blob: 83362a21f77b32a445e05943d48f24c7b855f756 [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
James Molloyb9505852011-09-07 17:24:38 +0000181 STI.reset(Tgt->createMCSubtargetInfo(tripleString, "", ""));
182
183 if (!STI)
184 return;
185
186 Disassembler.reset(Tgt->createMCDisassembler(*STI));
Sean Callananee5dfd42010-02-01 08:49:35 +0000187
188 if (!Disassembler)
189 return;
Sean Callanan9899f702010-04-13 21:21:57 +0000190
191 InstInfos = Disassembler->getEDInfo();
Sean Callananee5dfd42010-02-01 08:49:35 +0000192
193 InstString.reset(new std::string);
194 InstStream.reset(new raw_string_ostream(*InstString));
James Molloyb9505852011-09-07 17:24:38 +0000195 InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo, *STI));
Sean Callananee5dfd42010-02-01 08:49:35 +0000196
197 if (!InstPrinter)
198 return;
199
200 GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
Evan Cheng94b95502011-07-26 00:24:13 +0000201 SpecificAsmLexer.reset(Tgt->createMCAsmLexer(*MRI, *AsmInfo));
Sean Callananee5dfd42010-02-01 08:49:35 +0000202 SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
Sean Callanand74667e2010-02-02 02:18:20 +0000203
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000204 initMaps(*MRI);
Sean Callananee5dfd42010-02-01 08:49:35 +0000205
206 Valid = true;
207}
208
209EDDisassembler::~EDDisassembler() {
Sean Callanan8f993b82010-04-08 00:48:21 +0000210 if (!valid())
Sean Callananee5dfd42010-02-01 08:49:35 +0000211 return;
212}
213
214namespace {
215 /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
216 /// as provided by the sd interface. See MemoryObject.
217 class EDMemoryObject : public llvm::MemoryObject {
218 private:
219 EDByteReaderCallback Callback;
220 void *Arg;
221 public:
222 EDMemoryObject(EDByteReaderCallback callback,
223 void *arg) : Callback(callback), Arg(arg) { }
224 ~EDMemoryObject() { }
225 uint64_t getBase() const { return 0x0; }
226 uint64_t getExtent() const { return (uint64_t)-1; }
227 int readByte(uint64_t address, uint8_t *ptr) const {
Sean Callanan8f993b82010-04-08 00:48:21 +0000228 if (!Callback)
Sean Callananee5dfd42010-02-01 08:49:35 +0000229 return -1;
230
Sean Callanan8f993b82010-04-08 00:48:21 +0000231 if (Callback(ptr, address, Arg))
Sean Callananee5dfd42010-02-01 08:49:35 +0000232 return -1;
233
234 return 0;
235 }
236 };
237}
238
239EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
240 uint64_t address,
241 void *arg) {
242 EDMemoryObject memoryObject(byteReader, arg);
243
244 MCInst* inst = new MCInst;
245 uint64_t byteSize;
246
James Molloyee064432011-09-01 22:01:14 +0000247 MCDisassembler::DecodeStatus S;
248 S = Disassembler->getInstruction(*inst, byteSize, memoryObject, address,
Owen Anderson98c5dda2011-09-15 23:38:46 +0000249 ErrorStream, nulls());
James Molloyee064432011-09-01 22:01:14 +0000250 switch (S) {
251 case MCDisassembler::Fail:
252 case MCDisassembler::SoftFail:
James Molloyc047dca2011-09-01 18:02:14 +0000253 // FIXME: Do something different on soft failure mode?
Sean Callananee5dfd42010-02-01 08:49:35 +0000254 delete inst;
255 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000256
James Molloyee064432011-09-01 22:01:14 +0000257 case MCDisassembler::Success: {
258 const llvm::EDInstInfo *thisInstInfo = NULL;
259
260 if (InstInfos) {
261 thisInstInfo = &InstInfos[inst->getOpcode()];
262 }
263
264 EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
265 return sdInst;
266 }
267 }
268 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000269}
270
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000271void EDDisassembler::initMaps(const MCRegisterInfo &registerInfo) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000272 unsigned numRegisters = registerInfo.getNumRegs();
273 unsigned registerIndex;
274
275 for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
276 const char* registerName = registerInfo.get(registerIndex).Name;
277
278 RegVec.push_back(registerName);
279 RegRMap[registerName] = registerIndex;
280 }
281
Sean Callanan8f993b82010-04-08 00:48:21 +0000282 switch (Key.Arch) {
283 default:
284 break;
285 case Triple::x86:
286 case Triple::x86_64:
Sean Callananee5dfd42010-02-01 08:49:35 +0000287 stackPointers.insert(registerIDWithName("SP"));
288 stackPointers.insert(registerIDWithName("ESP"));
289 stackPointers.insert(registerIDWithName("RSP"));
290
291 programCounters.insert(registerIDWithName("IP"));
292 programCounters.insert(registerIDWithName("EIP"));
293 programCounters.insert(registerIDWithName("RIP"));
Sean Callanan8f993b82010-04-08 00:48:21 +0000294 break;
295 case Triple::arm:
296 case Triple::thumb:
