blob: 227e229f09d9950acc5c241361b7c369d66a750d [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"
Evan Cheng94b95502011-07-26 00:24:13 +000031#include "llvm/MC/MCTargetAsmLexer.h"
32#include "llvm/MC/MCTargetAsmParser.h"
Sean Callananee5dfd42010-02-01 08:49:35 +000033#include "llvm/Support/MemoryBuffer.h"
34#include "llvm/Support/MemoryObject.h"
35#include "llvm/Support/SourceMgr.h"
Evan Cheng3e74d6f2011-08-24 18:08:43 +000036#include "llvm/Support/TargetRegistry.h"
37#include "llvm/Support/TargetSelect.h"
Sean Callananee5dfd42010-02-01 08:49:35 +000038using namespace llvm;
39
40bool EDDisassembler::sInitialized = false;
41EDDisassembler::DisassemblerMap_t EDDisassembler::sDisassemblers;
42
Sean Callanan9899f702010-04-13 21:21:57 +000043struct TripleMap {
Sean Callananee5dfd42010-02-01 08:49:35 +000044 Triple::ArchType Arch;
45 const char *String;
Sean Callananee5dfd42010-02-01 08:49:35 +000046};
47
Sean Callanan9899f702010-04-13 21:21:57 +000048static struct TripleMap triplemap[] = {
49 { Triple::x86, "i386-unknown-unknown" },
50 { Triple::x86_64, "x86_64-unknown-unknown" },
51 { Triple::arm, "arm-unknown-unknown" },
52 { Triple::thumb, "thumb-unknown-unknown" },
53 { Triple::InvalidArch, NULL, }
Sean Callananee5dfd42010-02-01 08:49:35 +000054};
55
Sean Callanan9899f702010-04-13 21:21:57 +000056/// infoFromArch - Returns the TripleMap corresponding to a given architecture,
Sean Callananee5dfd42010-02-01 08:49:35 +000057/// or NULL if there is an error
58///
59/// @arg arch - The Triple::ArchType for the desired architecture
Sean Callanan9899f702010-04-13 21:21:57 +000060static const char *tripleFromArch(Triple::ArchType arch) {
Sean Callananee5dfd42010-02-01 08:49:35 +000061 unsigned int infoIndex;
62
Sean Callanan9899f702010-04-13 21:21:57 +000063 for (infoIndex = 0; triplemap[infoIndex].String != NULL; ++infoIndex) {
64 if (arch == triplemap[infoIndex].Arch)
65 return triplemap[infoIndex].String;
Sean Callananee5dfd42010-02-01 08:49:35 +000066 }
67
68 return NULL;
69}
70
71/// getLLVMSyntaxVariant - gets the constant to use to get an assembly printer
72/// for the desired assembly syntax, suitable for passing to
73/// Target::createMCInstPrinter()
74///
75/// @arg arch - The target architecture
76/// @arg syntax - The assembly syntax in sd form
77static int getLLVMSyntaxVariant(Triple::ArchType arch,
Chris Lattner847da552010-07-20 18:25:19 +000078 EDDisassembler::AssemblySyntax syntax) {
Sean Callananee5dfd42010-02-01 08:49:35 +000079 switch (syntax) {
80 default:
81 return -1;
82 // Mappings below from X86AsmPrinter.cpp
Chris Lattner847da552010-07-20 18:25:19 +000083 case EDDisassembler::kEDAssemblySyntaxX86ATT:
Sean Callananee5dfd42010-02-01 08:49:35 +000084 if (arch == Triple::x86 || arch == Triple::x86_64)
85 return 0;
86 else
87 return -1;
Chris Lattner847da552010-07-20 18:25:19 +000088 case EDDisassembler::kEDAssemblySyntaxX86Intel:
Sean Callananee5dfd42010-02-01 08:49:35 +000089 if (arch == Triple::x86 || arch == Triple::x86_64)
90 return 1;
91 else
92 return -1;
Chris Lattner847da552010-07-20 18:25:19 +000093 case EDDisassembler::kEDAssemblySyntaxARMUAL:
Sean Callanan8f993b82010-04-08 00:48:21 +000094 if (arch == Triple::arm || arch == Triple::thumb)
95 return 0;
96 else
97 return -1;
Sean Callananee5dfd42010-02-01 08:49:35 +000098 }
99}
100
Sean Callananee5dfd42010-02-01 08:49:35 +0000101void EDDisassembler::initialize() {
102 if (sInitialized)
103 return;
104
105 sInitialized = true;
106
Sean Callanan9899f702010-04-13 21:21:57 +0000107 InitializeAllTargetInfos();
Evan Chenge78085a2011-07-22 21:58:54 +0000108 InitializeAllTargetMCs();
Sean Callanan9899f702010-04-13 21:21:57 +0000109 InitializeAllAsmParsers();
110 InitializeAllDisassemblers();
Sean Callananee5dfd42010-02-01 08:49:35 +0000111}
112
113#undef BRINGUP_TARGET
114
115EDDisassembler *EDDisassembler::getDisassembler(Triple::ArchType arch,
