blob: 017cdb90f620b4b7c77bb26535d369bf55d1b069 [file] [log] [blame]
Chad Rosier3d45a772012-08-17 21:27:25 +00001//===--- SemaStmtAsm.cpp - Semantic Analysis for Asm Statements -----------===//
Chad Rosier4b5e48d2012-08-17 21:19:40 +00002//
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 semantic analysis for inline asm statements.
11//
12//===----------------------------------------------------------------------===//
13
14#include "clang/Sema/SemaInternal.h"
15#include "clang/Sema/Scope.h"
16#include "clang/Sema/ScopeInfo.h"
17#include "clang/Sema/Initialization.h"
18#include "clang/Sema/Lookup.h"
19#include "clang/AST/TypeLoc.h"
20#include "clang/Lex/Preprocessor.h"
21#include "clang/Basic/TargetInfo.h"
22#include "llvm/ADT/ArrayRef.h"
23#include "llvm/ADT/BitVector.h"
24#include "llvm/ADT/SmallString.h"
25#include "llvm/MC/MCAsmInfo.h"
26#include "llvm/MC/MCContext.h"
Chad Rosier6e97be72012-08-22 23:42:09 +000027#include "llvm/MC/MCExpr.h"
Chad Rosier4b5e48d2012-08-17 21:19:40 +000028#include "llvm/MC/MCInst.h"
29#include "llvm/MC/MCInstPrinter.h"
30#include "llvm/MC/MCInstrInfo.h"
31#include "llvm/MC/MCObjectFileInfo.h"
32#include "llvm/MC/MCRegisterInfo.h"
33#include "llvm/MC/MCStreamer.h"
34#include "llvm/MC/MCSubtargetInfo.h"
Chad Rosier6e97be72012-08-22 23:42:09 +000035#include "llvm/MC/MCSymbol.h"
Chad Rosier4b5e48d2012-08-17 21:19:40 +000036#include "llvm/MC/MCTargetAsmParser.h"
37#include "llvm/MC/MCParser/MCAsmLexer.h"
38#include "llvm/MC/MCParser/MCAsmParser.h"
Chad Rosierfb700262012-09-11 23:13:15 +000039#include "llvm/MC/MCParser/MCParsedAsmOperand.h"
Chad Rosier4b5e48d2012-08-17 21:19:40 +000040#include "llvm/Support/SourceMgr.h"
41#include "llvm/Support/TargetRegistry.h"
42#include "llvm/Support/TargetSelect.h"
43using namespace clang;
44using namespace sema;
45
46/// CheckAsmLValue - GNU C has an extremely ugly extension whereby they silently
47/// ignore "noop" casts in places where an lvalue is required by an inline asm.
48/// We emulate this behavior when -fheinous-gnu-extensions is specified, but
49/// provide a strong guidance to not use it.
50///
51/// This method checks to see if the argument is an acceptable l-value and
52/// returns false if it is a case we can handle.
53static bool CheckAsmLValue(const Expr *E, Sema &S) {
54 // Type dependent expressions will be checked during instantiation.
55 if (E->isTypeDependent())
56 return false;
57
58 if (E->isLValue())
59 return false; // Cool, this is an lvalue.
60
61 // Okay, this is not an lvalue, but perhaps it is the result of a cast that we
62 // are supposed to allow.
63 const Expr *E2 = E->IgnoreParenNoopCasts(S.Context);
64 if (E != E2 && E2->isLValue()) {
65 if (!S.getLangOpts().HeinousExtensions)
66 S.Diag(E2->getLocStart(), diag::err_invalid_asm_cast_lvalue)
67 << E->getSourceRange();
68 else
69 S.Diag(E2->getLocStart(), diag::warn_invalid_asm_cast_lvalue)
70 << E->getSourceRange();
71 // Accept, even if we emitted an error diagnostic.
72 return false;
73 }
74
75 // None of the above, just randomly invalid non-lvalue.
76 return true;
77}
78
79/// isOperandMentioned - Return true if the specified operand # is mentioned
80/// anywhere in the decomposed asm string.
81static bool isOperandMentioned(unsigned OpNo,
Chad Rosierdf5faf52012-08-25 00:11:56 +000082 ArrayRef<GCCAsmStmt::AsmStringPiece> AsmStrPieces) {
Chad Rosier4b5e48d2012-08-17 21:19:40 +000083 for (unsigned p = 0, e = AsmStrPieces.size(); p != e; ++p) {
Chad Rosierdf5faf52012-08-25 00:11:56 +000084 const GCCAsmStmt::AsmStringPiece &Piece = AsmStrPieces[p];
Chad Rosier4b5e48d2012-08-17 21:19:40 +000085 if (!Piece.isOperand()) continue;
86
87 // If this is a reference to the input and if the input was the smaller
88 // one, then we have to reject this asm.
