blob: 1226f1a2e324c0092c1a2305063f7834d9592290 [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"
Craig Topper17463b32012-04-02 06:09:36 +000025#include "llvm/MC/MCInstrInfo.h"
Evan Cheng0e6a0522011-07-18 20:57:22 +000026#include "llvm/MC/MCRegisterInfo.h"
Sean Callananee5dfd42010-02-01 08:49:35 +000027#include "llvm/MC/MCStreamer.h"
Evan Chengffc0e732011-07-09 05:47:46 +000028#include "llvm/MC/MCSubtargetInfo.h"
Sean Callananee5dfd42010-02-01 08:49:35 +000029#include "llvm/MC/MCParser/AsmLexer.h"
Sean Callananee5dfd42010-02-01 08:49:35 +000030#include "llvm/MC/MCParser/MCAsmParser.h"
31#include "llvm/MC/MCParser/MCParsedAsmOperand.h"
Evan Cheng94b95502011-07-26 00:24:13 +000032#include "llvm/MC/MCTargetAsmLexer.h"
33#include "llvm/MC/MCTargetAsmParser.h"
Sean Callananee5dfd42010-02-01 08:49:35 +000034#include "llvm/Support/MemoryBuffer.h"
35#include "llvm/Support/MemoryObject.h"
36#include "llvm/Support/SourceMgr.h"
Evan Cheng3e74d6f2011-08-24 18:08:43 +000037#include "llvm/Support/TargetRegistry.h"
Sean Callananee5dfd42010-02-01 08:49:35 +000038using namespace llvm;
39
Sean Callananee5dfd42010-02-01 08:49:35 +000040EDDisassembler::DisassemblerMap_t EDDisassembler::sDisassemblers;
41
Sean Callanan9899f702010-04-13 21:21:57 +000042struct TripleMap {
Sean Callananee5dfd42010-02-01 08:49:35 +000043 Triple::ArchType Arch;
44 const char *String;
Sean Callananee5dfd42010-02-01 08:49:35 +000045};
46
Craig Toppere3298102012-05-24 06:35:32 +000047static const struct TripleMap triplemap[] = {
Sean Callanan9899f702010-04-13 21:21:57 +000048 { Triple::x86, "i386-unknown-unknown" },
49 { Triple::x86_64, "x86_64-unknown-unknown" },
50 { Triple::arm, "arm-unknown-unknown" },
Chandler Carruth124e51c2012-02-21 03:39:36 +000051 { Triple::thumb, "thumb-unknown-unknown" }
Sean Callananee5dfd42010-02-01 08:49:35 +000052};
53
Sean Callanan9899f702010-04-13 21:21:57 +000054/// infoFromArch - Returns the TripleMap corresponding to a given architecture,
Sean Callananee5dfd42010-02-01 08:49:35 +000055/// or NULL if there is an error
56///
57/// @arg arch - The Triple::ArchType for the desired architecture
Sean Callanan9899f702010-04-13 21:21:57 +000058static const char *tripleFromArch(Triple::ArchType arch) {
Sean Callananee5dfd42010-02-01 08:49:35 +000059 unsigned int infoIndex;
60
Sean Callanan9899f702010-04-13 21:21:57 +000061 for (infoIndex = 0; triplemap[infoIndex].String != NULL; ++infoIndex) {
62 if (arch == triplemap[infoIndex].Arch)
63 return triplemap[infoIndex].String;
Sean Callananee5dfd42010-02-01 08:49:35 +000064 }
65
66 return NULL;
67}
68
69/// getLLVMSyntaxVariant - gets the constant to use to get an assembly printer
70/// for the desired assembly syntax, suitable for passing to
71/// Target::createMCInstPrinter()
72///
73/// @arg arch - The target architecture
74/// @arg syntax - The assembly syntax in sd form
75static int getLLVMSyntaxVariant(Triple::ArchType arch,
Chris Lattner847da552010-07-20 18:25:19 +000076 EDDisassembler::AssemblySyntax syntax) {
