blob: cae54164ed9385aae9f5dd44a926fd85f1dcf99b [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 Cheng43966132011-07-19 06:37:02 +0000111 InitializeAllMCCodeGenInfos();
Evan Cheng1abf2cb2011-07-14 23:50:31 +0000112 InitializeAllMCAsmInfos();
Evan Cheng0e6a0522011-07-18 20:57:22 +0000113 InitializeAllMCRegisterInfos();
Sean Callananc872143d2011-07-19 00:00:41 +0000114 InitializeAllMCSubtargetInfos();
Sean Callanan9899f702010-04-13 21:21:57 +0000115 InitializeAllAsmPrinters();
116 InitializeAllAsmParsers();
117 InitializeAllDisassemblers();
Sean Callananee5dfd42010-02-01 08:49:35 +0000118}
119
120#undef BRINGUP_TARGET
121
122EDDisassembler *EDDisassembler::getDisassembler(Triple::ArchType arch,
Chris Lattner847da552010-07-20 18:25:19 +0000123 AssemblySyntax syntax) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000124 CPUKey key;
125 key.Arch = arch;
126 key.Syntax = syntax;
127
128 EDDisassembler::DisassemblerMap_t::iterator i = sDisassemblers.find(key);
129
130 if (i != sDisassemblers.end()) {
131 return i->second;
Sean Callanan8f993b82010-04-08 00:48:21 +0000132 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000133 EDDisassembler* sdd = new EDDisassembler(key);
Sean Callanan8f993b82010-04-08 00:48:21 +0000134 if (!sdd->valid()) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000135 delete sdd;
136 return NULL;
137 }
138
139 sDisassemblers[key] = sdd;
140
141 return sdd;
142 }
143
144 return NULL;
145}
146
147EDDisassembler *EDDisassembler::getDisassembler(StringRef str,
Chris Lattner847da552010-07-20 18:25:19 +0000148 AssemblySyntax syntax) {
149 return getDisassembler(Triple(str).getArch(), syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000150}
151
Sean Callananee5dfd42010-02-01 08:49:35 +0000152EDDisassembler::EDDisassembler(CPUKey &key) :
Sean Callanan8f993b82010-04-08 00:48:21 +0000153 Valid(false),
154 HasSemantics(false),
155 ErrorStream(nulls()),
156 Key(key) {
Sean Callanan9899f702010-04-13 21:21:57 +0000157 const char *triple = tripleFromArch(key.Arch);
Sean Callanan8f993b82010-04-08 00:48:21 +0000158
Sean Callanan9899f702010-04-13 21:21:57 +0000159 if (!triple)
160 return;
Sean Callananee5dfd42010-02-01 08:49:35 +0000161
Sean Callanan4285b292010-04-09 00:11:15 +0000162 LLVMSyntaxVariant = getLLVMSyntaxVariant(key.Arch, key.Syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000163
Sean Callanan4285b292010-04-09 00:11:15 +0000164 if (LLVMSyntaxVariant < 0)
Sean Callananee5dfd42010-02-01 08:49:35 +0000165 return;
166
167 std::string tripleString(triple);
168 std::string errorString;
169
170 Tgt = TargetRegistry::lookupTarget(tripleString,
171 errorString);
172
173 if (!Tgt)
174 return;
175
Evan Cheng276365d2011-06-30 01:53:36 +0000176 std::string CPU;
Sean Callananee5dfd42010-02-01 08:49:35 +0000177 std::string featureString;
Evan Cheng276365d2011-06-30 01:53:36 +0000178 TargetMachine.reset(Tgt->createTargetMachine(tripleString, CPU,
Daniel Dunbar02612432010-07-19 00:33:43 +0000179 featureString));
Evan Chengebdeeab2011-07-08 01:53:10 +0000180
Daniel Dunbar02612432010-07-19 00:33:43 +0000181 const TargetRegisterInfo *registerInfo = TargetMachine->getRegisterInfo();
Sean Callananee5dfd42010-02-01 08:49:35 +0000182
183 if (!registerInfo)
184 return;
Sean Callanan8f993b82010-04-08 00:48:21 +0000185
186 initMaps(*registerInfo);
Sean Callananee5dfd42010-02-01 08:49:35 +0000187
Evan Cheng1abf2cb2011-07-14 23:50:31 +0000188 AsmInfo.reset(Tgt->createMCAsmInfo(tripleString));
Sean Callananee5dfd42010-02-01 08:49:35 +0000189
190 if (!AsmInfo)
191 return;
Evan Chengebdeeab2011-07-08 01:53:10 +0000192
Evan Cheng0e6a0522011-07-18 20:57:22 +0000193 MRI.reset(Tgt->createMCRegInfo(tripleString));
194
195 if (!MRI)
196 return;
197
Sean Callananee5dfd42010-02-01 08:49:35 +0000198 Disassembler.reset(Tgt->createMCDisassembler());
199
200 if (!Disassembler)
201 return;
Sean Callanan9899f702010-04-13 21:21:57 +0000202
203 InstInfos = Disassembler->getEDInfo();
Sean Callananee5dfd42010-02-01 08:49:35 +0000204
205 InstString.reset(new std::string);
