blob: c7221d86a86efd722c91172e16fb5b301c573ae4 [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;
107
108 EDDisassembler::DisassemblerMap_t::iterator i = sDisassemblers.find(key);
Sean Callanan3f4f34c2012-02-04 07:45:35 +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 }
113 else {
114 EDDisassembler *sdd = new EDDisassembler(key);
Sean Callanan8f993b82010-04-08 00:48:21 +0000115 if (!sdd->valid()) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000116 delete sdd;
117 return NULL;
118 }
119
120 sDisassemblers[key] = sdd;
121
122 return sdd;
123 }
124
125 return NULL;
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000126
Chris Lattner847da552010-07-20 18:25:19 +0000127 return getDisassembler(Triple(str).getArch(), syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000128}
129
Sean Callananee5dfd42010-02-01 08:49:35 +0000130EDDisassembler::EDDisassembler(CPUKey &key) :
Sean Callanan8f993b82010-04-08 00:48:21 +0000131 Valid(false),
132 HasSemantics(false),
133 ErrorStream(nulls()),
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000134 Key(key),
135 TgtTriple(key.Triple.c_str()) {
136 if (TgtTriple.getArch() == Triple::InvalidArch)
Sean Callanan9899f702010-04-13 21:21:57 +0000137 return;
Sean Callananee5dfd42010-02-01 08:49:35 +0000138
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000139 LLVMSyntaxVariant = getLLVMSyntaxVariant(TgtTriple.getArch(), key.Syntax);
Sean Callananee5dfd42010-02-01 08:49:35 +0000140
Sean Callanan4285b292010-04-09 00:11:15 +0000141 if (LLVMSyntaxVariant < 0)
Sean Callananee5dfd42010-02-01 08:49:35 +0000142 return;
143
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000144 std::string tripleString(key.Triple);
Sean Callananee5dfd42010-02-01 08:49:35 +0000145 std::string errorString;
146
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000147 Tgt = TargetRegistry::lookupTarget(key.Triple,
Sean Callananee5dfd42010-02-01 08:49:35 +0000148 errorString);
149
150 if (!Tgt)
151 return;
152
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000153 MRI.reset(Tgt->createMCRegInfo(tripleString));
Evan Chengebdeeab2011-07-08 01:53:10 +0000154
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000155 if (!MRI)
Sean Callananee5dfd42010-02-01 08:49:35 +0000156 return;
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000157
158 initMaps(*MRI);
Sean Callananee5dfd42010-02-01 08:49:35 +0000159
Evan Cheng1abf2cb2011-07-14 23:50:31 +0000160 AsmInfo.reset(Tgt->createMCAsmInfo(tripleString));
Sean Callananee5dfd42010-02-01 08:49:35 +0000161
162 if (!AsmInfo)
163 return;
Evan Chengebdeeab2011-07-08 01:53:10 +0000164
James Molloyb9505852011-09-07 17:24:38 +0000165 STI.reset(Tgt->createMCSubtargetInfo(tripleString, "", ""));
166
167 if (!STI)
168 return;
169
170 Disassembler.reset(Tgt->createMCDisassembler(*STI));
Sean Callananee5dfd42010-02-01 08:49:35 +0000171
172 if (!Disassembler)
173 return;
Sean Callanan9899f702010-04-13 21:21:57 +0000174
175 InstInfos = Disassembler->getEDInfo();
Sean Callananee5dfd42010-02-01 08:49:35 +0000176
177 InstString.reset(new std::string);
178 InstStream.reset(new raw_string_ostream(*InstString));
James Molloyb9505852011-09-07 17:24:38 +0000179 InstPrinter.reset(Tgt->createMCInstPrinter(LLVMSyntaxVariant, *AsmInfo, *STI));
Sean Callananee5dfd42010-02-01 08:49:35 +0000180
181 if (!InstPrinter)
182 return;
183
184 GenericAsmLexer.reset(new AsmLexer(*AsmInfo));
Evan Cheng94b95502011-07-26 00:24:13 +0000185 SpecificAsmLexer.reset(Tgt->createMCAsmLexer(*MRI, *AsmInfo));
Sean Callananee5dfd42010-02-01 08:49:35 +0000186 SpecificAsmLexer->InstallLexer(*GenericAsmLexer);
Sean Callanand74667e2010-02-02 02:18:20 +0000187
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000188 initMaps(*MRI);
Sean Callananee5dfd42010-02-01 08:49:35 +0000189
190 Valid = true;
191}
192
193EDDisassembler::~EDDisassembler() {
Sean Callanan8f993b82010-04-08 00:48:21 +0000194 if (!valid())
Sean Callananee5dfd42010-02-01 08:49:35 +0000195 return;
196}
197
198namespace {
199 /// EDMemoryObject - a subclass of MemoryObject that allows use of a callback
200 /// as provided by the sd interface. See MemoryObject.
