blob: 35cd3d77f8d8ec4fdaa3c510c01da9815439469d [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));
James Molloyb9505852011-09-07 17:24:38 +0000171 InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo, *STI));
Sean Callananee5dfd42010-02-01 08:49:35 +0000172
173 if (!InstPrinter)
174 return;
175
176 GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
Evan Cheng94b95502011-07-26 00:24:13 +0000177 SpecificAsmLexer.reset(Tgt->createMCAsmLexer(*MRI, *AsmInfo));
Sean Callananee5dfd42010-02-01 08:49:35 +0000178 SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
Sean Callanand74667e2010-02-02 02:18:20 +0000179
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000180 initMaps(*MRI);
Sean Callananee5dfd42010-02-01 08:49:35 +0000181
182 Valid = true;
183}
184
185EDDisassembler::~EDDisassembler() {
Sean Callanan8f993b82010-04-08 00:48:21 +0000186 if (!valid())
Sean Callananee5dfd42010-02-01 08:49:35 +0000187 return;
188}
189
190namespace {
191 /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
192 /// as provided by the sd interface. See MemoryObject.
193 class EDMemoryObject : public llvm::MemoryObject {
194 private:
195 EDByteReaderCallback Callback;
196 void *Arg;
197 public:
198 EDMemoryObject(EDByteReaderCallback callback,
199 void *arg) : Callback(callback), Arg(arg) { }
200 ~EDMemoryObject() { }
201 uint64_t getBase() const { return 0x0; }
Derek Schuff2ea93872012-02-06 22:30:29 +0000202 uint64_t getExtent() { return (uint64_t)-1; }
203 int readByte(uint64_t address, uint8_t *ptr) {
Sean Callanan8f993b82010-04-08 00:48:21 +0000204 if (!Callback)
Sean Callananee5dfd42010-02-01 08:49:35 +0000205 return -1;
206
Sean Callanan8f993b82010-04-08 00:48:21 +0000207 if (Callback(ptr, address, Arg))
Sean Callananee5dfd42010-02-01 08:49:35 +0000208 return -1;
209
210 return 0;
211 }
212 };
213}
214
215EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
216 uint64_t address,
217 void *arg) {
218 EDMemoryObject memoryObject(byteReader, arg);
219
220 MCInst* inst = new MCInst;
221 uint64_t byteSize;
222
James Molloyee064432011-09-01 22:01:14 +0000223 MCDisassembler::DecodeStatus S;
224 S = Disassembler->getInstruction(*inst, byteSize, memoryObject, address,
Owen Anderson98c5dda2011-09-15 23:38:46 +0000225 ErrorStream, nulls());
James Molloyee064432011-09-01 22:01:14 +0000226 switch (S) {
227 case MCDisassembler::Fail:
228 case MCDisassembler::SoftFail:
James Molloyc047dca2011-09-01 18:02:14 +0000229 // FIXME: Do something different on soft failure mode?
Sean Callananee5dfd42010-02-01 08:49:35 +0000230 delete inst;
231 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000232
James Molloyee064432011-09-01 22:01:14 +0000233 case MCDisassembler::Success: {
234 const llvm::EDInstInfo *thisInstInfo = NULL;
235
236 if (InstInfos) {
237 thisInstInfo = &InstInfos[inst->getOpcode()];
238 }
239
240 EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
241 return sdInst;
242 }
243 }
244 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000245}
246
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000247void EDDisassembler::initMaps(const MCRegisterInfo &registerInfo) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000248 unsigned numRegisters = registerInfo.getNumRegs();
249 unsigned registerIndex;
250
251 for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
252 const char* registerName = registerInfo.get(registerIndex).Name;
253
254 RegVec.push_back(registerName);
255 RegRMap[registerName] = registerIndex;
256 }
257
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000258 switch (TgtTriple.getArch()) {
Sean Callanan8f993b82010-04-08 00:48:21 +0000259 default:
260 break;
261 case Triple::x86:
262 case Triple::x86_64:
Sean Callananee5dfd42010-02-01 08:49:35 +0000263 stackPointers.insert(registerIDWithName("SP"));
264 stackPointers.insert(registerIDWithName("ESP"));
265 stackPointers.insert(registerIDWithName("RSP"));
266
267 programCounters.insert(registerIDWithName("IP"));
268 programCounters.insert(registerIDWithName("EIP"));
269 programCounters.insert(registerIDWithName("RIP"));
Sean Callanan8f993b82010-04-08 00:48:21 +0000270 break;
271 case Triple::arm:
272 case Triple::thumb:
273 stackPointers.insert(registerIDWithName("SP"));
274
275 programCounters.insert(registerIDWithName("PC"));
276 break;
Sean Callananee5dfd42010-02-01 08:49:35 +0000277 }
278}
279
280const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
