blob: 2e52bec3b92c59f891f925aea0e061801484f97f [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 Callanan9899f702010-04-13 21:21:57 +0000113 InitializeAllAsmPrinters();
114 InitializeAllAsmParsers();
115 InitializeAllDisassemblers();
Sean Callananee5dfd42010-02-01 08:49:35 +0000116}
117
118#undef BRINGUP_TARGET
119
120EDDisassembler *EDDisassembler::getDisassembler(Triple::ArchType arch,
Chris Lattner847da552010-07-20 18:25:19 +0000121 AssemblySyntax syntax) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000122 CPUKey key;
123 key.Arch = arch;
124 key.Syntax = syntax;
125
126 EDDisassembler::DisassemblerMap_t::iterator i = sDisassemblers.find(key);
127
128 if (i != sDisassemblers.end()) {
129 return i->second;
Sean Callanan8f993b82010-04-08 00:48:21 +0000130 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000131 EDDisassembler* sdd = new EDDisassembler(key);
Sean Callanan8f993b82010-04-08 00:48:21 +0000132 if (!sdd->valid()) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000133 delete sdd;
134 return NULL;
135 }
136
137 sDisassemblers[key] = sdd;
138
139 return sdd;
140 }
141
142 return NULL;
143}
144
145EDDisassembler *EDDisassembler::getDisassembler(StringRef str,
Chris Lattner847da552010-07-20 18:25:19 +0000146 AssemblySyntax syntax) {
147 return getDisassembler(Triple(str).getArch(), syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000148}
149
Sean Callananee5dfd42010-02-01 08:49:35 +0000150EDDisassembler::EDDisassembler(CPUKey &key) :
Sean Callanan8f993b82010-04-08 00:48:21 +0000151 Valid(false),
152 HasSemantics(false),
153 ErrorStream(nulls()),
154 Key(key) {
Sean Callanan9899f702010-04-13 21:21:57 +0000155 const char *triple = tripleFromArch(key.Arch);
Sean Callanan8f993b82010-04-08 00:48:21 +0000156
Sean Callanan9899f702010-04-13 21:21:57 +0000157 if (!triple)
158 return;
Sean Callananee5dfd42010-02-01 08:49:35 +0000159
Sean Callanan4285b292010-04-09 00:11:15 +0000160 LLVMSyntaxVariant = getLLVMSyntaxVariant(key.Arch, key.Syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000161
Sean Callanan4285b292010-04-09 00:11:15 +0000162 if (LLVMSyntaxVariant < 0)
Sean Callananee5dfd42010-02-01 08:49:35 +0000163 return;
164
165 std::string tripleString(triple);
166 std::string errorString;
167
168 Tgt = TargetRegistry::lookupTarget(tripleString,
169 errorString);
170
171 if (!Tgt)
172 return;
173
Evan Cheng276365d2011-06-30 01:53:36 +0000174 std::string CPU;
Sean Callananee5dfd42010-02-01 08:49:35 +0000175 std::string featureString;
Evan Cheng276365d2011-06-30 01:53:36 +0000176 TargetMachine.reset(Tgt->createTargetMachine(tripleString, CPU,
Daniel Dunbar02612432010-07-19 00:33:43 +0000177 featureString));
Evan Chengebdeeab2011-07-08 01:53:10 +0000178
Daniel Dunbar02612432010-07-19 00:33:43 +0000179 const TargetRegisterInfo *registerInfo = TargetMachine->getRegisterInfo();
Sean Callananee5dfd42010-02-01 08:49:35 +0000180
181 if (!registerInfo)
182 return;
Sean Callanan8f993b82010-04-08 00:48:21 +0000183
184 initMaps(*registerInfo);
Sean Callananee5dfd42010-02-01 08:49:35 +0000185
Evan Cheng1abf2cb2011-07-14 23:50:31 +0000186 AsmInfo.reset(Tgt->createMCAsmInfo(tripleString));
Sean Callananee5dfd42010-02-01 08:49:35 +0000187
188 if (!AsmInfo)
189 return;
Evan Chengebdeeab2011-07-08 01:53:10 +0000190
Evan Cheng0e6a0522011-07-18 20:57:22 +0000191 MRI.reset(Tgt->createMCRegInfo(tripleString));
192
193 if (!MRI)
194 return;
195
Sean Callananee5dfd42010-02-01 08:49:35 +0000196 Disassembler.reset(Tgt->createMCDisassembler());
197
198 if (!Disassembler)
199 return;
Sean Callanan9899f702010-04-13 21:21:57 +0000200
201 InstInfos = Disassembler->getEDInfo();
Sean Callananee5dfd42010-02-01 08:49:35 +0000202
203 InstString.reset(new std::string);
204 InstStream.reset(new raw_string_ostream(*InstString));
Evan Chengb2627992011-07-06 19:45:42 +0000205 InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo));
Sean Callananee5dfd42010-02-01 08:49:35 +0000206
207 if (!InstPrinter)
208 return;
209
210 GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
211 SpecificAsmLexer.reset(Tgt->createAsmLexer(*AsmInfo));
212 SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
Sean Callanand74667e2010-02-02 02:18:20 +0000213
Daniel Dunbar02612432010-07-19 00:33:43 +0000214 initMaps(*TargetMachine->getRegisterInfo());
Sean Callananee5dfd42010-02-01 08:49:35 +0000215
216 Valid = true;
217}
218
219EDDisassembler::~EDDisassembler() {
Sean Callanan8f993b82010-04-08 00:48:21 +0000220 if (!valid())
Sean Callananee5dfd42010-02-01 08:49:35 +0000221 return;
222}
223
224namespace {
225 /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
226 /// as provided by the sd interface. See MemoryObject.