89 if (Piece.getOperandNo() == OpNo)
90 return true;
91 }
92 return false;
93}
94
Chad Rosierdf5faf52012-08-25 00:11:56 +000095StmtResult Sema::ActOnGCCAsmStmt(SourceLocation AsmLoc, bool IsSimple,
96 bool IsVolatile, unsigned NumOutputs,
97 unsigned NumInputs, IdentifierInfo **Names,
98 MultiExprArg constraints, MultiExprArg exprs,
99 Expr *asmString, MultiExprArg clobbers,
100 SourceLocation RParenLoc) {
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000101 unsigned NumClobbers = clobbers.size();
102 StringLiteral **Constraints =
Benjamin Kramer5354e772012-08-23 23:38:35 +0000103 reinterpret_cast<StringLiteral**>(constraints.data());
104 Expr **Exprs = exprs.data();
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000105 StringLiteral *AsmString = cast<StringLiteral>(asmString);
Benjamin Kramer5354e772012-08-23 23:38:35 +0000106 StringLiteral **Clobbers = reinterpret_cast<StringLiteral**>(clobbers.data());
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000107
108 SmallVector<TargetInfo::ConstraintInfo, 4> OutputConstraintInfos;
109
110 // The parser verifies that there is a string literal here.
111 if (!AsmString->isAscii())
112 return StmtError(Diag(AsmString->getLocStart(),diag::err_asm_wide_character)
113 << AsmString->getSourceRange());
114
115 for (unsigned i = 0; i != NumOutputs; i++) {
116 StringLiteral *Literal = Constraints[i];
117 if (!Literal->isAscii())
118 return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
119 << Literal->getSourceRange());
120
121 StringRef OutputName;
122 if (Names[i])
123 OutputName = Names[i]->getName();
124
125 TargetInfo::ConstraintInfo Info(Literal->getString(), OutputName);
126 if (!Context.getTargetInfo().validateOutputConstraint(Info))
127 return StmtError(Diag(Literal->getLocStart(),
128 diag::err_asm_invalid_output_constraint)
129 << Info.getConstraintStr());
130
131 // Check that the output exprs are valid lvalues.
132 Expr *OutputExpr = Exprs[i];
133 if (CheckAsmLValue(OutputExpr, *this)) {
134 return StmtError(Diag(OutputExpr->getLocStart(),
135 diag::err_asm_invalid_lvalue_in_output)
136 << OutputExpr->getSourceRange());
137 }
138
139 OutputConstraintInfos.push_back(Info);
140 }
141
142 SmallVector<TargetInfo::ConstraintInfo, 4> InputConstraintInfos;
143
144 for (unsigned i = NumOutputs, e = NumOutputs + NumInputs; i != e; i++) {
145 StringLiteral *Literal = Constraints[i];
146 if (!Literal->isAscii())
147 return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
148 << Literal->getSourceRange());
149
150 StringRef InputName;
151 if (Names[i])
152 InputName = Names[i]->getName();
153
154 TargetInfo::ConstraintInfo Info(Literal->getString(), InputName);
155 if (!Context.getTargetInfo().validateInputConstraint(OutputConstraintInfos.data(),
156 NumOutputs, Info)) {
157 return StmtError(Diag(Literal->getLocStart(),
158 diag::err_asm_invalid_input_constraint)
159 << Info.getConstraintStr());
160 }
161
162 Expr *InputExpr = Exprs[i];
163
164 // Only allow void types for memory constraints.
165 if (Info.allowsMemory() && !Info.allowsRegister()) {
166 if (CheckAsmLValue(InputExpr, *this))
167 return StmtError(Diag(InputExpr->getLocStart(),
168 diag::err_asm_invalid_lvalue_in_input)
169 << Info.getConstraintStr()
170 << InputExpr->getSourceRange());
171 }
172
173 if (Info.allowsRegister()) {
174 if (InputExpr->getType()->isVoidType()) {
175 return StmtError(Diag(InputExpr->getLocStart(),
176 diag::err_asm_invalid_type_in_input)
177 << InputExpr->getType() << Info.getConstraintStr()
178 << InputExpr->getSourceRange());
179 }
180 }
181
182 ExprResult Result = DefaultFunctionArrayLvalueConversion(Exprs[i]);
183 if (Result.isInvalid())
184 return StmtError();
185
186 Exprs[i] = Result.take();
187 InputConstraintInfos.push_back(Info);
188 }
189
190 // Check that the clobbers are valid.
191 for (unsigned i = 0; i != NumClobbers; i++) {
192 StringLiteral *Literal = Clobbers[i];
193 if (!Literal->isAscii())
194 return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
195 << Literal->getSourceRange());
196
197 StringRef Clobber = Literal->getString();
198
199 if (!Context.getTargetInfo().isValidClobber(Clobber))
200 return StmtError(Diag(Literal->getLocStart(),
201 diag::err_asm_unknown_register_name) << Clobber);
202 }
203
Chad Rosierdf5faf52012-08-25 00:11:56 +0000204 GCCAsmStmt *NS =
205 new (Context) GCCAsmStmt(Context, AsmLoc, IsSimple, IsVolatile, NumOutputs,
206 NumInputs, Names, Constraints, Exprs, AsmString,
207 NumClobbers, Clobbers, RParenLoc);
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000208 // Validate the asm string, ensuring it makes sense given the operands we
209 // have.
