TargetSelect.h"
Sean Callananee5dfd42010-02-01 08:49:35 +000040using namespace llvm;
41
42bool EDDisassembler::sInitialized = false;
43EDDisassembler::DisassemblerMap_t EDDisassembler::sDisassemblers;
44
Sean Callanan9899f702010-04-13 21:21:57 +000045struct TripleMap {
Sean Callananee5dfd42010-02-01 08:49:35 +000046 Triple::ArchType Arch;
47 const char *String;
Sean Callananee5dfd42010-02-01 08:49:35 +000048};
49
Sean Callanan9899f702010-04-13 21:21:57 +000050static struct TripleMap triplemap[] = {
51 { Triple::x86, "i386-unknown-unknown" },
52 { Triple::x86_64, "x86_64-unknown-unknown" },
53 { Triple::arm, "arm-unknown-unknown" },
54 { Triple::thumb, "thumb-unknown-unknown" },
55 { Triple::InvalidArch, NULL, }
Sean Callananee5dfd42010-02-01 08:49:35 +000056};
57
Sean Callanan9899f702010-04-13 21:21:57 +000058/// infoFromArch - Returns the TripleMap corresponding to a given architecture,
Sean Callananee5dfd42010-02-01 08:49:35 +000059/// or NULL if there is an error
60///
61/// @arg arch - The Triple::ArchType for the desired architecture
Sean Callanan9899f702010-04-13 21:21:57 +000062static const char *tripleFromArch(Triple::ArchType arch) {
Sean Callananee5dfd42010-02-01 08:49:35 +000063 unsigned int infoIndex;
64
Sean Callanan9899f702010-04-13 21:21:57 +000065 for (infoIndex = 0; triplemap[infoIndex].String != NULL; ++infoIndex) {
66 if (arch == triplemap[infoIndex].Arch)
67 return triplemap[infoIndex].String;
Sean Callananee5dfd42010-02-01 08:49:35 +000068 }
69
70 return NULL;
71}
72
73/// getLLVMSyntaxVariant - gets the constant to use to get an assembly printer
74/// for the desired assembly syntax, suitable for passing to
75/// Target::createMCInstPrinter()
76///
77/// @arg arch - The target architecture
78/// @arg syntax - The assembly syntax in sd form
79static int getLLVMSyntaxVariant(Triple::ArchType arch,
Chris Lattner847da552010-07-20 18:25:19 +000080 EDDisassembler::AssemblySyntax syntax) {
Sean Callananee5dfd42010-02-01 08:49:35 +000081 switch (syntax) {
82 default:
83 return -1;
84 // Mappings below from X86AsmPrinter.cpp
Chris Lattner847da552010-07-20 18:25:19 +000085 case EDDisassembler::kEDAssemblySyntaxX86ATT:
Sean Callananee5dfd42010-02-01 08:49:35 +000086 if (arch == Triple::x86 || arch == Triple::x86_64)
87 return 0;
88 else
89 return -1;
Chris Lattner847da552010-07-20 18:25:19 +000090 case EDDisassembler::kEDAssemblySyntaxX86Intel:
Sean Callananee5dfd42010-02-01 08:49:35 +000091 if (arch == Triple::x86 || arch == Triple::x86_64)
92 return 1;
93 else
94 return -1;
Chris Lattner847da552010-07-20 18:25:19 +000095 case EDDisassembler::kEDAssemblySyntaxARMUAL:
Sean Callanan8f993b82010-04-08 00:48:21 +000096 if (arch == Triple::arm || arch == Triple::thumb)
97 return 0;
98 else
99 return -1;
Sean Callananee5dfd42010-02-01 08:49:35 +0000100 }
101}
102
Sean Callananee5dfd42010-02-01 08:49:35 +0000103void EDDisassembler::initialize() {
104 if (sInitialized)
105 return;
106
107 sInitialized = true;
108
Sean Callanan9899f702010-04-13 21:21:57 +0000109 InitializeAllTargetInfos();
110 InitializeAllTargets();
Evan Cheng1abf2cb2011-07-14 23:50:31 +0000111 InitializeAllMCAsmInfos();
Evan Cheng0e6a0522011-07-18 20:57:22 +0000112 InitializeAllMCRegisterInfos();
Sean Callananc872143d2011-07-19 00:00:41 +0000113 InitializeAllMCSubtargetInfos();
Sean Callanan9899f702010-04-13 21:21:57 +0000114 InitializeAllAsmPrinters();
115 InitializeAllAsmParsers();
116 InitializeAllDisassemblers();
Sean Callananee5dfd42010-02-01 08:49:35 +0000117}
118
119#undef BRINGUP_TARGET
120
121EDDisassembler *EDDisassembler::getDisassembler(Triple::ArchType arch,
Chris Lattner847da552010-07-20 18:25:19 +0000122 AssemblySyntax syntax) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000123 CPUKey key;
124 key.Arch = arch;
125 key.Syntax = syntax;
126
127 EDDisassembler::DisassemblerMap_t::iterator i = sDisassemblers.find(key);
128
129 if (i != sDisassemblers.end()) {
130 return i->second;