206 InstStream.reset(new raw_string_ostream(*InstString));
Evan Chengb2627992011-07-06 19:45:42 +0000207 InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo));
Sean Callananee5dfd42010-02-01 08:49:35 +0000208
209 if (!InstPrinter)
210 return;
211
212 GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
213 SpecificAsmLexer.reset(Tgt->createAsmLexer(*AsmInfo));
214 SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
Sean Callanand74667e2010-02-02 02:18:20 +0000215
Daniel Dunbar02612432010-07-19 00:33:43 +0000216 initMaps(*TargetMachine->getRegisterInfo());
Sean Callananee5dfd42010-02-01 08:49:35 +0000217
218 Valid = true;
219}
220
221EDDisassembler::~EDDisassembler() {
Sean Callanan8f993b82010-04-08 00:48:21 +0000222 if (!valid())
Sean Callananee5dfd42010-02-01 08:49:35 +0000223 return;
224}
225
226namespace {
227 /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
228 /// as provided by the sd interface. See MemoryObject.
229 class EDMemoryObject : public llvm::MemoryObject {
230 private:
231 EDByteReaderCallback Callback;
232 void *Arg;
233 public:
234 EDMemoryObject(EDByteReaderCallback callback,
235 void *arg) : Callback(callback), Arg(arg) { }
236 ~EDMemoryObject() { }
237 uint64_t getBase() const { return 0x0; }
238 uint64_t getExtent() const { return (uint64_t)-1; }
239 int readByte(uint64_t address, uint8_t *ptr) const {
Sean Callanan8f993b82010-04-08 00:48:21 +0000240 if (!Callback)
Sean Callananee5dfd42010-02-01 08:49:35 +0000241 return -1;
242
Sean Callanan8f993b82010-04-08 00:48:21 +0000243 if (Callback(ptr, address, Arg))
Sean Callananee5dfd42010-02-01 08:49:35 +0000244 return -1;
245
246 return 0;
247 }
248 };
249}
250
251EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
252 uint64_t address,
253 void *arg) {
254 EDMemoryObject memoryObject(byteReader, arg);
255
256 MCInst* inst = new MCInst;
257 uint64_t byteSize;
258
259 if (!Disassembler->getInstruction(*inst,
260 byteSize,
261 memoryObject,
262 address,
263 ErrorStream)) {
264 delete inst;
265 return NULL;
Sean Callanan8f993b82010-04-08 00:48:21 +0000266 } else {
Shantonu Sen0da9f132011-04-05 18:40:45 +0000267 const llvm::EDInstInfo *thisInstInfo = NULL;
Sean Callanan8f993b82010-04-08 00:48:21 +0000268
Shantonu Sen0da9f132011-04-05 18:40:45 +0000269 if (InstInfos) {
270 thisInstInfo = &InstInfos[inst->getOpcode()];
271 }
Sean Callananee5dfd42010-02-01 08:49:35 +0000272
273 EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
274 return sdInst;
275 }
276}
277
278void EDDisassembler::initMaps(const TargetRegisterInfo &registerInfo) {
279 unsigned numRegisters = registerInfo.getNumRegs();
280 unsigned registerIndex;
281
282 for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
283 const char* registerName = registerInfo.get(registerIndex).Name;
284
285 RegVec.push_back(registerName);
286 RegRMap[registerName] = registerIndex;
287 }
288
Sean Callanan8f993b82010-04-08 00:48:21 +0000289 switch (Key.Arch) {
290 default:
291 break;
292 case Triple::x86:
293 case Triple::x86_64:
Sean Callananee5dfd42010-02-01 08:49:35 +0000294 stackPointers.insert(registerIDWithName("SP"));
295 stackPointers.insert(registerIDWithName("ESP"));
296 stackPointers.insert(registerIDWithName("RSP"));
297
298 programCounters.insert(registerIDWithName("IP"));
299 programCounters.insert(registerIDWithName("EIP"));
300 programCounters.insert(registerIDWithName("RIP"));
Sean Callanan8f993b82010-04-08 00:48:21 +0000301 break;
302 case Triple::arm:
303 case Triple::thumb:
304 stackPointers.insert(registerIDWithName("SP"));
305
306 programCounters.insert(registerIDWithName("PC"));
307 break;
Sean Callananee5dfd42010-02-01 08:49:35 +0000308 }
309}
310
311const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
312 if (registerID >= RegVec.size())
313 return NULL;
314 else
315 return RegVec[registerID].c_str();
316}
317
318unsigned EDDisassembler::registerIDWithName(const char *name) const {
319 regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
320 if (iter == RegRMap.end())
321 return 0;
322 else
