blob: 8698335733642d418beca3cfd4c8db0473b2bc10 [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;
82 else
83 return -1;
Chris Lattner847da552010-07-20 18:25:19 +000084 case EDDisassembler::kEDAssemblySyntaxX86Intel:
Sean Callananee5dfd42010-02-01 08:49:35 +000085 if (arch == Triple::x86 || arch == Triple::x86_64)
86 return 1;
87 else
88 return -1;
Chris Lattner847da552010-07-20 18:25:19 +000089 case EDDisassembler::kEDAssemblySyntaxARMUAL:
Sean Callanan8f993b82010-04-08 00:48:21 +000090 if (arch == Triple::arm || arch == Triple::thumb)
91 return 0;
92 else
93 return -1;
Sean Callananee5dfd42010-02-01 08:49:35 +000094 }
95}
96
Sean Callananee5dfd42010-02-01 08:49:35 +000097EDDisassembler *EDDisassembler::getDisassembler(Triple::ArchType arch,
Chris Lattner847da552010-07-20 18:25:19 +000098 AssemblySyntax syntax) {
Sean Callananee5dfd42010-02-01 08:49:35 +000099 CPUKey key;
100 key.Arch = arch;
101 key.Syntax = syntax;
102
103 EDDisassembler::DisassemblerMap_t::iterator i = sDisassemblers.find(key);
104
105 if (i != sDisassemblers.end()) {
106 return i->second;
Sean Callanan8f993b82010-04-08 00:48:21 +0000107 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000108 EDDisassembler* sdd = new EDDisassembler(key);
Sean Callanan8f993b82010-04-08 00:48:21 +0000109 if (!sdd->valid()) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000110 delete sdd;
111 return NULL;
112 }
113
114 sDisassemblers[key] = sdd;
115
116 return sdd;
117 }
118
119 return NULL;
120}
121
122EDDisassembler *EDDisassembler::getDisassembler(StringRef str,
Chris Lattner847da552010-07-20 18:25:19 +0000123 AssemblySyntax syntax) {
124 return getDisassembler(Triple(str).getArch(), syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000125}
126
Sean Callananee5dfd42010-02-01 08:49:35 +0000127EDDisassembler::EDDisassembler(CPUKey &key) :
Sean Callanan8f993b82010-04-08 00:48:21 +0000128 Valid(false),
129 HasSemantics(false),
130 ErrorStream(nulls()),
131 Key(key) {
Sean Callanan9899f702010-04-13 21:21:57 +0000132 const char *triple = tripleFromArch(key.Arch);
Sean Callanan8f993b82010-04-08 00:48:21 +0000133
Sean Callanan9899f702010-04-13 21:21:57 +0000134 if (!triple)
135 return;
Sean Callananee5dfd42010-02-01 08:49:35 +0000136
Sean Callanan4285b292010-04-09 00:11:15 +0000137 LLVMSyntaxVariant = getLLVMSyntaxVariant(key.Arch, key.Syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000138
Sean Callanan4285b292010-04-09 00:11:15 +0000139 if (LLVMSyntaxVariant < 0)
Sean Callananee5dfd42010-02-01 08:49:35 +0000140 return;
141
142 std::string tripleString(triple);
143 std::string errorString;
144
145 Tgt = TargetRegistry::lookupTarget(tripleString,
146 errorString);
147
148 if (!Tgt)
149 return;
150
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000151 MRI.reset(Tgt->createMCRegInfo(tripleString));
Evan Chengebdeeab2011-07-08 01:53:10 +0000152
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000153 if (!MRI)
Sean Callananee5dfd42010-02-01 08:49:35 +0000154 return;
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000155
156 initMaps(*MRI);
Sean Callananee5dfd42010-02-01 08:49:35 +0000157
