blob: 4c2dae840ae718f4c162e1d4d56d2b7721653a18 [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"
Sean Callananee5dfd42010-02-01 08:49:35 +000037using namespace llvm;
38
Sean Callananee5dfd42010-02-01 08:49:35 +000039EDDisassembler::DisassemblerMap_t EDDisassembler::sDisassemblers;
40
Sean Callanan9899f702010-04-13 21:21:57 +000041struct TripleMap {
Sean Callananee5dfd42010-02-01 08:49:35 +000042 Triple::ArchType Arch;
43 const char *String;
Sean Callananee5dfd42010-02-01 08:49:35 +000044};
45
Sean Callanan9899f702010-04-13 21:21:57 +000046static struct TripleMap triplemap[] = {
47 { Triple::x86, "i386-unknown-unknown" },
48 { Triple::x86_64, "x86_64-unknown-unknown" },
49 { Triple::arm, "arm-unknown-unknown" },
Chandler Carruth124e51c2012-02-21 03:39:36 +000050 { Triple::thumb, "thumb-unknown-unknown" }
Sean Callananee5dfd42010-02-01 08:49:35 +000051};
52
Sean Callanan9899f702010-04-13 21:21:57 +000053/// infoFromArch - Returns the TripleMap corresponding to a given architecture,
Sean Callananee5dfd42010-02-01 08:49:35 +000054/// or NULL if there is an error
55///
56/// @arg arch - The Triple::ArchType for the desired architecture
Sean Callanan9899f702010-04-13 21:21:57 +000057static const char *tripleFromArch(Triple::ArchType arch) {
Sean Callananee5dfd42010-02-01 08:49:35 +000058 unsigned int infoIndex;
59
Sean Callanan9899f702010-04-13 21:21:57 +000060 for (infoIndex = 0; triplemap[infoIndex].String != NULL; ++infoIndex) {
61 if (arch == triplemap[infoIndex].Arch)
62 return triplemap[infoIndex].String;
Sean Callananee5dfd42010-02-01 08:49:35 +000063 }
64
65 return NULL;
66}
67
68/// getLLVMSyntaxVariant - gets the constant to use to get an assembly printer
69/// for the desired assembly syntax, suitable for passing to
70/// Target::createMCInstPrinter()
71///
72/// @arg arch - The target architecture
73/// @arg syntax - The assembly syntax in sd form
74static int getLLVMSyntaxVariant(Triple::ArchType arch,
Chris Lattner847da552010-07-20 18:25:19 +000075 EDDisassembler::AssemblySyntax syntax) {
Sean Callananee5dfd42010-02-01 08:49:35 +000076 switch (syntax) {
Sean Callananee5dfd42010-02-01 08:49:35 +000077 // Mappings below from X86AsmPrinter.cpp
Chris Lattner847da552010-07-20 18:25:19 +000078 case EDDisassembler::kEDAssemblySyntaxX86ATT:
Sean Callananee5dfd42010-02-01 08:49:35 +000079 if (arch == Triple::x86 || arch == Triple::x86_64)
80 return 0;
Benjamin Kramere4ad5822012-01-10 20:47:20 +000081 break;
Chris Lattner847da552010-07-20 18:25:19 +000082 case EDDisassembler::kEDAssemblySyntaxX86Intel:
Sean Callananee5dfd42010-02-01 08:49:35 +000083 if (arch == Triple::x86 || arch == Triple::x86_64)
84 return 1;
Benjamin Kramere4ad5822012-01-10 20:47:20 +000085 break;
Chris Lattner847da552010-07-20 18:25:19 +000086 case EDDisassembler::kEDAssemblySyntaxARMUAL:
Sean Callanan8f993b82010-04-08 00:48:21 +000087 if (arch == Triple::arm || arch == Triple::thumb)
88 return 0;
Benjamin Kramere4ad5822012-01-10 20:47:20 +000089 break;
Sean Callananee5dfd42010-02-01 08:49:35 +000090 }
Benjamin Kramere4ad5822012-01-10 20:47:20 +000091
92 return -1;
Sean Callananee5dfd42010-02-01 08:49:35 +000093}
94
Sean Callananee5dfd42010-02-01 08:49:35 +000095EDDisassembler *EDDisassembler::getDisassembler(Triple::ArchType arch,
Chris Lattner847da552010-07-20 18:25:19 +000096 AssemblySyntax syntax) {
Sean Callanan3f4f34c2012-02-04 07:45:35 +000097 const char *triple = tripleFromArch(arch);
98 return getDisassembler(StringRef(triple), syntax);
99}
100
101EDDisassembler *EDDisassembler::getDisassembler(StringRef str,
102 AssemblySyntax syntax) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000103 CPUKey key;
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000104 key.Triple = str.str();
Sean Callananee5dfd42010-02-01 08:49:35 +0000105 key.Syntax = syntax;
Ahmed Charlesb0934ab2012-02-19 11:37:01 +0000106