297 stackPointers.insert(registerIDWithName("SP"));
298
299 programCounters.insert(registerIDWithName("PC"));
300 break;
Sean Callananee5dfd42010-02-01 08:49:35 +0000301 }
302}
303
304const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
305 if (registerID >= RegVec.size())
306 return NULL;
307 else
308 return RegVec[registerID].c_str();
309}
310
311unsigned EDDisassembler::registerIDWithName(const char *name) const {
312 regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
313 if (iter == RegRMap.end())
314 return 0;
315 else
316 return (*iter).second;
317}
318
319bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
320 return (stackPointers.find(registerID) != stackPointers.end());
321}
322
323bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
324 return (programCounters.find(registerID) != programCounters.end());
325}
326
Chris Lattnerd3740872010-04-04 05:04:31 +0000327int EDDisassembler::printInst(std::string &str, MCInst &inst) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000328 PrinterMutex.acquire();
329
Owen Anderson98c5dda2011-09-15 23:38:46 +0000330 InstPrinter->printInst(&inst, *InstStream, "");
Sean Callananee5dfd42010-02-01 08:49:35 +0000331 InstStream->flush();
332 str = *InstString;
333 InstString->clear();
334
335 PrinterMutex.release();
336
337 return 0;
338}
339
Sean Callanan457856c2011-04-09 00:37:25 +0000340static void diag_handler(const SMDiagnostic &diag,
341 void *context)
342{
343 if (context) {
344 EDDisassembler *disassembler = static_cast<EDDisassembler*>(context);
345 diag.Print("", disassembler->ErrorStream);
346 }
347}
348
Sean Callananee5dfd42010-02-01 08:49:35 +0000349int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
350 SmallVectorImpl<AsmToken> &tokens,
351 const std::string &str) {
352 int ret = 0;
353
Sean Callanan8f993b82010-04-08 00:48:21 +0000354 switch (Key.Arch) {
355 default:
356 return -1;
357 case Triple::x86:
358 case Triple::x86_64:
359 case Triple::arm:
360 case Triple::thumb:
361 break;
362 }
363
Sean Callananee5dfd42010-02-01 08:49:35 +0000364 const char *cStr = str.c_str();
365 MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
366
367 StringRef instName;
368 SMLoc instLoc;
369
370 SourceMgr sourceMgr;
Sean Callanan457856c2011-04-09 00:37:25 +0000371 sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this));
Sean Callananee5dfd42010-02-01 08:49:35 +0000372 sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
Evan Cheng203576a2011-07-20 19:50:42 +0000373 MCContext context(*AsmInfo, *MRI, NULL);
Chris Lattnerc18409a2010-03-11 22:53:35 +0000374 OwningPtr<MCStreamer> streamer(createNullStreamer(context));
Jim Grosbach1b84cce2011-08-16 18:33:49 +0000375 OwningPtr<MCAsmParser> genericParser(createMCAsmParser(sourceMgr,
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000376 context, *streamer,
377 *AsmInfo));
Evan Chengebdeeab2011-07-08 01:53:10 +0000378
379 StringRef triple = tripleFromArch(Key.Arch);
Evan Chengffc0e732011-07-09 05:47:46 +0000380 OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(triple, "", ""));
Evan Cheng94b95502011-07-26 00:24:13 +0000381 OwningPtr<MCTargetAsmParser>
382 TargetParser(Tgt->createMCAsmParser(*STI, *genericParser));
Sean Callananee5dfd42010-02-01 08:49:35 +0000383
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000384 AsmToken OpcodeToken = genericParser->Lex();
385 AsmToken NextToken = genericParser->Lex(); // consume next token, because specificParser expects us to
Sean Callanana8702562010-04-24 01:00:16 +0000386
Sean Callanan8f993b82010-04-08 00:48:21 +0000387 if (OpcodeToken.is(AsmToken::Identifier)) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000388 instName = OpcodeToken.getString();
389 instLoc = OpcodeToken.getLoc();
Sean Callanana8702562010-04-24 01:00:16 +0000390
391 if (NextToken.isNot(AsmToken::Eof) &&
392 TargetParser->ParseInstruction(instName, instLoc, operands))
Sean Callananee5dfd42010-02-01 08:49:35 +0000393 ret = -1;
Sean Callanan8f993b82010-04-08 00:48:21 +0000394 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000395 ret = -1;
396 }
397
Sean Callananee5dfd42010-02-01 08:49:35 +0000398 ParserMutex.acquire();
399
400 if (!ret) {
401 GenericAsmLexer->setBuffer(buf);
402
403 while (SpecificAsmLexer->Lex(),
404 SpecificAsmLexer->isNot(AsmToken::Eof) &&
405 SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
406 if (SpecificAsmLexer->is(AsmToken::Error)) {
407 ret = -1;
408 break;
409 }
410 tokens.push_back(SpecificAsmLexer->getTok());
411 }
412 }
413
414 ParserMutex.release();
415
416 return ret;
417}
418
419int EDDisassembler::llvmSyntaxVariant() const {
420 return LLVMSyntaxVariant;
421}