blob: ea77902112c46a7cae71aa428faaae1e5a08884e [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" },
50 { Triple::thumb, "thumb-unknown-unknown" },
51 { Triple::InvalidArch, NULL, }
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()) {
131 if (TgtTriple.getArch() == Triple::InvalidArch)
Sean Callanan9899f702010-04-13 21:21:57 +0000132 return;
Sean Callananee5dfd42010-02-01 08:49:35 +0000133
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000134 LLVMSyntaxVariant = getLLVMSyntaxVariant(TgtTriple.getArch(), key.Syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000135
Sean Callanan4285b292010-04-09 00:11:15 +0000136 if (LLVMSyntaxVariant < 0)
Sean Callananee5dfd42010-02-01 08:49:35 +0000137 return;
138
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000139 std::string tripleString(key.Triple);
Sean Callananee5dfd42010-02-01 08:49:35 +0000140 std::string errorString;
141
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000142 Tgt = TargetRegistry::lookupTarget(key.Triple,
Sean Callananee5dfd42010-02-01 08:49:35 +0000143 errorString);
144
145 if (!Tgt)
146 return;
147
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000148 MRI.reset(Tgt->createMCRegInfo(tripleString));
Evan Chengebdeeab2011-07-08 01:53:10 +0000149
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000150 if (!MRI)
Sean Callananee5dfd42010-02-01 08:49:35 +0000151 return;
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000152
153 initMaps(*MRI);
Sean Callananee5dfd42010-02-01 08:49:35 +0000154
Evan Cheng1abf2cb2011-07-14 23:50:31 +0000155 AsmInfo.reset(Tgt->createMCAsmInfo(tripleString));
Sean Callananee5dfd42010-02-01 08:49:35 +0000156
157 if (!AsmInfo)
158 return;
Evan Chengebdeeab2011-07-08 01:53:10 +0000159
James Molloyb9505852011-09-07 17:24:38 +0000160 STI.reset(Tgt->createMCSubtargetInfo(tripleString, "", ""));
161
162 if (!STI)
163 return;
164
165 Disassembler.reset(Tgt->createMCDisassembler(*STI));
Sean Callananee5dfd42010-02-01 08:49:35 +0000166
167 if (!Disassembler)
168 return;
Sean Callanan9899f702010-04-13 21:21:57 +0000169
170 InstInfos = Disassembler->getEDInfo();
Sean Callananee5dfd42010-02-01 08:49:35 +0000171
172 InstString.reset(new std::string);
173 InstStream.reset(new raw_string_ostream(*InstString));
James Molloyb9505852011-09-07 17:24:38 +0000174 InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo, *STI));
Sean Callananee5dfd42010-02-01 08:49:35 +0000175
176 if (!InstPrinter)
177 return;
178
179 GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
Evan Cheng94b95502011-07-26 00:24:13 +0000180 SpecificAsmLexer.reset(Tgt->createMCAsmLexer(*MRI, *AsmInfo));
Sean Callananee5dfd42010-02-01 08:49:35 +0000181 SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
Sean Callanand74667e2010-02-02 02:18:20 +0000182
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000183 initMaps(*MRI);
Sean Callananee5dfd42010-02-01 08:49:35 +0000184
185 Valid = true;
186}
187
188EDDisassembler::~EDDisassembler() {
Sean Callanan8f993b82010-04-08 00:48:21 +0000189 if (!valid())
Sean Callananee5dfd42010-02-01 08:49:35 +0000190 return;
191}
192
193namespace {
194 /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
195 /// as provided by the sd interface. See MemoryObject.