Sean Callananee5dfd42010-02-01 08:49:35 +0000107 EDDisassembler::DisassemblerMap_t::iterator i = sDisassemblers.find(key);
Ahmed Charlesb0934ab2012-02-19 11:37:01 +0000108
Sean Callananee5dfd42010-02-01 08:49:35 +0000109 if (i != sDisassemblers.end()) {
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000110 return i->second;
111 }
Ahmed Charlesb0934ab2012-02-19 11:37:01 +0000112
113 EDDisassembler *sdd = new EDDisassembler(key);
114 if (!sdd->valid()) {
115 delete sdd;
116 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000117 }
Ahmed Charlesb0934ab2012-02-19 11:37:01 +0000118
119 sDisassemblers[key] = sdd;
120
121 return sdd;
Sean Callananee5dfd42010-02-01 08:49:35 +0000122}
123
Sean Callananee5dfd42010-02-01 08:49:35 +0000124EDDisassembler::EDDisassembler(CPUKey &key) :
Sean Callanan8f993b82010-04-08 00:48:21 +0000125 Valid(false),
126 HasSemantics(false),
127 ErrorStream(nulls()),
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000128 Key(key),
129 TgtTriple(key.Triple.c_str()) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000130
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000131 LLVMSyntaxVariant = getLLVMSyntaxVariant(TgtTriple.getArch(), key.Syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000132
Sean Callanan4285b292010-04-09 00:11:15 +0000133 if (LLVMSyntaxVariant < 0)
Sean Callananee5dfd42010-02-01 08:49:35 +0000134 return;
135
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000136 std::string tripleString(key.Triple);
Sean Callananee5dfd42010-02-01 08:49:35 +0000137 std::string errorString;
138
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000139 Tgt = TargetRegistry::lookupTarget(key.Triple,
Sean Callananee5dfd42010-02-01 08:49:35 +0000140 errorString);
141
142 if (!Tgt)
143 return;
144
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000145 MRI.reset(Tgt->createMCRegInfo(tripleString));
Evan Chengebdeeab2011-07-08 01:53:10 +0000146
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000147 if (!MRI)
Sean Callananee5dfd42010-02-01 08:49:35 +0000148 return;
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000149
150 initMaps(*MRI);
Sean Callananee5dfd42010-02-01 08:49:35 +0000151
Evan Cheng1abf2cb2011-07-14 23:50:31 +0000152 AsmInfo.reset(Tgt->createMCAsmInfo(tripleString));
Sean Callananee5dfd42010-02-01 08:49:35 +0000153
154 if (!AsmInfo)
155 return;
Evan Chengebdeeab2011-07-08 01:53:10 +0000156
James Molloyb9505852011-09-07 17:24:38 +0000157 STI.reset(Tgt->createMCSubtargetInfo(tripleString, "", ""));
158
159 if (!STI)
160 return;
161
162 Disassembler.reset(Tgt->createMCDisassembler(*STI));
Sean Callananee5dfd42010-02-01 08:49:35 +0000163
164 if (!Disassembler)
165 return;
Sean Callanan9899f702010-04-13 21:21:57 +0000166
167 InstInfos = Disassembler->getEDInfo();
Sean Callananee5dfd42010-02-01 08:49:35 +0000168
169 InstString.reset(new std::string);
170 InstStream.reset(new raw_string_ostream(*InstString));
Jim Grosbachc6449b62012-03-05 19:33:20 +0000171 InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo,
172 *MRI, *STI));
Sean Callananee5dfd42010-02-01 08:49:35 +0000173
174 if (!InstPrinter)
175 return;
176
177 GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
Evan Cheng94b95502011-07-26 00:24:13 +0000178 SpecificAsmLexer.reset(Tgt->createMCAsmLexer(*MRI, *AsmInfo));
Sean Callananee5dfd42010-02-01 08:49:35 +0000179 SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
Sean Callanand74667e2010-02-02 02:18:20 +0000180
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000181 initMaps(*MRI);
Sean Callananee5dfd42010-02-01 08:49:35 +0000182
183 Valid = true;
184}
185
186EDDisassembler::~EDDisassembler() {
Sean Callanan8f993b82010-04-08 00:48:21 +0000187 if (!valid())
Sean Callananee5dfd42010-02-01 08:49:35 +0000188 return;
189}
190
191namespace {
192 /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
193 /// as provided by the sd interface. See MemoryObject.