Sean Callanan8f993b82010-04-08 00:48:21 +0000131 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000132 EDDisassembler* sdd = new EDDisassembler(key);
Sean Callanan8f993b82010-04-08 00:48:21 +0000133 if (!sdd->valid()) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000134 delete sdd;
135 return NULL;
136 }
137
138 sDisassemblers[key] = sdd;
139
140 return sdd;
141 }
142
143 return NULL;
144}
145
146EDDisassembler *EDDisassembler::getDisassembler(StringRef str,
Chris Lattner847da552010-07-20 18:25:19 +0000147 AssemblySyntax syntax) {
148 return getDisassembler(Triple(str).getArch(), syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000149}
150
Sean Callananee5dfd42010-02-01 08:49:35 +0000151EDDisassembler::EDDisassembler(CPUKey &key) :
Sean Callanan8f993b82010-04-08 00:48:21 +0000152 Valid(false),
153 HasSemantics(false),
154 ErrorStream(nulls()),
155 Key(key) {
Sean Callanan9899f702010-04-13 21:21:57 +0000156 const char *triple = tripleFromArch(key.Arch);
Sean Callanan8f993b82010-04-08 00:48:21 +0000157
Sean Callanan9899f702010-04-13 21:21:57 +0000158 if (!triple)
159 return;
Sean Callananee5dfd42010-02-01 08:49:35 +0000160
Sean Callanan4285b292010-04-09 00:11:15 +0000161 LLVMSyntaxVariant = getLLVMSyntaxVariant(key.Arch, key.Syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000162
Sean Callanan4285b292010-04-09 00:11:15 +0000163 if (LLVMSyntaxVariant < 0)
Sean Callananee5dfd42010-02-01 08:49:35 +0000164 return;
165
166 std::string tripleString(triple);
167 std::string errorString;
168
169 Tgt = TargetRegistry::lookupTarget(tripleString,
170 errorString);
171
172 if (!Tgt)
173 return;
174
Evan Cheng276365d2011-06-30 01:53:36 +0000175 std::string CPU;
Sean Callananee5dfd42010-02-01 08:49:35 +0000176 std::string featureString;
Evan Cheng276365d2011-06-30 01:53:36 +0000177 TargetMachine.reset(Tgt->createTargetMachine(tripleString, CPU,
Daniel Dunbar02612432010-07-19 00:33:43 +0000178 featureString));
Evan Chengebdeeab2011-07-08 01:53:10 +0000179
Daniel Dunbar02612432010-07-19 00:33:43 +0000180 const TargetRegisterInfo *registerInfo = TargetMachine->getRegisterInfo();
Sean Callananee5dfd42010-02-01 08:49:35 +0000181
182 if (!registerInfo)
183 return;
Sean Callanan8f993b82010-04-08 00:48:21 +0000184
185 initMaps(*registerInfo);
Sean Callananee5dfd42010-02-01 08:49:35 +0000186
Evan Cheng1abf2cb2011-07-14 23:50:31 +0000187 AsmInfo.reset(Tgt->createMCAsmInfo(tripleString));
Sean Callananee5dfd42010-02-01 08:49:35 +0000188
189 if (!AsmInfo)
190 return;
Evan Chengebdeeab2011-07-08 01:53:10 +0000191
Evan Cheng0e6a0522011-07-18 20:57:22 +0000192 MRI.reset(Tgt->createMCRegInfo(tripleString));
193
194 if (!MRI)
195 return;
196
Sean Callananee5dfd42010-02-01 08:49:35 +0000197 Disassembler.reset(Tgt->createMCDisassembler());
198
199 if (!Disassembler)
200 return;
Sean Callanan9899f702010-04-13 21:21:57 +0000201
202 InstInfos = Disassembler->getEDInfo();
Sean Callananee5dfd42010-02-01 08:49:35 +0000203
204 InstString.reset(new std::string);
205 InstStream.reset(new raw_string_ostream(*InstString));
Evan Chengb2627992011-07-06 19:45:42 +0000206 InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo));
Sean Callananee5dfd42010-02-01 08:49:35 +0000207
208 if (!InstPrinter)
209 return;
210
211 GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
212 SpecificAsmLexer.reset(Tgt->createAsmLexer(*AsmInfo));
213 SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
Sean Callanand74667e2010-02-02 02:18:20 +0000214
Daniel Dunbar02612432010-07-19 00:33:43 +0000215 initMaps(*TargetMachine->getRegisterInfo());
Sean Callananee5dfd42010-02-01 08:49:35 +0000216
217 Valid = true;
218}
219
220EDDisassembler::~EDDisassembler() {
Sean Callanan8f993b82010-04-08 00:48:21 +0000221 if (!valid())
Sean Callananee5dfd42010-02-01 08:49:35 +0000222 return;
223}
224
225namespace {
226 /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
227 /// as provided by the sd interface. See MemoryObject.
