blob: e51276611f5608c821c3e15700c8ce884ab25963 [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
16#include "llvm/ADT/OwningPtr.h"
17#include "llvm/ADT/SmallVector.h"
18#include "llvm/MC/MCAsmInfo.h"
19#include "llvm/MC/MCContext.h"
20#include "llvm/MC/MCDisassembler.h"
21#include "llvm/MC/MCExpr.h"
22#include "llvm/MC/MCInst.h"
23#include "llvm/MC/MCInstPrinter.h"
24#include "llvm/MC/MCStreamer.h"
25#include "llvm/MC/MCParser/AsmLexer.h"
26#include "llvm/MC/MCParser/AsmParser.h"
27#include "llvm/MC/MCParser/MCAsmParser.h"
28#include "llvm/MC/MCParser/MCParsedAsmOperand.h"
29#include "llvm/Support/MemoryBuffer.h"
30#include "llvm/Support/MemoryObject.h"
31#include "llvm/Support/SourceMgr.h"
32#include "llvm/Target/TargetAsmLexer.h"
33#include "llvm/Target/TargetAsmParser.h"
34#include "llvm/Target/TargetRegistry.h"
35#include "llvm/Target/TargetMachine.h"
36#include "llvm/Target/TargetRegisterInfo.h"
37#include "llvm/Target/TargetSelect.h"
38
39#include "EDDisassembler.h"
40#include "EDInst.h"
41
42#include "../../lib/Target/X86/X86GenEDInfo.inc"
43
44using namespace llvm;
45
46bool EDDisassembler::sInitialized = false;
47EDDisassembler::DisassemblerMap_t EDDisassembler::sDisassemblers;
48
49struct InfoMap {
50 Triple::ArchType Arch;
51 const char *String;
52 const InstInfo *Info;
53};
54
55static struct InfoMap infomap[] = {
56 { Triple::x86, "i386-unknown-unknown", instInfoX86 },
57 { Triple::x86_64, "x86_64-unknown-unknown", instInfoX86 },
58 { Triple::InvalidArch, NULL, NULL }
59};
60
61/// infoFromArch - Returns the InfoMap corresponding to a given architecture,
62/// or NULL if there is an error
63///
64/// @arg arch - The Triple::ArchType for the desired architecture
65static const InfoMap *infoFromArch(Triple::ArchType arch) {
66 unsigned int infoIndex;
67
68 for (infoIndex = 0; infomap[infoIndex].String != NULL; ++infoIndex) {
69 if(arch == infomap[infoIndex].Arch)
70 return &infomap[infoIndex];
71 }
72
73 return NULL;
74}
75
76/// getLLVMSyntaxVariant - gets the constant to use to get an assembly printer
77/// for the desired assembly syntax, suitable for passing to
78/// Target::createMCInstPrinter()
79///
80/// @arg arch - The target architecture
81/// @arg syntax - The assembly syntax in sd form
82static int getLLVMSyntaxVariant(Triple::ArchType arch,
83 EDAssemblySyntax_t syntax) {
84 switch (syntax) {
85 default:
86 return -1;
87 // Mappings below from X86AsmPrinter.cpp
88 case kEDAssemblySyntaxX86ATT:
89 if (arch == Triple::x86 || arch == Triple::x86_64)
90 return 0;
91 else
92 return -1;
93 case kEDAssemblySyntaxX86Intel:
94 if (arch == Triple::x86 || arch == Triple::x86_64)
95 return 1;
96 else
97 return -1;
98 }
99}
100
101#define BRINGUP_TARGET(tgt) \
102 LLVMInitialize##tgt##TargetInfo(); \
103 LLVMInitialize##tgt##Target(); \
104 LLVMInitialize##tgt##AsmPrinter(); \
105 LLVMInitialize##tgt##AsmParser(); \
106 LLVMInitialize##tgt##Disassembler();
107
108void EDDisassembler::initialize() {
109 if (sInitialized)
110 return;
111
112 sInitialized = true;
113
114 BRINGUP_TARGET(X86)
115}
116
117#undef BRINGUP_TARGET
118
119EDDisassembler *EDDisassembler::getDisassembler(Triple::ArchType arch,
120 EDAssemblySyntax_t syntax) {
121 CPUKey key;
122 key.Arch = arch;
123 key.Syntax = syntax;
124
125 EDDisassembler::DisassemblerMap_t::iterator i = sDisassemblers.find(key);
126
127 if (i != sDisassemblers.end()) {
128 return i->second;
129 }
130 else {
131 EDDisassembler* sdd = new EDDisassembler(key);
132 if(!sdd->valid()) {
133 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,
146 EDAssemblySyntax_t syntax) {
147 Triple triple(str);
148