194 class EDMemoryObject : public llvm::MemoryObject {
195 private:
196 EDByteReaderCallback Callback;
197 void *Arg;
198 public:
199 EDMemoryObject(EDByteReaderCallback callback,
200 void *arg) : Callback(callback), Arg(arg) { }
201 ~EDMemoryObject() { }
202 uint64_t getBase() const { return 0x0; }
Derek Schuffadef06a2012-02-29 01:09:06 +0000203 uint64_t getExtent() const { return (uint64_t)-1; }
204 int readByte(uint64_t address, uint8_t *ptr) const {
Sean Callanan8f993b82010-04-08 00:48:21 +0000205 if (!Callback)
Sean Callananee5dfd42010-02-01 08:49:35 +0000206 return -1;
207
Sean Callanan8f993b82010-04-08 00:48:21 +0000208 if (Callback(ptr, address, Arg))
Sean Callananee5dfd42010-02-01 08:49:35 +0000209 return -1;
210
211 return 0;
212 }
213 };
214}
215
216EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
217 uint64_t address,
218 void *arg) {
219 EDMemoryObject memoryObject(byteReader, arg);
220
221 MCInst* inst = new MCInst;
222 uint64_t byteSize;
223
James Molloyee064432011-09-01 22:01:14 +0000224 MCDisassembler::DecodeStatus S;
225 S = Disassembler->getInstruction(*inst, byteSize, memoryObject, address,
Owen Anderson98c5dda2011-09-15 23:38:46 +0000226 ErrorStream, nulls());
James Molloyee064432011-09-01 22:01:14 +0000227 switch (S) {
228 case MCDisassembler::Fail:
229 case MCDisassembler::SoftFail:
James Molloyc047dca2011-09-01 18:02:14 +0000230 // FIXME: Do something different on soft failure mode?
Sean Callananee5dfd42010-02-01 08:49:35 +0000231 delete inst;
232 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000233
James Molloyee064432011-09-01 22:01:14 +0000234 case MCDisassembler::Success: {
235 const llvm::EDInstInfo *thisInstInfo = NULL;
236
237 if (InstInfos) {
238 thisInstInfo = &InstInfos[inst->getOpcode()];
239 }
240
241 EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
242 return sdInst;
243 }
244 }
245 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000246}
247
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000248void EDDisassembler::initMaps(const MCRegisterInfo &registerInfo) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000249 unsigned numRegisters = registerInfo.getNumRegs();
250 unsigned registerIndex;
251
252 for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
253 const char* registerName = registerInfo.get(registerIndex).Name;
254
255 RegVec.push_back(registerName);
256 RegRMap[registerName] = registerIndex;
257 }
258
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000259 switch (TgtTriple.getArch()) {
Sean Callanan8f993b82010-04-08 00:48:21 +0000260 default:
261 break;
262 case Triple::x86:
263 case Triple::x86_64:
Sean Callananee5dfd42010-02-01 08:49:35 +0000264 stackPointers.insert(registerIDWithName("SP"));
265 stackPointers.insert(registerIDWithName("ESP"));
266 stackPointers.insert(registerIDWithName("RSP"));
267
268 programCounters.insert(registerIDWithName("IP"));
269 programCounters.insert(registerIDWithName("EIP"));
270 programCounters.insert(registerIDWithName("RIP"));
Sean Callanan8f993b82010-04-08 00:48:21 +0000271 break;
272 case Triple::arm:
273 case Triple::thumb:
274 stackPointers.insert(registerIDWithName("SP"));
275
276 programCounters.insert(registerIDWithName("PC"));
277 break;
Sean Callananee5dfd42010-02-01 08:49:35 +0000278 }
279}
280
281const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
282 if (registerID >= RegVec.size())
283 return NULL;
284 else
285 return RegVec[registerID].c_str();
286}
287
288unsigned EDDisassembler::registerIDWithName(const char *name) const {
289 regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
290 if (iter == RegRMap.end())
291 return 0;
292 else
293 return (*iter).second;
294}
295
296bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
297 return (stackPointers.find(registerID) != stackPointers.end());
298}
299
300bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