281 if (registerID >= RegVec.size())
282 return NULL;
283 else
284 return RegVec[registerID].c_str();
285}
286
287unsigned EDDisassembler::registerIDWithName(const char *name) const {
288 regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
289 if (iter == RegRMap.end())
290 return 0;
291 else
292 return (*iter).second;
293}
294
295bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
296 return (stackPointers.find(registerID) != stackPointers.end());
297}
298
299bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
300 return (programCounters.find(registerID) != programCounters.end());
301}
302
Chris Lattnerd3740872010-04-04 05:04:31 +0000303int EDDisassembler::printInst(std::string &str, MCInst &inst) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000304 PrinterMutex.acquire();
305
Owen Anderson98c5dda2011-09-15 23:38:46 +0000306 InstPrinter->printInst(&inst, *InstStream, "");
Sean Callananee5dfd42010-02-01 08:49:35 +0000307 InstStream->flush();
308 str = *InstString;
309 InstString->clear();
310
311 PrinterMutex.release();
312
313 return 0;
314}
315
Chris Lattnerd8b7aa22011-10-16 04:47:35 +0000316static void diag_handler(const SMDiagnostic &diag, void *context) {
317 if (context)
318 diag.print("", static_cast<EDDisassembler*>(context)->ErrorStream);
Sean Callanan457856c2011-04-09 00:37:25 +0000319}
320
Sean Callananee5dfd42010-02-01 08:49:35 +0000321int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
322 SmallVectorImpl<AsmToken> &tokens,
323 const std::string &str) {
324 int ret = 0;
325
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000326 switch (TgtTriple.getArch()) {
Sean Callanan8f993b82010-04-08 00:48:21 +0000327 default:
328 return -1;
329 case Triple::x86:
330 case Triple::x86_64:
331 case Triple::arm:
332 case Triple::thumb:
333 break;
334 }
335
Sean Callananee5dfd42010-02-01 08:49:35 +0000336 const char *cStr = str.c_str();
337 MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
338
339 StringRef instName;
340 SMLoc instLoc;
341
342 SourceMgr sourceMgr;
Sean Callanan457856c2011-04-09 00:37:25 +0000343 sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this));
Sean Callananee5dfd42010-02-01 08:49:35 +0000344 sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
Evan Cheng203576a2011-07-20 19:50:42 +0000345 MCContext context(*AsmInfo, *MRI, NULL);
Chris Lattnerc18409a2010-03-11 22:53:35 +0000346 OwningPtr<MCStreamer> streamer(createNullStreamer(context));
Jim Grosbach1b84cce2011-08-16 18:33:49 +0000347 OwningPtr<MCAsmParser> genericParser(createMCAsmParser(sourceMgr,
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000348 context, *streamer,
349 *AsmInfo));
Evan Chengebdeeab2011-07-08 01:53:10 +0000350
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000351 OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(Key.Triple.c_str(), "", ""));
Evan Cheng94b95502011-07-26 00:24:13 +0000352 OwningPtr<MCTargetAsmParser>
353 TargetParser(Tgt->createMCAsmParser(*STI, *genericParser));
Sean Callananee5dfd42010-02-01 08:49:35 +0000354
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000355 AsmToken OpcodeToken = genericParser->Lex();
356 AsmToken NextToken = genericParser->Lex(); // consume next token, because specificParser expects us to
Sean Callanana8702562010-04-24 01:00:16 +0000357
Sean Callanan8f993b82010-04-08 00:48:21 +0000358 if (OpcodeToken.is(AsmToken::Identifier)) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000359 instName = OpcodeToken.getString();
360 instLoc = OpcodeToken.getLoc();
Sean Callanana8702562010-04-24 01:00:16 +0000361
362 if (NextToken.isNot(AsmToken::Eof) &&
363 TargetParser->ParseInstruction(instName, instLoc, operands))
Sean Callananee5dfd42010-02-01 08:49:35 +0000364 ret = -1;
Sean Callanan8f993b82010-04-08 00:48:21 +0000365 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000366 ret = -1;
367 }
368
Sean Callananee5dfd42010-02-01 08:49:35 +0000369 ParserMutex.acquire();
370
371 if (!ret) {
372 GenericAsmLexer->setBuffer(buf);
373
374 while (SpecificAsmLexer->Lex(),
375 SpecificAsmLexer->isNot(AsmToken::Eof) &&
376 SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
377 if (SpecificAsmLexer->is(AsmToken::Error)) {
378 ret = -1;
379 break;
380 }
381 tokens.push_back(SpecificAsmLexer->getTok());
382 }
383 }
384
385 ParserMutex.release();
386
387 return ret;
388}
389
390int EDDisassembler::llvmSyntaxVariant() const {
391 return LLVMSyntaxVariant;
392}