196 class EDMemoryObject : public llvm::MemoryObject {
197 private:
198 EDByteReaderCallback Callback;
199 void *Arg;
200 public:
201 EDMemoryObject(EDByteReaderCallback callback,
202 void *arg) : Callback(callback), Arg(arg) { }
203 ~EDMemoryObject() { }
204 uint64_t getBase() const { return 0x0; }
Derek Schuff2ea93872012-02-06 22:30:29 +0000205 uint64_t getExtent() { return (uint64_t)-1; }
206 int readByte(uint64_t address, uint8_t *ptr) {
Sean Callanan8f993b82010-04-08 00:48:21 +0000207 if (!Callback)
Sean Callananee5dfd42010-02-01 08:49:35 +0000208 return -1;
209
Sean Callanan8f993b82010-04-08 00:48:21 +0000210 if (Callback(ptr, address, Arg))
Sean Callananee5dfd42010-02-01 08:49:35 +0000211 return -1;
212
213 return 0;
214 }
215 };
216}
217
218EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
219 uint64_t address,
220 void *arg) {
221 EDMemoryObject memoryObject(byteReader, arg);
222
223 MCInst* inst = new MCInst;
224 uint64_t byteSize;
225
James Molloyee064432011-09-01 22:01:14 +0000226 MCDisassembler::DecodeStatus S;
227 S = Disassembler->getInstruction(*inst, byteSize, memoryObject, address,
Owen Anderson98c5dda2011-09-15 23:38:46 +0000228 ErrorStream, nulls());
James Molloyee064432011-09-01 22:01:14 +0000229 switch (S) {
230 case MCDisassembler::Fail:
231 case MCDisassembler::SoftFail:
James Molloyc047dca2011-09-01 18:02:14 +0000232 // FIXME: Do something different on soft failure mode?
Sean Callananee5dfd42010-02-01 08:49:35 +0000233 delete inst;
234 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000235
James Molloyee064432011-09-01 22:01:14 +0000236 case MCDisassembler::Success: {
237 const llvm::EDInstInfo *thisInstInfo = NULL;
238
239 if (InstInfos) {
240 thisInstInfo = &InstInfos[inst->getOpcode()];
241 }
242
243 EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
244 return sdInst;
245 }
246 }
247 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000248}
249
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000250void EDDisassembler::initMaps(const MCRegisterInfo &registerInfo) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000251 unsigned numRegisters = registerInfo.getNumRegs();
252 unsigned registerIndex;
253
254 for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
255 const char* registerName = registerInfo.get(registerIndex).Name;
256
257 RegVec.push_back(registerName);
258 RegRMap[registerName] = registerIndex;
259 }
260
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000261 switch (TgtTriple.getArch()) {
Sean Callanan8f993b82010-04-08 00:48:21 +0000262 default:
263 break;
264 case Triple::x86:
265 case Triple::x86_64:
Sean Callananee5dfd42010-02-01 08:49:35 +0000266 stackPointers.insert(registerIDWithName("SP"));
267 stackPointers.insert(registerIDWithName("ESP"));
268 stackPointers.insert(registerIDWithName("RSP"));
269
270 programCounters.insert(registerIDWithName("IP"));
271 programCounters.insert(registerIDWithName("EIP"));
272 programCounters.insert(registerIDWithName("RIP"));
Sean Callanan8f993b82010-04-08 00:48:21 +0000273 break;
274 case Triple::arm:
275 case Triple::thumb:
276 stackPointers.insert(registerIDWithName("SP"));
277
278 programCounters.insert(registerIDWithName("PC"));
279 break;
Sean Callananee5dfd42010-02-01 08:49:35 +0000280 }
281}
282
283const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
284 if (registerID >= RegVec.size())
285 return NULL;
286 else
287 return RegVec[registerID].c_str();
288}
289
290unsigned EDDisassembler::registerIDWithName(const char *name) const {
291 regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
292 if (iter == RegRMap.end())
293 return 0;
294 else
295 return (*iter).second;
296}
297
298bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
299 return (stackPointers.find(registerID) != stackPointers.end());
300}
301
302bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