Sean Callananee5dfd42010-02-01 08:49:35 +000077 switch (syntax) {
Sean Callananee5dfd42010-02-01 08:49:35 +000078 // Mappings below from X86AsmPrinter.cpp
Chris Lattner847da552010-07-20 18:25:19 +000079 case EDDisassembler::kEDAssemblySyntaxX86ATT:
Sean Callananee5dfd42010-02-01 08:49:35 +000080 if (arch == Triple::x86 || arch == Triple::x86_64)
81 return 0;
Benjamin Kramere4ad5822012-01-10 20:47:20 +000082 break;
Chris Lattner847da552010-07-20 18:25:19 +000083 case EDDisassembler::kEDAssemblySyntaxX86Intel:
Sean Callananee5dfd42010-02-01 08:49:35 +000084 if (arch == Triple::x86 || arch == Triple::x86_64)
85 return 1;
Benjamin Kramere4ad5822012-01-10 20:47:20 +000086 break;
Chris Lattner847da552010-07-20 18:25:19 +000087 case EDDisassembler::kEDAssemblySyntaxARMUAL:
Sean Callanan8f993b82010-04-08 00:48:21 +000088 if (arch == Triple::arm || arch == Triple::thumb)
89 return 0;
Benjamin Kramere4ad5822012-01-10 20:47:20 +000090 break;
Sean Callananee5dfd42010-02-01 08:49:35 +000091 }
Benjamin Kramere4ad5822012-01-10 20:47:20 +000092
93 return -1;
Sean Callananee5dfd42010-02-01 08:49:35 +000094}
95
Sean Callananee5dfd42010-02-01 08:49:35 +000096EDDisassembler *EDDisassembler::getDisassembler(Triple::ArchType arch,
Chris Lattner847da552010-07-20 18:25:19 +000097 AssemblySyntax syntax) {
Sean Callanan3f4f34c2012-02-04 07:45:35 +000098 const char *triple = tripleFromArch(arch);
99 return getDisassembler(StringRef(triple), syntax);
100}
101
102EDDisassembler *EDDisassembler::getDisassembler(StringRef str,
103 AssemblySyntax syntax) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000104 CPUKey key;
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000105 key.Triple = str.str();
Sean Callananee5dfd42010-02-01 08:49:35 +0000106 key.Syntax = syntax;
Ahmed Charlesb0934ab2012-02-19 11:37:01 +0000107
Sean Callananee5dfd42010-02-01 08:49:35 +0000108 EDDisassembler::DisassemblerMap_t::iterator i = sDisassemblers.find(key);
Ahmed Charlesb0934ab2012-02-19 11:37:01 +0000109
Sean Callananee5dfd42010-02-01 08:49:35 +0000110 if (i != sDisassemblers.end()) {
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000111 return i->second;
112 }
Ahmed Charlesb0934ab2012-02-19 11:37:01 +0000113
114 EDDisassembler *sdd = new EDDisassembler(key);
115 if (!sdd->valid()) {
116 delete sdd;
117 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000118 }
Ahmed Charlesb0934ab2012-02-19 11:37:01 +0000119
120 sDisassemblers[key] = sdd;
121
122 return sdd;
Sean Callananee5dfd42010-02-01 08:49:35 +0000123}
124
Sean Callananee5dfd42010-02-01 08:49:35 +0000125EDDisassembler::EDDisassembler(CPUKey &key) :
Sean Callanan8f993b82010-04-08 00:48:21 +0000126 Valid(false),
127 HasSemantics(false),
128 ErrorStream(nulls()),
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000129 Key(key),
130 TgtTriple(key.Triple.c_str()) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000131