323 return (*iter).second;
324}
325
326bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
327 return (stackPointers.find(registerID) != stackPointers.end());
328}
329
330bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
331 return (programCounters.find(registerID) != programCounters.end());
332}
333
Chris Lattnerd3740872010-04-04 05:04:31 +0000334int EDDisassembler::printInst(std::string &str, MCInst &inst) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000335 PrinterMutex.acquire();
336
Chris Lattnerd3740872010-04-04 05:04:31 +0000337 InstPrinter->printInst(&inst, *InstStream);
Sean Callananee5dfd42010-02-01 08:49:35 +0000338 InstStream->flush();
339 str = *InstString;
340 InstString->clear();
341
342 PrinterMutex.release();
343
344 return 0;
345}
346
Sean Callanan457856c2011-04-09 00:37:25 +0000347static void diag_handler(const SMDiagnostic &diag,
348 void *context)
349{
350 if (context) {
351 EDDisassembler *disassembler = static_cast<EDDisassembler*>(context);
352 diag.Print("", disassembler->ErrorStream);
353 }
354}
355
Sean Callananee5dfd42010-02-01 08:49:35 +0000356int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
357 SmallVectorImpl<AsmToken> &tokens,
358 const std::string &str) {
359 int ret = 0;
360
Sean Callanan8f993b82010-04-08 00:48:21 +0000361 switch (Key.Arch) {
362 default:
363 return -1;
364 case Triple::x86:
365 case Triple::x86_64:
366 case Triple::arm:
367 case Triple::thumb:
368 break;
369 }
370
Sean Callananee5dfd42010-02-01 08:49:35 +0000371 const char *cStr = str.c_str();
372 MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
373
374 StringRef instName;
375 SMLoc instLoc;
376
377 SourceMgr sourceMgr;
Sean Callanan457856c2011-04-09 00:37:25 +0000378 sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this));
Sean Callananee5dfd42010-02-01 08:49:35 +0000379 sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
Evan Cheng203576a2011-07-20 19:50:42 +0000380 MCContext context(*AsmInfo, *MRI, NULL);
Chris Lattnerc18409a2010-03-11 22:53:35 +0000381 OwningPtr<MCStreamer> streamer(createNullStreamer(context));
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000382 OwningPtr<MCAsmParser> genericParser(createMCAsmParser(*Tgt, sourceMgr,
383 context, *streamer,
384 *AsmInfo));
Evan Chengebdeeab2011-07-08 01:53:10 +0000385
386 StringRef triple = tripleFromArch(Key.Arch);
Evan Chengffc0e732011-07-09 05:47:46 +0000387 OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(triple, "", ""));
388 OwningPtr<TargetAsmParser> TargetParser(Tgt->createAsmParser(*STI,
Evan Chengebdeeab2011-07-08 01:53:10 +0000389 *genericParser));
Sean Callananee5dfd42010-02-01 08:49:35 +0000390
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000391 AsmToken OpcodeToken = genericParser->Lex();
392 AsmToken NextToken = genericParser->Lex(); // consume next token, because specificParser expects us to
Sean Callanana8702562010-04-24 01:00:16 +0000393
Sean Callanan8f993b82010-04-08 00:48:21 +0000394 if (OpcodeToken.is(AsmToken::Identifier)) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000395 instName = OpcodeToken.getString();
396 instLoc = OpcodeToken.getLoc();
Sean Callanana8702562010-04-24 01:00:16 +0000397
398 if (NextToken.isNot(AsmToken::Eof) &&
399 TargetParser->ParseInstruction(instName, instLoc, operands))
Sean Callananee5dfd42010-02-01 08:49:35 +0000400 ret = -1;
Sean Callanan8f993b82010-04-08 00:48:21 +0000401 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000402 ret = -1;
403 }
404
Sean Callananee5dfd42010-02-01 08:49:35 +0000405 ParserMutex.acquire();
406
407 if (!ret) {
408 GenericAsmLexer->setBuffer(buf);
409
410 while (SpecificAsmLexer->Lex(),
411 SpecificAsmLexer->isNot(AsmToken::Eof) &&
412 SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
413 if (SpecificAsmLexer->is(AsmToken::Error)) {
414 ret = -1;
415 break;
416 }
417 tokens.push_back(SpecificAsmLexer->getTok());
418 }
419 }
420
421 ParserMutex.release();
422
423 return ret;
424}
425
426int EDDisassembler::llvmSyntaxVariant() const {
427 return LLVMSyntaxVariant;
428}