Evan Cheng1abf2cb2011-07-14 23:50:31 +0000158 AsmInfo.reset(Tgt->createMCAsmInfo(tripleString));
Sean Callananee5dfd42010-02-01 08:49:35 +0000159
160 if (!AsmInfo)
161 return;
Evan Chengebdeeab2011-07-08 01:53:10 +0000162
James Molloyb9505852011-09-07 17:24:38 +0000163 STI.reset(Tgt->createMCSubtargetInfo(tripleString, "", ""));
164
165 if (!STI)
166 return;
167
168 Disassembler.reset(Tgt->createMCDisassembler(*STI));
Sean Callananee5dfd42010-02-01 08:49:35 +0000169
170 if (!Disassembler)
171 return;
Sean Callanan9899f702010-04-13 21:21:57 +0000172
173 InstInfos = Disassembler->getEDInfo();
Sean Callananee5dfd42010-02-01 08:49:35 +0000174
175 InstString.reset(new std::string);
176 InstStream.reset(new raw_string_ostream(*InstString));
James Molloyb9505852011-09-07 17:24:38 +0000177 InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo, *STI));
Sean Callananee5dfd42010-02-01 08:49:35 +0000178
179 if (!InstPrinter)
180 return;
181
182 GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
Evan Cheng94b95502011-07-26 00:24:13 +0000183 SpecificAsmLexer.reset(Tgt->createMCAsmLexer(*MRI, *AsmInfo));
Sean Callananee5dfd42010-02-01 08:49:35 +0000184 SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
Sean Callanand74667e2010-02-02 02:18:20 +0000185
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000186 initMaps(*MRI);
Sean Callananee5dfd42010-02-01 08:49:35 +0000187
188 Valid = true;
189}
190
191EDDisassembler::~EDDisassembler() {
Sean Callanan8f993b82010-04-08 00:48:21 +0000192 if (!valid())
Sean Callananee5dfd42010-02-01 08:49:35 +0000193 return;
194}
195
196namespace {
197 /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
198 /// as provided by the sd interface. See MemoryObject.
199 class EDMemoryObject : public llvm::MemoryObject {
200 private:
201 EDByteReaderCallback Callback;
202 void *Arg;
203 public:
204 EDMemoryObject(EDByteReaderCallback callback,
205 void *arg) : Callback(callback), Arg(arg) { }
206 ~EDMemoryObject() { }
207 uint64_t getBase() const { return 0x0; }
208 uint64_t getExtent() const { return (uint64_t)-1; }
209 int readByte(uint64_t address, uint8_t *ptr) const {
Sean Callanan8f993b82010-04-08 00:48:21 +0000210 if (!Callback)
Sean Callananee5dfd42010-02-01 08:49:35 +0000211 return -1;
212
Sean Callanan8f993b82010-04-08 00:48:21 +0000213 if (Callback(ptr, address, Arg))
Sean Callananee5dfd42010-02-01 08:49:35 +0000214 return -1;
215
216 return 0;
217 }
218 };
219}
220
221EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
222 uint64_t address,
223 void *arg) {
224 EDMemoryObject memoryObject(byteReader, arg);
225
226 MCInst* inst = new MCInst;
227 uint64_t byteSize;
228
James Molloyee064432011-09-01 22:01:14 +0000229 MCDisassembler::DecodeStatus S;
230 S = Disassembler->getInstruction(*inst, byteSize, memoryObject, address,
Owen Anderson98c5dda2011-09-15 23:38:46 +0000231 ErrorStream, nulls());
James Molloyee064432011-09-01 22:01:14 +0000232 switch (S) {
233 case MCDisassembler::Fail:
234 case MCDisassembler::SoftFail:
James Molloyc047dca2011-09-01 18:02:14 +0000235 // FIXME: Do something different on soft failure mode?