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000132 LLVMSyntaxVariant = getLLVMSyntaxVariant(TgtTriple.getArch(), key.Syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000133
Sean Callanan4285b292010-04-09 00:11:15 +0000134 if (LLVMSyntaxVariant < 0)
Sean Callananee5dfd42010-02-01 08:49:35 +0000135 return;
136
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000137 std::string tripleString(key.Triple);
Sean Callananee5dfd42010-02-01 08:49:35 +0000138 std::string errorString;
139
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000140 Tgt = TargetRegistry::lookupTarget(key.Triple,
Sean Callananee5dfd42010-02-01 08:49:35 +0000141 errorString);
142
143 if (!Tgt)
144 return;
145
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000146 MRI.reset(Tgt->createMCRegInfo(tripleString));
Evan Chengebdeeab2011-07-08 01:53:10 +0000147
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000148 if (!MRI)
Sean Callananee5dfd42010-02-01 08:49:35 +0000149 return;
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000150
151 initMaps(*MRI);
Sean Callananee5dfd42010-02-01 08:49:35 +0000152
Evan Cheng1abf2cb2011-07-14 23:50:31 +0000153 AsmInfo.reset(Tgt->createMCAsmInfo(tripleString));
Sean Callananee5dfd42010-02-01 08:49:35 +0000154
155 if (!AsmInfo)
156 return;
Evan Chengebdeeab2011-07-08 01:53:10 +0000157
James Molloyb9505852011-09-07 17:24:38 +0000158 STI.reset(Tgt->createMCSubtargetInfo(tripleString, "", ""));
159
160 if (!STI)
161 return;
162
163 Disassembler.reset(Tgt->createMCDisassembler(*STI));
Sean Callananee5dfd42010-02-01 08:49:35 +0000164
165 if (!Disassembler)
166 return;
Sean Callanan9899f702010-04-13 21:21:57 +0000167
168 InstInfos = Disassembler->getEDInfo();
Craig Topper17463b32012-04-02 06:09:36 +0000169
170 MII.reset(Tgt->createMCInstrInfo());
171
172 if (!MII)
173 return;
174
Sean Callananee5dfd42010-02-01 08:49:35 +0000175 InstString.reset(new std::string);
176 InstStream.reset(new raw_string_ostream(*InstString));
Jim Grosbachc6449b62012-03-05 19:33:20 +0000177 InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo,
Craig Topper17463b32012-04-02 06:09:36 +0000178 *MII, *MRI, *STI));
Sean Callananee5dfd42010-02-01 08:49:35 +0000179
180 if (!InstPrinter)
181 return;
182
183 GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
Evan Cheng94b95502011-07-26 00:24:13 +0000184 SpecificAsmLexer.reset(Tgt->createMCAsmLexer(*MRI, *AsmInfo));
Sean Callananee5dfd42010-02-01 08:49:35 +0000185 SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
Sean Callanand74667e2010-02-02 02:18:20 +0000186
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000187 initMaps(*MRI);
Sean Callananee5dfd42010-02-01 08:49:35 +0000188
189 Valid = true;
190}
191
192EDDisassembler::~EDDisassembler() {
Sean Callanan8f993b82010-04-08 00:48:21 +0000193 if (!valid())
Sean Callananee5dfd42010-02-01 08:49:35 +0000194 return;
195}
196
197namespace {
198 /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
199 /// as provided by the sd interface. See MemoryObject.