Chris Lattner847da552010-07-20 18:25:19 +0000116 AssemblySyntax syntax) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000117 CPUKey key;
118 key.Arch = arch;
119 key.Syntax = syntax;
120
121 EDDisassembler::DisassemblerMap_t::iterator i = sDisassemblers.find(key);
122
123 if (i != sDisassemblers.end()) {
124 return i->second;
Sean Callanan8f993b82010-04-08 00:48:21 +0000125 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000126 EDDisassembler* sdd = new EDDisassembler(key);
Sean Callanan8f993b82010-04-08 00:48:21 +0000127 if (!sdd->valid()) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000128 delete sdd;
129 return NULL;
130 }
131
132 sDisassemblers[key] = sdd;
133
134 return sdd;
135 }
136
137 return NULL;
138}
139
140EDDisassembler *EDDisassembler::getDisassembler(StringRef str,
Chris Lattner847da552010-07-20 18:25:19 +0000141 AssemblySyntax syntax) {
142 return getDisassembler(Triple(str).getArch(), syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000143}
144
Sean Callananee5dfd42010-02-01 08:49:35 +0000145EDDisassembler::EDDisassembler(CPUKey &key) :
Sean Callanan8f993b82010-04-08 00:48:21 +0000146 Valid(false),
147 HasSemantics(false),
148 ErrorStream(nulls()),
149 Key(key) {
Sean Callanan9899f702010-04-13 21:21:57 +0000150 const char *triple = tripleFromArch(key.Arch);
Sean Callanan8f993b82010-04-08 00:48:21 +0000151
Sean Callanan9899f702010-04-13 21:21:57 +0000152 if (!triple)
153 return;
Sean Callananee5dfd42010-02-01 08:49:35 +0000154
Sean Callanan4285b292010-04-09 00:11:15 +0000155 LLVMSyntaxVariant = getLLVMSyntaxVariant(key.Arch, key.Syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000156
Sean Callanan4285b292010-04-09 00:11:15 +0000157 if (LLVMSyntaxVariant < 0)
Sean Callananee5dfd42010-02-01 08:49:35 +0000158 return;
159
160 std::string tripleString(triple);
161 std::string errorString;
162
163 Tgt = TargetRegistry::lookupTarget(tripleString,
164 errorString);
165
166 if (!Tgt)
167 return;
168
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000169 MRI.reset(Tgt->createMCRegInfo(tripleString));
Evan Chengebdeeab2011-07-08 01:53:10 +0000170
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000171 if (!MRI)
Sean Callananee5dfd42010-02-01 08:49:35 +0000172 return;
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000173
174 initMaps(*MRI);
Sean Callananee5dfd42010-02-01 08:49:35 +0000175
Evan Cheng1abf2cb2011-07-14 23:50:31 +0000176 AsmInfo.reset(Tgt->createMCAsmInfo(tripleString));
Sean Callananee5dfd42010-02-01 08:49:35 +0000177
178 if (!AsmInfo)
179 return;
Evan Chengebdeeab2011-07-08 01:53:10 +0000180
Sean Callananee5dfd42010-02-01 08:49:35 +0000181 Disassembler.reset(Tgt->createMCDisassembler());
182
183 if (!Disassembler)
184 return;
Sean Callanan9899f702010-04-13 21:21:57 +0000185
186 InstInfos = Disassembler->getEDInfo();
Sean Callananee5dfd42010-02-01 08:49:35 +0000187
188 InstString.reset(new std::string);
189 InstStream.reset(new raw_string_ostream(*InstString));
Evan Chengb2627992011-07-06 19:45:42 +0000190 InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo));
Sean Callananee5dfd42010-02-01 08:49:35 +0000191
192 if (!InstPrinter)
193 return;
194
195 GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
Evan Cheng94b95502011-07-26 00:24:13 +0000196 SpecificAsmLexer.reset(Tgt->createMCAsmLexer(*MRI, *AsmInfo));
Sean Callananee5dfd42010-02-01 08:49:35 +0000197 SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
Sean Callanand74667e2010-02-02 02:18:20 +0000198
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000199 initMaps(*MRI);
Sean Callananee5dfd42010-02-01 08:49:35 +0000200
201 Valid = true;
202}
203
204EDDisassembler::~EDDisassembler() {
Sean Callanan8f993b82010-04-08 00:48:21 +0000205 if (!valid())
Sean Callananee5dfd42010-02-01 08:49:35 +0000206 return;
207}
208
209namespace {
210 /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
211 /// as provided by the sd interface. See MemoryObject.