227 class EDMemoryObject : public llvm::MemoryObject {
228 private:
229 EDByteReaderCallback Callback;
230 void *Arg;
231 public:
232 EDMemoryObject(EDByteReaderCallback callback,
233 void *arg) : Callback(callback), Arg(arg) { }
234 ~EDMemoryObject() { }
235 uint64_t getBase() const { return 0x0; }
236 uint64_t getExtent() const { return (uint64_t)-1; }
237 int readByte(uint64_t address, uint8_t *ptr) const {
Sean Callanan8f993b82010-04-08 00:48:21 +0000238 if (!Callback)
Sean Callananee5dfd42010-02-01 08:49:35 +0000239 return -1;
240
Sean Callanan8f993b82010-04-08 00:48:21 +0000241 if (Callback(ptr, address, Arg))
Sean Callananee5dfd42010-02-01 08:49:35 +0000242 return -1;
243
244 return 0;
245 }
246 };
247}
248
249EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
250 uint64_t address,
251 void *arg) {
252 EDMemoryObject memoryObject(byteReader, arg);
253
254 MCInst* inst = new MCInst;
255 uint64_t byteSize;
256
257 if (!Disassembler->getInstruction(*inst,
258 byteSize,
259 memoryObject,
260 address,
261 ErrorStream)) {
262 delete inst;
263 return NULL;
Sean Callanan8f993b82010-04-08 00:48:21 +0000264 } else {
Shantonu Sen0da9f132011-04-05 18:40:45 +0000265 const llvm::EDInstInfo *thisInstInfo = NULL;
Sean Callanan8f993b82010-04-08 00:48:21 +0000266
Shantonu Sen0da9f132011-04-05 18:40:45 +0000267 if (InstInfos) {
268 thisInstInfo = &InstInfos[inst->getOpcode()];
269 }
Sean Callananee5dfd42010-02-01 08:49:35 +0000270
271 EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
272 return sdInst;
273 }
274}
275
276void EDDisassembler::initMaps(const TargetRegisterInfo &registerInfo) {
277 unsigned numRegisters = registerInfo.getNumRegs();
278 unsigned registerIndex;
279
280 for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
281 const char* registerName = registerInfo.get(registerIndex).Name;
282
283 RegVec.push_back(registerName);
284 RegRMap[registerName] = registerIndex;
285 }
286
Sean Callanan8f993b82010-04-08 00:48:21 +0000287 switch (Key.Arch) {
288 default:
289 break;
290 case Triple::x86:
291 case Triple::x86_64:
Sean Callananee5dfd42010-02-01 08:49:35 +0000292 stackPointers.insert(registerIDWithName("SP"));
293 stackPointers.insert(registerIDWithName("ESP"));
294 stackPointers.insert(registerIDWithName("RSP"));
295
296 programCounters.insert(registerIDWithName("IP"));
297 programCounters.insert(registerIDWithName("EIP"));
298 programCounters.insert(registerIDWithName("RIP"));
Sean Callanan8f993b82010-04-08 00:48:21 +0000299 break;
300 case Triple::arm:
301 case Triple::thumb:
302 stackPointers.insert(registerIDWithName("SP"));
303
304 programCounters.insert(registerIDWithName("PC"));
305 break;
Sean Callananee5dfd42010-02-01 08:49:35 +0000306 }
307}
308
309const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
310 if (registerID >= RegVec.size())
311 return NULL;
312 else
313 return RegVec[registerID].c_str();
314}
315
316unsigned EDDisassembler::registerIDWithName(const char *name) const {
317 regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
318 if (iter == RegRMap.end())
319 return 0;
320 else
321 return (*iter).second;
322}
323
324bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
325 return (stackPointers.find(registerID) != stackPointers.end());
326}
327
328bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
329 return (programCounters.find(registerID) != programCounters.end());
330}
331
Chris Lattnerd3740872010-04-04 05:04:31 +0000332int EDDisassembler::printInst(std::string &str, MCInst &inst) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000333 PrinterMutex.acquire();