200 class EDMemoryObject : public llvm::MemoryObject {
201 private:
202 EDByteReaderCallback Callback;
203 void *Arg;
204 public:
205 EDMemoryObject(EDByteReaderCallback callback,
206 void *arg) : Callback(callback), Arg(arg) { }
207 ~EDMemoryObject() { }
208 uint64_t getBase() const { return 0x0; }
Derek Schuffadef06a2012-02-29 01:09:06 +0000209 uint64_t getExtent() const { return (uint64_t)-1; }
210 int readByte(uint64_t address, uint8_t *ptr) const {
Sean Callanan8f993b82010-04-08 00:48:21 +0000211 if (!Callback)
Sean Callananee5dfd42010-02-01 08:49:35 +0000212 return -1;
213
Sean Callanan8f993b82010-04-08 00:48:21 +0000214 if (Callback(ptr, address, Arg))
Sean Callananee5dfd42010-02-01 08:49:35 +0000215 return -1;
216
217 return 0;
218 }
219 };
220}
221
222EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
223 uint64_t address,
224 void *arg) {
225 EDMemoryObject memoryObject(byteReader, arg);
226
227 MCInst* inst = new MCInst;
228 uint64_t byteSize;
229
James Molloyee064432011-09-01 22:01:14 +0000230 MCDisassembler::DecodeStatus S;
231 S = Disassembler->getInstruction(*inst, byteSize, memoryObject, address,
Owen Anderson98c5dda2011-09-15 23:38:46 +0000232 ErrorStream, nulls());
James Molloyee064432011-09-01 22:01:14 +0000233 switch (S) {
234 case MCDisassembler::Fail:
235 case MCDisassembler::SoftFail:
James Molloyc047dca2011-09-01 18:02:14 +0000236 // FIXME: Do something different on soft failure mode?
Sean Callananee5dfd42010-02-01 08:49:35 +0000237 delete inst;
238 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000239
James Molloyee064432011-09-01 22:01:14 +0000240 case MCDisassembler::Success: {
241 const llvm::EDInstInfo *thisInstInfo = NULL;
242
243 if (InstInfos) {
244 thisInstInfo = &InstInfos[inst->getOpcode()];
245 }
246
247 EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
248 return sdInst;
249 }
250 }
251 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000252}
253
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000254void EDDisassembler::initMaps(const MCRegisterInfo &registerInfo) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000255 unsigned numRegisters = registerInfo.getNumRegs();
256 unsigned registerIndex;
257
258 for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
Jakob Stoklund Olesend5ce3ff2012-05-25 00:21:41 +0000259 const char* registerName = registerInfo.getName(registerIndex);
Sean Callananee5dfd42010-02-01 08:49:35 +0000260
261 RegVec.push_back(registerName);
262 RegRMap[registerName] = registerIndex;
263 }
264
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000265 switch (TgtTriple.getArch()) {
Sean Callanan8f993b82010-04-08 00:48:21 +0000266 default:
267 break;
268 case Triple::x86:
269 case Triple::x86_64:
Sean Callananee5dfd42010-02-01 08:49:35 +0000270 stackPointers.insert(registerIDWithName("SP"));
271 stackPointers.insert(registerIDWithName("ESP"));
272 stackPointers.insert(registerIDWithName("RSP"));
273
274 programCounters.insert(registerIDWithName("IP"));
275 programCounters.insert(registerIDWithName("EIP"));
276 programCounters.insert(registerIDWithName("RIP"));
Sean Callanan8f993b82010-04-08 00:48:21 +0000277 break;
278 case Triple::arm:
279 case Triple::thumb:
280 stackPointers.insert(registerIDWithName("SP"));
281
282 programCounters.insert(registerIDWithName("PC"));
283 break;
Sean Callananee5dfd42010-02-01 08:49:35 +0000284 }
285}
286
287const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
288 if (registerID >= RegVec.size())
289 return NULL;
290 else
291 return RegVec[registerID].c_str();
292}
293
294unsigned EDDisassembler::registerIDWithName(const char *name) const {
295 regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
296 if (iter == RegRMap.end())
297 return 0;
298 else
299 return (*iter).second;
300}
301
302bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
303 return (stackPointers.find(registerID) != stackPointers.end());
304}
305
306bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
307 return (programCounters.find(registerID) != programCounters.end());
308}
309
Chris Lattnerd3740872010-04-04 05:04:31 +0000310int EDDisassembler::printInst(std::string &str, MCInst &inst) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000311 PrinterMutex.acquire();
312
Owen Anderson98c5dda2011-09-15 23:38:46 +0000313 InstPrinter->printInst(&inst, *InstStream, "");
Sean Callananee5dfd42010-02-01 08:49:35 +0000314 InstStream->flush();
315 str = *InstString;
316 InstString->clear();
317
318 PrinterMutex.release();
319
320 return 0;
321}
322
Chris Lattnerd8b7aa22011-10-16 04:47:35 +0000323static void diag_handler(const SMDiagnostic &diag, void *context) {
324 if (context)
325 diag.print("", static_cast<EDDisassembler*>(context)->ErrorStream);
Sean Callanan457856c2011-04-09 00:37:25 +0000326}
327
Sean Callananee5dfd42010-02-01 08:49:35 +0000328int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
329 SmallVectorImpl<AsmToken> &tokens,
330 const std::string &str) {
331 int ret = 0;
332
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000333 switch (TgtTriple.getArch()) {
Sean Callanan8f993b82010-04-08 00:48:21 +0000334 default:
335 return -1;
336 case Triple::x86:
337 case Triple::x86_64:
338 case Triple::arm:
339 case Triple::thumb:
340 break;
341 }
342
Sean Callananee5dfd42010-02-01 08:49:35 +0000343 const char *cStr = str.c_str();
344 MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
345
346 StringRef instName;
347 SMLoc instLoc;
348
349 SourceMgr sourceMgr;
Sean Callanan457856c2011-04-09 00:37:25 +0000350 sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this));
Sean Callananee5dfd42010-02-01 08:49:35 +0000351 sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
Evan Cheng203576a2011-07-20 19:50:42 +0000352 MCContext context(*AsmInfo, *MRI, NULL);
Chris Lattnerc18409a2010-03-11 22:53:35 +0000353 OwningPtr<MCStreamer> streamer(createNullStreamer(context));
Jim Grosbach1b84cce2011-08-16 18:33:49 +0000354 OwningPtr<MCAsmParser> genericParser(createMCAsmParser(sourceMgr,
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000355 context, *streamer,
356 *AsmInfo));
Evan Chengebdeeab2011-07-08 01:53:10 +0000357
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000358 OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(Key.Triple.c_str(), "", ""));
Evan Cheng94b95502011-07-26 00:24:13 +0000359 OwningPtr<MCTargetAsmParser>
360 TargetParser(Tgt->createMCAsmParser(*STI, *genericParser));
Sean Callananee5dfd42010-02-01 08:49:35 +0000361
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000362 AsmToken OpcodeToken = genericParser->Lex();
363 AsmToken NextToken = genericParser->Lex(); // consume next token, because specificParser expects us to
Sean Callanana8702562010-04-24 01:00:16 +0000364
Sean Callanan8f993b82010-04-08 00:48:21 +0000365 if (OpcodeToken.is(AsmToken::Identifier)) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000366 instName = OpcodeToken.getString();
367 instLoc = OpcodeToken.getLoc();
Sean Callanana8702562010-04-24 01:00:16 +0000368
369 if (NextToken.isNot(AsmToken::Eof) &&
370 TargetParser->ParseInstruction(instName, instLoc, operands))
Sean Callananee5dfd42010-02-01 08:49:35 +0000371 ret = -1;
Sean Callanan8f993b82010-04-08 00:48:21 +0000372 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000373 ret = -1;
374 }
375
Sean Callananee5dfd42010-02-01 08:49:35 +0000376 ParserMutex.acquire();
377
378 if (!ret) {
379 GenericAsmLexer->setBuffer(buf);
380
381 while (SpecificAsmLexer->Lex(),
382 SpecificAsmLexer->isNot(AsmToken::Eof) &&
383 SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
384 if (SpecificAsmLexer->is(AsmToken::Error)) {
385 ret = -1;
386 break;
387 }
388 tokens.push_back(SpecificAsmLexer->getTok());
389 }
390 }
391
392 ParserMutex.release();
393
394 return ret;
395}
396
397int EDDisassembler::llvmSyntaxVariant() const {
398 return LLVMSyntaxVariant;
399}