149 return getDisassembler(triple.getArch(), syntax);
150}
151
152namespace {
153 class EDAsmParser : public MCAsmParser {
154 AsmLexer Lexer;
155 MCContext Context;
156 OwningPtr<MCStreamer> Streamer;
157 public:
158 // Mandatory functions
159 EDAsmParser(const MCAsmInfo &MAI) : Lexer(MAI) {
160 Streamer.reset(createNullStreamer(Context));
161 }
162 virtual ~EDAsmParser() { }
163 MCAsmLexer &getLexer() { return Lexer; }
164 MCContext &getContext() { return Context; }
165 MCStreamer &getStreamer() { return *Streamer; }
166 void Warning(SMLoc L, const Twine &Msg) { }
167 bool Error(SMLoc L, const Twine &Msg) { return true; }
168 const AsmToken &Lex() { return Lexer.Lex(); }
169 bool ParseExpression(const MCExpr *&Res, SMLoc &EndLoc) {
170 AsmToken token = Lex();
171 if(token.isNot(AsmToken::Integer))
172 return true;
173 Res = MCConstantExpr::Create(token.getIntVal(), Context);
174 return false;
175 }
176 bool ParseParenExpression(const MCExpr *&Res, SMLoc &EndLoc) {
177 assert(0 && "I can't ParseParenExpression()s!");
178 }
179 bool ParseAbsoluteExpression(int64_t &Res) {
180 assert(0 && "I can't ParseAbsoluteExpression()s!");
181 }
182
183 /// setBuffer - loads a buffer into the parser
184 /// @arg buf - The buffer to read tokens from
185 void setBuffer(const MemoryBuffer &buf) { Lexer.setBuffer(&buf); }
186 /// parseInstName - When the lexer is positioned befor an instruction
187 /// name (with possible intervening whitespace), reads past the name,
188 /// returning 0 on success and -1 on failure
189 /// @arg name - A reference to a string that is filled in with the
190 /// instruction name
191 /// @arg loc - A reference to a location that is filled in with the
192 /// position of the instruction name
193 int parseInstName(StringRef &name, SMLoc &loc) {
194 AsmToken tok = Lexer.Lex();
195 if(tok.isNot(AsmToken::Identifier)) {
196 return -1;
197 }
198 name = tok.getString();
199 loc = tok.getLoc();
200 return 0;
201 }
202 };
203}
204
205EDDisassembler::EDDisassembler(CPUKey &key) :
206 Valid(false), ErrorString(), ErrorStream(ErrorString), Key(key) {
207 const InfoMap *infoMap = infoFromArch(key.Arch);
208
209 if (!infoMap)
210 return;
211
212 const char *triple = infoMap->String;
213
214 int syntaxVariant = getLLVMSyntaxVariant(key.Arch, key.Syntax);
215
216 if (syntaxVariant < 0)
217 return;
218
219 std::string tripleString(triple);
220 std::string errorString;
221
222 Tgt = TargetRegistry::lookupTarget(tripleString,
223 errorString);
224
225 if (!Tgt)
226 return;
227
228 std::string featureString;
229
230 OwningPtr<const TargetMachine>
231 targetMachine(Tgt->createTargetMachine(tripleString,
232 featureString));
233
234 const TargetRegisterInfo *registerInfo = targetMachine->getRegisterInfo();
235
236 if (!registerInfo)
237 return;
238
239 AsmInfo.reset(Tgt->createAsmInfo(tripleString));
240
241 if (!AsmInfo)
242 return;
243
244 Disassembler.reset(Tgt->createMCDisassembler());
245
246 if (!Disassembler)
247 return;
248
249 InstString.reset(new std::string);
250 InstStream.reset(new raw_string_ostream(*InstString));
251
252 InstPrinter.reset(Tgt->createMCInstPrinter(syntaxVariant,
253 *AsmInfo,
254 *InstStream));
255
256 if (!InstPrinter)
257 return;
258
259 GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
260 SpecificAsmLexer.reset(Tgt->createAsmLexer(*AsmInfo));
261 SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
262
263 InstInfos = infoMap->Info;
264
265 Valid = true;
266}
267
268EDDisassembler::~EDDisassembler() {
269 if(!valid())
270 return;
271}
272
273namespace {
274 /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
275 /// as provided by the sd interface. See MemoryObject.