201 class EDMemoryObject : public llvm::MemoryObject {
202 private:
203 EDByteReaderCallback Callback;
204 void *Arg;
205 public:
206 EDMemoryObject(EDByteReaderCallback callback,
207 void *arg) : Callback(callback), Arg(arg) { }
208 ~EDMemoryObject() { }
209 uint64_t getBase() const { return 0x0; }
210 uint64_t getExtent() const { return (uint64_t)-1; }
211 int readByte(uint64_t address, uint8_t *ptr) const {
Sean Callanan8f993b82010-04-08 00:48:21 +0000212 if (!Callback)
Sean Callananee5dfd42010-02-01 08:49:35 +0000213 return -1;
214
Sean Callanan8f993b82010-04-08 00:48:21 +0000215 if (Callback(ptr, address, Arg))
Sean Callananee5dfd42010-02-01 08:49:35 +0000216 return -1;
217
218 return 0;
219 }
220 };
221}
222
223EDInst *EDDisassembler::createInst(EDByteReaderCallback byteReader,
224 uint64_t address,
225 void *arg) {
226 EDMemoryObject memoryObject(byteReader, arg);
227
228 MCInst* inst = new MCInst;
229 uint64_t byteSize;
230
James Molloyee064432011-09-01 22:01:14 +0000231 MCDisassembler::DecodeStatus S;
232 S = Disassembler->getInstruction(*inst, byteSize, memoryObject, address,
Owen Anderson98c5dda2011-09-15 23:38:46 +0000233 ErrorStream, nulls());
James Molloyee064432011-09-01 22:01:14 +0000234 switch (S) {
235 case MCDisassembler::Fail:
236 case MCDisassembler::SoftFail:
James Molloyc047dca2011-09-01 18:02:14 +0000237 // FIXME: Do something different on soft failure mode?
Sean Callananee5dfd42010-02-01 08:49:35 +0000238 delete inst;
239 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000240
James Molloyee064432011-09-01 22:01:14 +0000241 case MCDisassembler::Success: {
242 const llvm::EDInstInfo *thisInstInfo = NULL;
243
244 if (InstInfos) {
245 thisInstInfo = &InstInfos[inst->getOpcode()];
246 }
247
248 EDInst* sdInst = new EDInst(inst, byteSize, *this, thisInstInfo);
249 return sdInst;
250 }
251 }
252 return NULL;
Sean Callananee5dfd42010-02-01 08:49:35 +0000253}
254
Evan Cheng1b0fc9b2011-07-25 20:53:02 +0000255void EDDisassembler::initMaps(const MCRegisterInfo &registerInfo) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000256 unsigned numRegisters = registerInfo.getNumRegs();
257 unsigned registerIndex;
258
259 for (registerIndex = 0; registerIndex < numRegisters; ++registerIndex) {
260 const char* registerName = registerInfo.get(registerIndex).Name;
261
262 RegVec.push_back(registerName);
263 RegRMap[registerName] = registerIndex;
264 }
265
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000266 switch (TgtTriple.getArch()) {
Sean Callanan8f993b82010-04-08 00:48:21 +0000267 default:
268 break;
269 case Triple::x86:
270 case Triple::x86_64:
Sean Callananee5dfd42010-02-01 08:49:35 +0000271 stackPointers.insert(registerIDWithName("SP"));
272 stackPointers.insert(registerIDWithName("ESP"));
273 stackPointers.insert(registerIDWithName("RSP"));
274
275 programCounters.insert(registerIDWithName("IP"));
276 programCounters.insert(registerIDWithName("EIP"));
277 programCounters.insert(registerIDWithName("RIP"));
Sean Callanan8f993b82010-04-08 00:48:21 +0000278 break;
279 case Triple::arm:
280 case Triple::thumb:
281 stackPointers.insert(registerIDWithName("SP"));
282
283 programCounters.insert(registerIDWithName("PC"));
284 break;
Sean Callananee5dfd42010-02-01 08:49:35 +0000285 }
286}
287
288const char *EDDisassembler::nameWithRegisterID(unsigned registerID) const {
289 if (registerID >= RegVec.size())
290 return NULL;
291 else
292 return RegVec[registerID].c_str();
293}
294
295unsigned EDDisassembler::registerIDWithName(const char *name) const {
296 regrmap_t::const_iterator iter = RegRMap.find(std::string(name));
297 if (iter == RegRMap.end())
298 return 0;
299 else
300 return (*iter).second;
301}
302
303bool EDDisassembler::registerIsStackPointer(unsigned registerID) {
304 return (stackPointers.find(registerID) != stackPointers.end());
305}
306
307bool EDDisassembler::registerIsProgramCounter(unsigned registerID) {