228 class EDMemoryObject : public llvm::MemoryObject {
229 private:
230 EDByteReaderCallback Callback;
231 void *Arg;
232 public:
233 EDMemoryObject(EDByteReaderCallback callback,
234 void *arg) : Callback(callback), Arg(arg) { }
235 ~EDMemoryObject() { }
236 uint64_t getBase() const { return 0x0; }
237 uint64_t getExtent() const { return (uint64_t)-1; }
238 int readByte(uint64_t address, uint8_t *ptr) const {
Sean Callanan8f993b82010-04-08 00:48:21 +0000239 if (!Callback)
Sean Callananee5dfd42010-02-01 08:49:35 +0000240 return -1;
241
Sean Callanan8f993b82010-04-08 00:48:21 +0000242 if (Callback(ptr, address, Arg))
Sean Callananee5dfd42010-02-01 08:49:35 +0000243 return -1;
244
245 return 0;
246 }
247 };
248}
249
250EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
251 uint64_t address,
252 void *arg) {
253 EDMemoryObject memoryObject(byteReader, arg);
254
255 MCInst* inst = new MCInst;
256 uint64_t byteSize;
257
258 if (!Disassembler->getInstruction(*inst,
259 byteSize,
260 memoryObject,
261 address,
262 ErrorStream)) {
263 delete inst;
264 return NULL;
Sean Callanan8f993b82010-04-08 00:48:21 +0000265 } else {
Shantonu Sen0da9f132011-04-05 18:40:45 +0000266 const llvm::EDInstInfo *thisInstInfo = NULL;
Sean Callanan8f993b82010-04-08 00:48:21 +0000267
Shantonu Sen0da9f132011-04-05 18:40:45 +0000268 if (InstInfos) {
269 thisInstInfo = &InstInfos[inst->getOpcode()];
270 }
Sean Callananee5dfd42010-02-01 08:49:35 +0000271
272 EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
273 return sdInst;
274 }
275}
276
277void EDDisassembler::initMaps(const TargetRegisterInfo &registerInfo) {
278 unsigned numRegisters = registerInfo.getNumRegs();
279 unsigned registerIndex;
280
281 for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
282 const char* registerName = registerInfo.get(registerIndex).Name;
283
284 RegVec.push_back(registerName);
285 RegRMap[registerName] = registerIndex;
286 }
287
Sean Callanan8f993b82010-04-08 00:48:21 +0000288 switch (Key.Arch) {
289 default:
290 break;
291 case Triple::x86:
292 case Triple::x86_64:
Sean Callananee5dfd42010-02-01 08:49:35 +0000293 stackPointers.insert(registerIDWithName("SP"));
294 stackPointers.insert(registerIDWithName("ESP"));
295 stackPointers.insert(registerIDWithName("RSP"));
296
297 programCounters.insert(registerIDWithName("IP"));
298 programCounters.insert(registerIDWithName("EIP"));
299 programCounters.insert(registerIDWithName("RIP"));
Sean Callanan8f993b82010-04-08 00:48:21 +0000300 break;
301 case Triple::arm:
302 case Triple::thumb:
303 stackPointers.insert(registerIDWithName("SP"));
304
305 programCounters.insert(registerIDWithName("PC"));
306 break;
Sean Callananee5dfd42010-02-01 08:49:35 +0000307 }
308}
309
310const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
311 if (registerID >= RegVec.size())
312 return NULL;
313 else
314 return RegVec[registerID].c_str();
315}
316
317unsigned EDDisassembler::registerIDWithName(const char *name) const {
318 regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
319 if (iter == RegRMap.end())
320 return 0;
321 else
322 return (*iter).second;
323}
324
325bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
326 return (stackPointers.find(registerID) != stackPointers.end());
327}
328
329bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
330 return (programCounters.find(registerID) != programCounters.end());
331}
332
Chris Lattnerd3740872010-04-04 05:04:31 +0000333int EDDisassembler::printInst(std::string &str, MCInst &inst) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000334 PrinterMutex.acquire();