212 class EDMemoryObject : public llvm::MemoryObject {
213 private:
214 EDByteReaderCallback Callback;
215 void *Arg;
216 public:
217 EDMemoryObject(EDByteReaderCallback callback,
218 void *arg) : Callback(callback), Arg(arg) { }
219 ~EDMemoryObject() { }
220 uint64_t getBase() const { return 0x0; }
221 uint64_t getExtent() const { return (uint64_t)-1; }
222 int readByte(uint64_t address, uint8_t *ptr) const {
Sean Callanan8f993b82010-04-08 00:48:21 +0000223 if (!Callback)
Sean Callananee5dfd42010-02-01 08:49:35 +0000224 return -1;
225
Sean Callanan8f993b82010-04-08 00:48:21 +0000226 if (Callback(ptr, address, Arg))
Sean Callananee5dfd42010-02-01 08:49:35 +0000227 return -1;
228
229 return 0;
230 }
231 };
232}
233
234EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
235 uint64_t address,
236 void *arg) {
237 EDMemoryObject memoryObject(byteReader, arg);
238
239 MCInst* inst = new MCInst;
240 uint64_t byteSize;
241
James Molloyc047dca2011-09-01 18:02:14 +0000242 MCDisassembler::DecodeStatus S;
243 S = Disassembler->getInstruction(*inst,
244 byteSize,
245 memoryObject,
246 address,
247 ErrorStream);
248 switch (S) {
249 case MCDisassembler::Fail:
250 case MCDisassembler::SoftFail:
251 // FIXME: Do something different on soft failure mode?
Sean Callananee5dfd42010-02-01 08:49:35 +0000252 delete inst;
253 return NULL;
James Molloyc047dca2011-09-01 18:02:14 +0000254 case MCDisassembler::Success: {
Shantonu Sen0da9f132011-04-05 18:40:45 +0000255 const llvm::EDInstInfo *thisInstInfo = NULL;
Sean Callanan8f993b82010-04-08 00:48:21 +0000256
Shantonu Sen0da9f132011-04-05 18:40:45 +0000257 if (InstInfos) {
258 thisInstInfo = &InstInfos[inst->getOpcode()];
259 }
Sean Callananee5dfd42010-02-01 08:49:35 +0000260
261 EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
262 return sdInst;
263 }
James Molloyc047dca2011-09-01 18:02:14 +0000264 }
Sean Callananee5dfd42010-02-01 08:49:35 +0000265}
266
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000267void EDDisassembler::initMaps(const MCRegisterInfo &registerInfo) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000268 unsigned numRegisters = registerInfo.getNumRegs();
269 unsigned registerIndex;
270
271 for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
272 const char* registerName = registerInfo.get(registerIndex).Name;
273
274 RegVec.push_back(registerName);
275 RegRMap[registerName] = registerIndex;
276 }
277
Sean Callanan8f993b82010-04-08 00:48:21 +0000278 switch (Key.Arch) {
279 default:
280 break;
281 case Triple::x86:
282 case Triple::x86_64:
Sean Callananee5dfd42010-02-01 08:49:35 +0000283 stackPointers.insert(registerIDWithName("SP"));
284 stackPointers.insert(registerIDWithName("ESP"));
285 stackPointers.insert(registerIDWithName("RSP"));
286
287 programCounters.insert(registerIDWithName("IP"));
288 programCounters.insert(registerIDWithName("EIP"));
289 programCounters.insert(registerIDWithName("RIP"));
Sean Callanan8f993b82010-04-08 00:48:21 +0000290 break;
291 case Triple::arm:
292 case Triple::thumb:
293 stackPointers.insert(registerIDWithName("SP"));
294
295 programCounters.insert(registerIDWithName("PC"));
296 break;
Sean Callananee5dfd42010-02-01 08:49:35 +0000297 }
298}
299
300const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
301 if (registerID >= RegVec.size())
302 return NULL;
303 else
304 return RegVec[registerID].c_str();
305}
306
307unsigned EDDisassembler::registerIDWithName(const char *name) const {
308 regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
309 if (iter == RegRMap.end())
310 return 0;
311 else
312 return (*iter).second;
313}
314
315bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
316 return (stackPointers.find(registerID) != stackPointers.end());
317}
318
319bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