308 return (programCounters.find(registerID) != programCounters.end());
309}
310
Chris Lattnerd3740872010-04-04 05:04:31 +0000311int EDDisassembler::printInst(std::string &str, MCInst &inst) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000312 PrinterMutex.acquire();
313
Owen Anderson98c5dda2011-09-15 23:38:46 +0000314 InstPrinter->printInst(&inst, *InstStream, "");
Sean Callananee5dfd42010-02-01 08:49:35 +0000315 InstStream->flush();
316 str = *InstString;
317 InstString->clear();
318
319 PrinterMutex.release();
320
321 return 0;
322}
323
Chris Lattnerd8b7aa22011-10-16 04:47:35 +0000324static void diag_handler(const SMDiagnostic &diag, void *context) {
325 if (context)
326 diag.print("", static_cast<EDDisassembler*>(context)->ErrorStream);
Sean Callanan457856c2011-04-09 00:37:25 +0000327}
328
Sean Callananee5dfd42010-02-01 08:49:35 +0000329int EDDisassembler::parseInst(SmallVectorImpl<MCParsedAsmOperand*> &operands,
330 SmallVectorImpl<AsmToken> &tokens,
331 const std::string &str) {
332 int ret = 0;
333
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000334 switch (TgtTriple.getArch()) {
Sean Callanan8f993b82010-04-08 00:48:21 +0000335 default:
336 return -1;
337 case Triple::x86:
338 case Triple::x86_64:
339 case Triple::arm:
340 case Triple::thumb:
341 break;
342 }
343
Sean Callananee5dfd42010-02-01 08:49:35 +0000344 const char *cStr = str.c_str();
345 MemoryBuffer *buf = MemoryBuffer::getMemBuffer(cStr, cStr + strlen(cStr));
346
347 StringRef instName;
348 SMLoc instLoc;
349
350 SourceMgr sourceMgr;
Sean Callanan457856c2011-04-09 00:37:25 +0000351 sourceMgr.setDiagHandler(diag_handler, static_cast<void*>(this));
Sean Callananee5dfd42010-02-01 08:49:35 +0000352 sourceMgr.AddNewSourceBuffer(buf, SMLoc()); // ownership of buf handed over
Evan Cheng203576a2011-07-20 19:50:42 +0000353 MCContext context(*AsmInfo, *MRI, NULL);
Chris Lattnerc18409a2010-03-11 22:53:35 +0000354 OwningPtr<MCStreamer> streamer(createNullStreamer(context));
Jim Grosbach1b84cce2011-08-16 18:33:49 +0000355 OwningPtr<MCAsmParser> genericParser(createMCAsmParser(sourceMgr,
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000356 context, *streamer,
357 *AsmInfo));
Evan Chengebdeeab2011-07-08 01:53:10 +0000358
Sean Callanan3f4f34c2012-02-04 07:45:35 +0000359 OwningPtr<MCSubtargetInfo> STI(Tgt->createMCSubtargetInfo(Key.Triple.c_str(), "", ""));
Evan Cheng94b95502011-07-26 00:24:13 +0000360 OwningPtr<MCTargetAsmParser>
361 TargetParser(Tgt->createMCAsmParser(*STI, *genericParser));
Sean Callananee5dfd42010-02-01 08:49:35 +0000362
Daniel Dunbar9fbb37e2010-07-18 18:31:33 +0000363 AsmToken OpcodeToken = genericParser->Lex();
364 AsmToken NextToken = genericParser->Lex(); // consume next token, because specificParser expects us to
Sean Callanana8702562010-04-24 01:00:16 +0000365
Sean Callanan8f993b82010-04-08 00:48:21 +0000366 if (OpcodeToken.is(AsmToken::Identifier)) {
Sean Callananee5dfd42010-02-01 08:49:35 +0000367 instName = OpcodeToken.getString();
368 instLoc = OpcodeToken.getLoc();
Sean Callanana8702562010-04-24 01:00:16 +0000369
370 if (NextToken.isNot(AsmToken::Eof) &&
371 TargetParser->ParseInstruction(instName, instLoc, operands))
Sean Callananee5dfd42010-02-01 08:49:35 +0000372 ret = -1;
Sean Callanan8f993b82010-04-08 00:48:21 +0000373 } else {
Sean Callananee5dfd42010-02-01 08:49:35 +0000374 ret = -1;
375 }
376
Sean Callananee5dfd42010-02-01 08:49:35 +0000377 ParserMutex.acquire();
378
379 if (!ret) {
380 GenericAsmLexer->setBuffer(buf);
381
382 while (SpecificAsmLexer->Lex(),
383 SpecificAsmLexer->isNot(AsmToken::Eof) &&
384 SpecificAsmLexer->isNot(AsmToken::EndOfStatement)) {
385 if (SpecificAsmLexer->is(AsmToken::Error)) {
386 ret = -1;
387 break;
388 }
389 tokens.push_back(SpecificAsmLexer->getTok());
390 }
391 }
392
393 ParserMutex.release();
394
395 return ret;
396}
397
398int EDDisassembler::llvmSyntaxVariant() const {
399 return LLVMSyntaxVariant;
400}