335
Chris Lattnerd3740872010-04-04 05:04:31 +0000336 InstPrinter->printInst(&inst, *InstStream);
Sean Callananee5dfd42010-02-01 08:49:35 +0000337 InstStream->flush();
338 str = *InstString;
339 InstString->clear();
340
341 PrinterMutex.release();
342
343 return 0;
344}
345
Sean Callanan457856c2011-04-09 00:37:25 +0000346static void diag_handler(const SMDiagnostic &diag,
347 void *context)
348{
349 if (context) {
350 EDDisassembler *disassembler = static_cast<EDDisassembler*>(context);
351 diag.Print("", disassembler->ErrorStream);
352 }
353}
354
Sean Callananee5dfd42010-02-01 08:49:35 +0000355int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
356 SmallVectorImpl<AsmToken> &tokens,
357 const std::string &str) {
358 int ret = 0;
359
Sean Callanan8f993b82010-04-08 00:48:21 +0000360 switch (Key.Arch) {
361 default:
362 return -1;
363 case Triple::x86:
364 case Triple::x86_64:
365 case Triple::arm:
366 case Triple::thumb:
367 break;
368 }
369
Sean Callananee5dfd42010-02-01 08:49:35 +0000370 const char *cStr = str.c_str();
371 MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
372
373 StringRef instName;
374 SMLoc instLoc;
375
376 SourceMgr sourceMgr;
Sean Callanan457856c2011-04-09 00:37:25 +0000377 sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this));
Sean Callananee5dfd42010-02-01 08:49:35 +0000378 sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
Evan Cheng0e6a0522011-07-18 20:57:22 +0000379 MCContext context(*AsmInfo, *MRI, NULL);
Chris Lattnerc18409a2010-03-11 22:53:35 +0000380 OwningPtr<MCStreamer> streamer(createNullStreamer(context));
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000381 OwningPtr<MCAsmParser> genericParser(createMCAsmParser(*Tgt, sourceMgr,
382 context, *streamer,
383 *AsmInfo));
Evan Chengebdeeab2011-07-08 01:53:10 +0000384
385 StringRef triple = tripleFromArch(Key.Arch);
Evan Chengffc0e732011-07-09 05:47:46 +0000386 OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(triple, "", ""));
387 OwningPtr<TargetAsmParser> TargetParser(Tgt->createAsmParser(*STI,
Evan Chengebdeeab2011-07-08 01:53:10 +0000388 *genericParser));
Sean Callananee5dfd42010-02-01 08:49:35 +0000389
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000390 AsmToken OpcodeToken = genericParser->Lex();
391 AsmToken NextToken = genericParser->Lex(); // consume next token, because specificParser expects us to
Sean Callanana8702562010-04-24 01:00:16 +0000392
Sean Callanan8f993b82010-04-08 00:48:21 +0000393 if (OpcodeToken.is(AsmToken::Identifier)) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000394 instName = OpcodeToken.getString();
395 instLoc = OpcodeToken.getLoc();
Sean Callanana8702562010-04-24 01:00:16 +0000396
397 if (NextToken.isNot(AsmToken::Eof) &&
398 TargetParser->ParseInstruction(instName, instLoc, operands))
Sean Callananee5dfd42010-02-01 08:49:35 +0000399 ret = -1;
Sean Callanan8f993b82010-04-08 00:48:21 +0000400 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000401 ret = -1;
402 }
403
Sean Callananee5dfd42010-02-01 08:49:35 +0000404 ParserMutex.acquire();
405
406 if (!ret) {
407 GenericAsmLexer->setBuffer(buf);
408
409 while (SpecificAsmLexer->Lex(),
410 SpecificAsmLexer->isNot(AsmToken::Eof) &&
411 SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
412 if (SpecificAsmLexer->is(AsmToken::Error)) {
413 ret = -1;
414 break;
415 }
416 tokens.push_back(SpecificAsmLexer->getTok());
417 }
418 }
419
420 ParserMutex.release();
421
422 return ret;
423}
424
425int EDDisassembler::llvmSyntaxVariant() const {
426 return LLVMSyntaxVariant;
427}