blob: 68a201b96a91751c7773c4bff134a133c039b5da [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 Molloyee064432011-09-01 22:01:14 +0000242 MCDisassembler::DecodeStatus S;
243 S = Disassembler->getInstruction(*inst, byteSize, memoryObject, address,
244 ErrorStream);
245 switch (S) {
246 case MCDisassembler::Fail:
247 case MCDisassembler::SoftFail:
James Molloyc047dca2011-09-01 18:02:14 +0000248 // FIXME: Do something different on soft failure mode?
Sean Callananee5dfd42010-02-01 08:49:35 +0000249 delete inst;
250 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000251
James Molloyee064432011-09-01 22:01:14 +0000252 case MCDisassembler::Success: {
253 const llvm::EDInstInfo *thisInstInfo = NULL;
254
255 if (InstInfos) {
256 thisInstInfo = &InstInfos[inst->getOpcode()];
257 }
258
259 EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
260 return sdInst;
261 }
262 }
263 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000264}
265
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000266void EDDisassembler::initMaps(const MCRegisterInfo &registerInfo) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000267 unsigned numRegisters = registerInfo.getNumRegs();
268 unsigned registerIndex;
269
270 for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
271 const char* registerName = registerInfo.get(registerIndex).Name;
272
273 RegVec.push_back(registerName);
274 RegRMap[registerName] = registerIndex;
275 }
276
Sean Callanan8f993b82010-04-08 00:48:21 +0000277 switch (Key.Arch) {
278 default:
279 break;
280 case Triple::x86:
281 case Triple::x86_64:
Sean Callananee5dfd42010-02-01 08:49:35 +0000282 stackPointers.insert(registerIDWithName("SP"));
283 stackPointers.insert(registerIDWithName("ESP"));
284 stackPointers.insert(registerIDWithName("RSP"));
285
286 programCounters.insert(registerIDWithName("IP"));
287 programCounters.insert(registerIDWithName("EIP"));
288 programCounters.insert(registerIDWithName("RIP"));
Sean Callanan8f993b82010-04-08 00:48:21 +0000289 break;
290 case Triple::arm:
291 case Triple::thumb:
292 stackPointers.insert(registerIDWithName("SP"));
293
294 programCounters.insert(registerIDWithName("PC"));
295 break;
Sean Callananee5dfd42010-02-01 08:49:35 +0000296 }
297}
298
299const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
300 if (registerID >= RegVec.size())
301 return NULL;
302 else
303 return RegVec[registerID].c_str();
304}
305
306unsigned EDDisassembler::registerIDWithName(const char *name) const {
307 regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
308 if (iter == RegRMap.end())
309 return 0;
310 else
311 return (*iter).second;
312}
313
314bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
315 return (stackPointers.find(registerID) != stackPointers.end());
316}
317
318bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
319 return (programCounters.find(registerID) != programCounters.end());
320}
321
Chris Lattnerd3740872010-04-04 05:04:31 +0000322int EDDisassembler::printInst(std::string &str, MCInst &inst) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000323 PrinterMutex.acquire();
324
Chris Lattnerd3740872010-04-04 05:04:31 +0000325 InstPrinter->printInst(&inst, *InstStream);
Sean Callananee5dfd42010-02-01 08:49:35 +0000326 InstStream->flush();
327 str = *InstString;
328 InstString->clear();
329
330 PrinterMutex.release();
331
332 return 0;
333}
334
Sean Callanan457856c2011-04-09 00:37:25 +0000335static void diag_handler(const SMDiagnostic &diag,
336 void *context)
337{
338 if (context) {
339 EDDisassembler *disassembler = static_cast<EDDisassembler*>(context);
340 diag.Print("", disassembler->ErrorStream);
341 }
342}
343
Sean Callananee5dfd42010-02-01 08:49:35 +0000344int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
345 SmallVectorImpl<AsmToken> &tokens,
346 const std::string &str) {
347 int ret = 0;
348
Sean Callanan8f993b82010-04-08 00:48:21 +0000349 switch (Key.Arch) {
350 default:
351 return -1;
352 case Triple::x86:
353 case Triple::x86_64:
354 case Triple::arm:
355 case Triple::thumb:
356 break;
357 }
358
Sean Callananee5dfd42010-02-01 08:49:35 +0000359 const char *cStr = str.c_str();
360 MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
361
362 StringRef instName;
363 SMLoc instLoc;
364
365 SourceMgr sourceMgr;
Sean Callanan457856c2011-04-09 00:37:25 +0000366 sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this));
Sean Callananee5dfd42010-02-01 08:49:35 +0000367 sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
Evan Cheng203576a2011-07-20 19:50:42 +0000368 MCContext context(*AsmInfo, *MRI, NULL);
Chris Lattnerc18409a2010-03-11 22:53:35 +0000369 OwningPtr<MCStreamer> streamer(createNullStreamer(context));
Jim Grosbach1b84cce2011-08-16 18:33:49 +0000370 OwningPtr<MCAsmParser> genericParser(createMCAsmParser(sourceMgr,
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000371 context, *streamer,
372 *AsmInfo));
Evan Chengebdeeab2011-07-08 01:53:10 +0000373
374 StringRef triple = tripleFromArch(Key.Arch);
Evan Chengffc0e732011-07-09 05:47:46 +0000375 OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(triple, "", ""));
Evan Cheng94b95502011-07-26 00:24:13 +0000376 OwningPtr<MCTargetAsmParser>
377 TargetParser(Tgt->createMCAsmParser(*STI, *genericParser));
Sean Callananee5dfd42010-02-01 08:49:35 +0000378
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000379 AsmToken OpcodeToken = genericParser->Lex();
380 AsmToken NextToken = genericParser->Lex(); // consume next token, because specificParser expects us to
Sean Callanana8702562010-04-24 01:00:16 +0000381
Sean Callanan8f993b82010-04-08 00:48:21 +0000382 if (OpcodeToken.is(AsmToken::Identifier)) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000383 instName = OpcodeToken.getString();
384 instLoc = OpcodeToken.getLoc();
Sean Callanana8702562010-04-24 01:00:16 +0000385
386 if (NextToken.isNot(AsmToken::Eof) &&
387 TargetParser->ParseInstruction(instName, instLoc, operands))
Sean Callananee5dfd42010-02-01 08:49:35 +0000388 ret = -1;
Sean Callanan8f993b82010-04-08 00:48:21 +0000389 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000390 ret = -1;
391 }
392
Sean Callananee5dfd42010-02-01 08:49:35 +0000393 ParserMutex.acquire();
394
395 if (!ret) {
396 GenericAsmLexer->setBuffer(buf);
397
398 while (SpecificAsmLexer->Lex(),
399 SpecificAsmLexer->isNot(AsmToken::Eof) &&
400 SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
401 if (SpecificAsmLexer->is(AsmToken::Error)) {
402 ret = -1;
403 break;
404 }
405 tokens.push_back(SpecificAsmLexer->getTok());
406 }
407 }
408
409 ParserMutex.release();
410
411 return ret;
412}
413
414int EDDisassembler::llvmSyntaxVariant() const {
415 return LLVMSyntaxVariant;
416}