301 return (programCounters.find(registerID) != programCounters.end());
302}
303
Chris Lattnerd3740872010-04-04 05:04:31 +0000304int EDDisassembler::printInst(std::string &str, MCInst &inst) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000305 PrinterMutex.acquire();
306
Owen Anderson98c5dda2011-09-15 23:38:46 +0000307 InstPrinter->printInst(&inst, *InstStream, "");
Sean Callananee5dfd42010-02-01 08:49:35 +0000308 InstStream->flush();
309 str = *InstString;
310 InstString->clear();
311
312 PrinterMutex.release();
313
314 return 0;
315}
316
Chris Lattnerd8b7aa22011-10-16 04:47:35 +0000317static void diag_handler(const SMDiagnostic &diag, void *context) {
318 if (context)
319 diag.print("", static_cast<EDDisassembler*>(context)->ErrorStream);
Sean Callanan457856c2011-04-09 00:37:25 +0000320}
321
Sean Callananee5dfd42010-02-01 08:49:35 +0000322int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
323 SmallVectorImpl<AsmToken> &tokens,
324 const std::string &str) {
325 int ret = 0;
326
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000327 switch (TgtTriple.getArch()) {
Sean Callanan8f993b82010-04-08 00:48:21 +0000328 default:
329 return -1;
330 case Triple::x86:
331 case Triple::x86_64:
332 case Triple::arm:
333 case Triple::thumb:
334 break;
335 }
336
Sean Callananee5dfd42010-02-01 08:49:35 +0000337 const char *cStr = str.c_str();
338 MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
339
340 StringRef instName;
341 SMLoc instLoc;
342
343 SourceMgr sourceMgr;
Sean Callanan457856c2011-04-09 00:37:25 +0000344 sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this));
Sean Callananee5dfd42010-02-01 08:49:35 +0000345 sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
Evan Cheng203576a2011-07-20 19:50:42 +0000346 MCContext context(*AsmInfo, *MRI, NULL);
Chris Lattnerc18409a2010-03-11 22:53:35 +0000347 OwningPtr<MCStreamer> streamer(createNullStreamer(context));
Jim Grosbach1b84cce2011-08-16 18:33:49 +0000348 OwningPtr<MCAsmParser> genericParser(createMCAsmParser(sourceMgr,
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000349 context, *streamer,
350 *AsmInfo));
Evan Chengebdeeab2011-07-08 01:53:10 +0000351
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000352 OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(Key.Triple.c_str(), "", ""));
Evan Cheng94b95502011-07-26 00:24:13 +0000353 OwningPtr<MCTargetAsmParser>
354 TargetParser(Tgt->createMCAsmParser(*STI, *genericParser));
Sean Callananee5dfd42010-02-01 08:49:35 +0000355
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000356 AsmToken OpcodeToken = genericParser->Lex();
357 AsmToken NextToken = genericParser->Lex(); // consume next token, because specificParser expects us to
Sean Callanana8702562010-04-24 01:00:16 +0000358
Sean Callanan8f993b82010-04-08 00:48:21 +0000359 if (OpcodeToken.is(AsmToken::Identifier)) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000360 instName = OpcodeToken.getString();
361 instLoc = OpcodeToken.getLoc();
Sean Callanana8702562010-04-24 01:00:16 +0000362
363 if (NextToken.isNot(AsmToken::Eof) &&
364 TargetParser->ParseInstruction(instName, instLoc, operands))
Sean Callananee5dfd42010-02-01 08:49:35 +0000365 ret = -1;
Sean Callanan8f993b82010-04-08 00:48:21 +0000366 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000367 ret = -1;
368 }
369
Sean Callananee5dfd42010-02-01 08:49:35 +0000370 ParserMutex.acquire();
371
372 if (!ret) {
373 GenericAsmLexer->setBuffer(buf);
374
375 while (SpecificAsmLexer->Lex(),
376 SpecificAsmLexer->isNot(AsmToken::Eof) &&
377 SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
378 if (SpecificAsmLexer->is(AsmToken::Error)) {
379 ret = -1;
380 break;
381 }
382 tokens.push_back(SpecificAsmLexer->getTok());
383 }
384 }
385
386 ParserMutex.release();
387
388 return ret;
389}
390
391int EDDisassembler::llvmSyntaxVariant() const {
392 return LLVMSyntaxVariant;
393}