Sean Callananee5dfd42010-02-01 08:49:35 +0000236 delete inst;
237 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000238
James Molloyee064432011-09-01 22:01:14 +0000239 case MCDisassembler::Success: {
240 const llvm::EDInstInfo *thisInstInfo = NULL;
241
242 if (InstInfos) {
243 thisInstInfo = &InstInfos[inst->getOpcode()];
244 }
245
246 EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
247 return sdInst;
248 }
249 }
250 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000251}
252
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000253void EDDisassembler::initMaps(const MCRegisterInfo &registerInfo) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000254 unsigned numRegisters = registerInfo.getNumRegs();
255 unsigned registerIndex;
256
257 for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
258 const char* registerName = registerInfo.get(registerIndex).Name;
259
260 RegVec.push_back(registerName);
261 RegRMap[registerName] = registerIndex;
262 }
263
Sean Callanan8f993b82010-04-08 00:48:21 +0000264 switch (Key.Arch) {
265 default:
266 break;
267 case Triple::x86:
268 case Triple::x86_64:
Sean Callananee5dfd42010-02-01 08:49:35 +0000269 stackPointers.insert(registerIDWithName("SP"));
270 stackPointers.insert(registerIDWithName("ESP"));
271 stackPointers.insert(registerIDWithName("RSP"));
272
273 programCounters.insert(registerIDWithName("IP"));
274 programCounters.insert(registerIDWithName("EIP"));
275 programCounters.insert(registerIDWithName("RIP"));
Sean Callanan8f993b82010-04-08 00:48:21 +0000276 break;
277 case Triple::arm:
278 case Triple::thumb:
279 stackPointers.insert(registerIDWithName("SP"));
280
281 programCounters.insert(registerIDWithName("PC"));
282 break;
Sean Callananee5dfd42010-02-01 08:49:35 +0000283 }
284}
285
286const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
287 if (registerID >= RegVec.size())
288 return NULL;
289 else
290 return RegVec[registerID].c_str();
291}
292
293unsigned EDDisassembler::registerIDWithName(const char *name) const {
294 regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
295 if (iter == RegRMap.end())
296 return 0;
297 else
298 return (*iter).second;
299}
300
301bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
302 return (stackPointers.find(registerID) != stackPointers.end());
303}
304
305bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
306 return (programCounters.find(registerID) != programCounters.end());
307}
308
Chris Lattnerd3740872010-04-04 05:04:31 +0000309int EDDisassembler::printInst(std::string &str, MCInst &inst) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000310 PrinterMutex.acquire();
311
Owen Anderson98c5dda2011-09-15 23:38:46 +0000312 InstPrinter->printInst(&inst, *InstStream, "");
Sean Callananee5dfd42010-02-01 08:49:35 +0000313 InstStream->flush();
314 str = *InstString;
315 InstString->clear();
316
317 PrinterMutex.release();
318
319 return 0;
320}
321
Chris Lattnerd8b7aa22011-10-16 04:47:35 +0000322static void diag_handler(const SMDiagnostic &diag, void *context) {
323 if (context)
324 diag.print("", static_cast<EDDisassembler*>(context)->ErrorStream);
Sean Callanan457856c2011-04-09 00:37:25 +0000325}
326
Sean Callananee5dfd42010-02-01 08:49:35 +0000327int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
328 SmallVectorImpl<AsmToken> &tokens,
329 const std::string &str) {
330 int ret = 0;
331
Sean Callanan8f993b82010-04-08 00:48:21 +0000332 switch (Key.Arch) {
333 default:
334 return -1;
335 case Triple::x86:
336 case Triple::x86_64:
337 case Triple::arm:
338 case Triple::thumb:
339 break;
340 }
341
Sean Callananee5dfd42010-02-01 08:49:35 +0000342 const char *cStr = str.c_str();
343 MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
344
345 StringRef instName;
346 SMLoc instLoc;
347
348 SourceMgr sourceMgr;
Sean Callanan457856c2011-04-09 00:37:25 +0000349 sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this));
Sean Callananee5dfd42010-02-01 08:49:35 +0000350 sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
Evan Cheng203576a2011-07-20 19:50:42 +0000351 MCContext context(*AsmInfo, *MRI, NULL);
Chris Lattnerc18409a2010-03-11 22:53:35 +0000352 OwningPtr<MCStreamer> streamer(createNullStreamer(context));
Jim Grosbach1b84cce2011-08-16 18:33:49 +0000353 OwningPtr<MCAsmParser> genericParser(createMCAsmParser(sourceMgr,
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000354 context, *streamer,
355 *AsmInfo));
Evan Chengebdeeab2011-07-08 01:53:10 +0000356
357 StringRef triple = tripleFromArch(Key.Arch);
Evan Chengffc0e732011-07-09 05:47:46 +0000358 OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(triple, "", ""));
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}