303 return (programCounters.find(registerID) != programCounters.end());
304}
305
Chris Lattnerd3740872010-04-04 05:04:31 +0000306int EDDisassembler::printInst(std::string &str, MCInst &inst) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000307 PrinterMutex.acquire();
308
Owen Anderson98c5dda2011-09-15 23:38:46 +0000309 InstPrinter->printInst(&inst, *InstStream, "");
Sean Callananee5dfd42010-02-01 08:49:35 +0000310 InstStream->flush();
311 str = *InstString;
312 InstString->clear();
313
314 PrinterMutex.release();
315
316 return 0;
317}
318
Chris Lattnerd8b7aa22011-10-16 04:47:35 +0000319static void diag_handler(const SMDiagnostic &diag, void *context) {
320 if (context)
321 diag.print("", static_cast<EDDisassembler*>(context)->ErrorStream);
Sean Callanan457856c2011-04-09 00:37:25 +0000322}
323
Sean Callananee5dfd42010-02-01 08:49:35 +0000324int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
325 SmallVectorImpl<AsmToken> &tokens,
326 const std::string &str) {
327 int ret = 0;
328
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000329 switch (TgtTriple.getArch()) {
Sean Callanan8f993b82010-04-08 00:48:21 +0000330 default:
331 return -1;
332 case Triple::x86:
333 case Triple::x86_64:
334 case Triple::arm:
335 case Triple::thumb:
336 break;
337 }
338
Sean Callananee5dfd42010-02-01 08:49:35 +0000339 const char *cStr = str.c_str();
340 MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
341
342 StringRef instName;
343 SMLoc instLoc;
344
345 SourceMgr sourceMgr;
Sean Callanan457856c2011-04-09 00:37:25 +0000346 sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this));
Sean Callananee5dfd42010-02-01 08:49:35 +0000347 sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
Evan Cheng203576a2011-07-20 19:50:42 +0000348 MCContext context(*AsmInfo, *MRI, NULL);
Chris Lattnerc18409a2010-03-11 22:53:35 +0000349 OwningPtr<MCStreamer> streamer(createNullStreamer(context));
Jim Grosbach1b84cce2011-08-16 18:33:49 +0000350 OwningPtr<MCAsmParser> genericParser(createMCAsmParser(sourceMgr,
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000351 context, *streamer,
352 *AsmInfo));
Evan Chengebdeeab2011-07-08 01:53:10 +0000353
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000354 OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(Key.Triple.c_str(), "", ""));
Evan Cheng94b95502011-07-26 00:24:13 +0000355 OwningPtr<MCTargetAsmParser>
356 TargetParser(Tgt->createMCAsmParser(*STI, *genericParser));
Sean Callananee5dfd42010-02-01 08:49:35 +0000357
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000358 AsmToken OpcodeToken = genericParser->Lex();
359 AsmToken NextToken = genericParser->Lex(); // consume next token, because specificParser expects us to
Sean Callanana8702562010-04-24 01:00:16 +0000360
Sean Callanan8f993b82010-04-08 00:48:21 +0000361 if (OpcodeToken.is(AsmToken::Identifier)) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000362 instName = OpcodeToken.getString();
363 instLoc = OpcodeToken.getLoc();
Sean Callanana8702562010-04-24 01:00:16 +0000364
365 if (NextToken.isNot(AsmToken::Eof) &&
366 TargetParser->ParseInstruction(instName, instLoc, operands))
Sean Callananee5dfd42010-02-01 08:49:35 +0000367 ret = -1;
Sean Callanan8f993b82010-04-08 00:48:21 +0000368 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000369 ret = -1;
370 }
371
Sean Callananee5dfd42010-02-01 08:49:35 +0000372 ParserMutex.acquire();
373
374 if (!ret) {
375 GenericAsmLexer->setBuffer(buf);
376
377 while (SpecificAsmLexer->Lex(),
378 SpecificAsmLexer->isNot(AsmToken::Eof) &&
379 SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
380 if (SpecificAsmLexer->is(AsmToken::Error)) {
381 ret = -1;
382 break;
383 }
384 tokens.push_back(SpecificAsmLexer->getTok());
385 }
386 }
387
388 ParserMutex.release();
389
390 return ret;
391}
392
393int EDDisassembler::llvmSyntaxVariant() const {
394 return LLVMSyntaxVariant;
395}