276 class EDMemoryObject : public llvm::MemoryObject {
277 private:
278 EDByteReaderCallback Callback;
279 void *Arg;
280 public:
281 EDMemoryObject(EDByteReaderCallback callback,
282 void *arg) : Callback(callback), Arg(arg) { }
283 ~EDMemoryObject() { }
284 uint64_t getBase() const { return 0x0; }
285 uint64_t getExtent() const { return (uint64_t)-1; }
286 int readByte(uint64_t address, uint8_t *ptr) const {
287 if(!Callback)
288 return -1;
289
290 if(Callback(ptr, address, Arg))
291 return -1;
292
293 return 0;
294 }
295 };
296}
297
298EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
299 uint64_t address,
300 void *arg) {
301 EDMemoryObject memoryObject(byteReader, arg);
302
303 MCInst* inst = new MCInst;
304 uint64_t byteSize;
305
306 if (!Disassembler->getInstruction(*inst,
307 byteSize,
308 memoryObject,
309 address,
310 ErrorStream)) {
311 delete inst;
312 return NULL;
313 }
314 else {
315 const InstInfo *thisInstInfo = &InstInfos[inst->getOpcode()];
316
317 EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
318 return sdInst;
319 }
320}
321
322void EDDisassembler::initMaps(const TargetRegisterInfo &registerInfo) {
323 unsigned numRegisters = registerInfo.getNumRegs();
324 unsigned registerIndex;
325
326 for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
327 const char* registerName = registerInfo.get(registerIndex).Name;
328
329 RegVec.push_back(registerName);
330 RegRMap[registerName] = registerIndex;
331 }
332
333 if (Key.Arch == Triple::x86 ||
334 Key.Arch == Triple::x86_64) {
335 stackPointers.insert(registerIDWithName("SP"));
336 stackPointers.insert(registerIDWithName("ESP"));
337 stackPointers.insert(registerIDWithName("RSP"));
338
339 programCounters.insert(registerIDWithName("IP"));
340 programCounters.insert(registerIDWithName("EIP"));
341 programCounters.insert(registerIDWithName("RIP"));
342 }
343}
344
345const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
346 if (registerID >= RegVec.size())
347 return NULL;
348 else
349 return RegVec[registerID].c_str();
350}
351
352unsigned EDDisassembler::registerIDWithName(const char *name) const {
353 regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
354 if (iter == RegRMap.end())
355 return 0;
356 else
357 return (*iter).second;
358}
359
360bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
361 return (stackPointers.find(registerID) != stackPointers.end());
362}
363
364bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
365 return (programCounters.find(registerID) != programCounters.end());
366}
367
368int EDDisassembler::printInst(std::string& str,
369 MCInst& inst) {
370 PrinterMutex.acquire();
371
372 InstPrinter->printInst(&inst);
373 InstStream->flush();
374 str = *InstString;
375 InstString->clear();
376
377 PrinterMutex.release();
378
379 return 0;
380}
381
382int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
383 SmallVectorImpl<AsmToken> &tokens,
384 const std::string &str) {
385 int ret = 0;
386
387 const char *cStr = str.c_str();
388 MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
389
390 StringRef instName;
391 SMLoc instLoc;
392
393 SourceMgr sourceMgr;
394 sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
395 MCContext context;
396 OwningPtr<MCStreamer> streamer
397 (createNullStreamer(context));
398 AsmParser genericParser(sourceMgr, context, *streamer, *AsmInfo);
399 OwningPtr<TargetAsmParser> specificParser
400 (Tgt->createAsmParser(genericParser));
401
402 AsmToken OpcodeToken = genericParser.Lex();
403
404 if(OpcodeToken.is(AsmToken::Identifier)) {
405 instName = OpcodeToken.getString();
406 instLoc = OpcodeToken.getLoc();
407 if (specificParser->ParseInstruction(instName, instLoc, operands))
408 ret = -1;
409 }
410 else {
411 ret = -1;
412 }
413
Sean Callananee5dfd42010-02-01 08:49:35 +0000414 ParserMutex.acquire();
415
416 if (!ret) {
417 GenericAsmLexer->setBuffer(buf);
418
419 while (SpecificAsmLexer->Lex(),
420 SpecificAsmLexer->isNot(AsmToken::Eof) &&
421 SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
422 if (SpecificAsmLexer->is(AsmToken::Error)) {
423 ret = -1;
424 break;
425 }
426 tokens.push_back(SpecificAsmLexer->getTok());
427 }
428 }
429
430 ParserMutex.release();
431
432 return ret;
433}
434
435int EDDisassembler::llvmSyntaxVariant() const {
436 return LLVMSyntaxVariant;
437}