334
Chris Lattnerd3740872010-04-04 05:04:31 +0000335 InstPrinter->printInst(&inst, *InstStream);
Sean Callananee5dfd42010-02-01 08:49:35 +0000336 InstStream->flush();
337 str = *InstString;
338 InstString->clear();
339
340 PrinterMutex.release();
341
342 return 0;
343}
344
Sean Callanan457856c2011-04-09 00:37:25 +0000345static void diag_handler(const SMDiagnostic &diag,
346 void *context)
347{
348 if (context) {
349 EDDisassembler *disassembler = static_cast<EDDisassembler*>(context);
350 diag.Print("", disassembler->ErrorStream);
351 }
352}
353
Sean Callananee5dfd42010-02-01 08:49:35 +0000354int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
355 SmallVectorImpl<AsmToken> &tokens,
356 const std::string &str) {
357 int ret = 0;
358
Sean Callanan8f993b82010-04-08 00:48:21 +0000359 switch (Key.Arch) {
360 default:
361 return -1;
362 case Triple::x86:
363 case Triple::x86_64:
364 case Triple::arm:
365 case Triple::thumb:
366 break;
367 }
368
Sean Callananee5dfd42010-02-01 08:49:35 +0000369 const char *cStr = str.c_str();
370 MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
371
372 StringRef instName;
373 SMLoc instLoc;
374
375 SourceMgr sourceMgr;
Sean Callanan457856c2011-04-09 00:37:25 +0000376 sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this));
Sean Callananee5dfd42010-02-01 08:49:35 +0000377 sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
Evan Cheng0e6a0522011-07-18 20:57:22 +0000378 MCContext context(*AsmInfo, *MRI, NULL);
Chris Lattnerc18409a2010-03-11 22:53:35 +0000379 OwningPtr<MCStreamer> streamer(createNullStreamer(context));
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000380 OwningPtr<MCAsmParser> genericParser(createMCAsmParser(*Tgt, sourceMgr,
381 context, *streamer,
382 *AsmInfo));
Evan Chengebdeeab2011-07-08 01:53:10 +0000383
384 StringRef triple = tripleFromArch(Key.Arch);
Evan Chengffc0e732011-07-09 05:47:46 +0000385 OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(triple, "", ""));
386 OwningPtr<TargetAsmParser> TargetParser(Tgt->createAsmParser(*STI,
Evan Chengebdeeab2011-07-08 01:53:10 +0000387 *genericParser));
Sean Callananee5dfd42010-02-01 08:49:35 +0000388
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000389 AsmToken OpcodeToken = genericParser->Lex();
390 AsmToken NextToken = genericParser->Lex(); // consume next token, because specificParser expects us to
Sean Callanana8702562010-04-24 01:00:16 +0000391
Sean Callanan8f993b82010-04-08 00:48:21 +0000392 if (OpcodeToken.is(AsmToken::Identifier)) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000393 instName = OpcodeToken.getString();
394 instLoc = OpcodeToken.getLoc();
Sean Callanana8702562010-04-24 01:00:16 +0000395
396 if (NextToken.isNot(AsmToken::Eof) &&
397 TargetParser->ParseInstruction(instName, instLoc, operands))
Sean Callananee5dfd42010-02-01 08:49:35 +0000398 ret = -1;
Sean Callanan8f993b82010-04-08 00:48:21 +0000399 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000400 ret = -1;
401 }
402
Sean Callananee5dfd42010-02-01 08:49:35 +0000403 ParserMutex.acquire();
404
405 if (!ret) {
406 GenericAsmLexer->setBuffer(buf);
407
408 while (SpecificAsmLexer->Lex(),
409 SpecificAsmLexer->isNot(AsmToken::Eof) &&
410 SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
411 if (SpecificAsmLexer->is(AsmToken::Error)) {
412 ret = -1;
413 break;
414 }
415 tokens.push_back(SpecificAsmLexer->getTok());
416 }
417 }
418
419 ParserMutex.release();
420
421 return ret;
422}
423
424int EDDisassembler::llvmSyntaxVariant() const {
425 return LLVMSyntaxVariant;
426}