320 return (programCounters.find(registerID) != programCounters.end());
321}
322
Chris Lattnerd3740872010-04-04 05:04:31 +0000323int EDDisassembler::printInst(std::string &str, MCInst &inst) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000324 PrinterMutex.acquire();
325
Chris Lattnerd3740872010-04-04 05:04:31 +0000326 InstPrinter->printInst(&inst, *InstStream);
Sean Callananee5dfd42010-02-01 08:49:35 +0000327 InstStream->flush();
328 str = *InstString;
329 InstString->clear();
330
331 PrinterMutex.release();
332
333 return 0;
334}
335
Sean Callanan457856c2011-04-09 00:37:25 +0000336static void diag_handler(const SMDiagnostic &diag,
337 void *context)
338{
339 if (context) {
340 EDDisassembler *disassembler = static_cast<EDDisassembler*>(context);
341 diag.Print("", disassembler->ErrorStream);
342 }
343}
344
Sean Callananee5dfd42010-02-01 08:49:35 +0000345int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
346 SmallVectorImpl<AsmToken> &tokens,
347 const std::string &str) {
348 int ret = 0;
349
Sean Callanan8f993b82010-04-08 00:48:21 +0000350 switch (Key.Arch) {
351 default:
352 return -1;
353 case Triple::x86:
354 case Triple::x86_64:
355 case Triple::arm:
356 case Triple::thumb:
357 break;
358 }
359
Sean Callananee5dfd42010-02-01 08:49:35 +0000360 const char *cStr = str.c_str();
361 MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
362
363 StringRef instName;
364 SMLoc instLoc;
365
366 SourceMgr sourceMgr;
Sean Callanan457856c2011-04-09 00:37:25 +0000367 sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this));
Sean Callananee5dfd42010-02-01 08:49:35 +0000368 sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
Evan Cheng203576a2011-07-20 19:50:42 +0000369 MCContext context(*AsmInfo, *MRI, NULL);
Chris Lattnerc18409a2010-03-11 22:53:35 +0000370 OwningPtr<MCStreamer> streamer(createNullStreamer(context));
Jim Grosbach1b84cce2011-08-16 18:33:49 +0000371 OwningPtr<MCAsmParser> genericParser(createMCAsmParser(sourceMgr,
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000372 context, *streamer,
373 *AsmInfo));
Evan Chengebdeeab2011-07-08 01:53:10 +0000374
375 StringRef triple = tripleFromArch(Key.Arch);
Evan Chengffc0e732011-07-09 05:47:46 +0000376 OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(triple, "", ""));
Evan Cheng94b95502011-07-26 00:24:13 +0000377 OwningPtr<MCTargetAsmParser>
378 TargetParser(Tgt->createMCAsmParser(*STI, *genericParser));
Sean Callananee5dfd42010-02-01 08:49:35 +0000379
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000380 AsmToken OpcodeToken = genericParser->Lex();
381 AsmToken NextToken = genericParser->Lex(); // consume next token, because specificParser expects us to
Sean Callanana8702562010-04-24 01:00:16 +0000382
Sean Callanan8f993b82010-04-08 00:48:21 +0000383 if (OpcodeToken.is(AsmToken::Identifier)) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000384 instName = OpcodeToken.getString();
385 instLoc = OpcodeToken.getLoc();
Sean Callanana8702562010-04-24 01:00:16 +0000386
387 if (NextToken.isNot(AsmToken::Eof) &&
388 TargetParser->ParseInstruction(instName, instLoc, operands))
Sean Callananee5dfd42010-02-01 08:49:35 +0000389 ret = -1;
Sean Callanan8f993b82010-04-08 00:48:21 +0000390 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000391 ret = -1;
392 }
393
Sean Callananee5dfd42010-02-01 08:49:35 +0000394 ParserMutex.acquire();
395
396 if (!ret) {
397 GenericAsmLexer->setBuffer(buf);
398
399 while (SpecificAsmLexer->Lex(),
400 SpecificAsmLexer->isNot(AsmToken::Eof) &&
401 SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
402 if (SpecificAsmLexer->is(AsmToken::Error)) {
403 ret = -1;
404 break;
405 }
406 tokens.push_back(SpecificAsmLexer->getTok());
407 }
408 }
409
410 ParserMutex.release();
411
412 return ret;
413}
414
415int EDDisassembler::llvmSyntaxVariant() const {
416 return LLVMSyntaxVariant;
417}