Chad Rosierdf5faf52012-08-25 00:11:56 +0000210 SmallVector<GCCAsmStmt::AsmStringPiece, 8> Pieces;
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000211 unsigned DiagOffs;
212 if (unsigned DiagID = NS->AnalyzeAsmString(Pieces, Context, DiagOffs)) {
213 Diag(getLocationOfStringLiteralByte(AsmString, DiagOffs), DiagID)
214 << AsmString->getSourceRange();
215 return StmtError();
216 }
217
218 // Validate tied input operands for type mismatches.
219 for (unsigned i = 0, e = InputConstraintInfos.size(); i != e; ++i) {
220 TargetInfo::ConstraintInfo &Info = InputConstraintInfos[i];
221
222 // If this is a tied constraint, verify that the output and input have
223 // either exactly the same type, or that they are int/ptr operands with the
224 // same size (int/long, int*/long, are ok etc).
225 if (!Info.hasTiedOperand()) continue;
226
227 unsigned TiedTo = Info.getTiedOperand();
228 unsigned InputOpNo = i+NumOutputs;
229 Expr *OutputExpr = Exprs[TiedTo];
230 Expr *InputExpr = Exprs[InputOpNo];
231
232 if (OutputExpr->isTypeDependent() || InputExpr->isTypeDependent())
233 continue;
234
235 QualType InTy = InputExpr->getType();
236 QualType OutTy = OutputExpr->getType();
237 if (Context.hasSameType(InTy, OutTy))
238 continue; // All types can be tied to themselves.
239
240 // Decide if the input and output are in the same domain (integer/ptr or
241 // floating point.
242 enum AsmDomain {
243 AD_Int, AD_FP, AD_Other
244 } InputDomain, OutputDomain;
245
246 if (InTy->isIntegerType() || InTy->isPointerType())
247 InputDomain = AD_Int;
248 else if (InTy->isRealFloatingType())
249 InputDomain = AD_FP;
250 else
251 InputDomain = AD_Other;
252
253 if (OutTy->isIntegerType() || OutTy->isPointerType())
254 OutputDomain = AD_Int;
255 else if (OutTy->isRealFloatingType())
256 OutputDomain = AD_FP;
257 else
258 OutputDomain = AD_Other;
259
260 // They are ok if they are the same size and in the same domain. This
261 // allows tying things like:
262 // void* to int*
263 // void* to int if they are the same size.
264 // double to long double if they are the same size.
265 //
266 uint64_t OutSize = Context.getTypeSize(OutTy);
267 uint64_t InSize = Context.getTypeSize(InTy);
268 if (OutSize == InSize && InputDomain == OutputDomain &&
269 InputDomain != AD_Other)
270 continue;
271
272 // If the smaller input/output operand is not mentioned in the asm string,
273 // then we can promote the smaller one to a larger input and the asm string
274 // won't notice.
275 bool SmallerValueMentioned = false;
276
277 // If this is a reference to the input and if the input was the smaller
278 // one, then we have to reject this asm.
279 if (isOperandMentioned(InputOpNo, Pieces)) {
280 // This is a use in the asm string of the smaller operand. Since we
281 // codegen this by promoting to a wider value, the asm will get printed
282 // "wrong".
283 SmallerValueMentioned |= InSize < OutSize;
284 }
285 if (isOperandMentioned(TiedTo, Pieces)) {
286 // If this is a reference to the output, and if the output is the larger
287 // value, then it's ok because we'll promote the input to the larger type.
288 SmallerValueMentioned |= OutSize < InSize;
289 }
290
291 // If the smaller value wasn't mentioned in the asm string, and if the
292 // output was a register, just extend the shorter one to the size of the
293 // larger one.
294 if (!SmallerValueMentioned && InputDomain != AD_Other &&
295 OutputConstraintInfos[TiedTo].allowsRegister())
296 continue;
297
298 // Either both of the operands were mentioned or the smaller one was
299 // mentioned. One more special case that we'll allow: if the tied input is
300 // integer, unmentioned, and is a constant, then we'll allow truncating it
301 // down to the size of the destination.
302 if (InputDomain == AD_Int && OutputDomain == AD_Int &&
303 !isOperandMentioned(InputOpNo, Pieces) &&
304 InputExpr->isEvaluatable(Context)) {
305 CastKind castKind =
306 (OutTy->isBooleanType() ? CK_IntegralToBoolean : CK_IntegralCast);
307 InputExpr = ImpCastExprToType(InputExpr, OutTy, castKind).take();
308 Exprs[InputOpNo] = InputExpr;
309 NS->setInputExpr(i, InputExpr);
310 continue;
311 }
312
313 Diag(InputExpr->getLocStart(),
314 diag::err_asm_tying_incompatible_types)
315 << InTy << OutTy << OutputExpr->getSourceRange()
316 << InputExpr->getSourceRange();
317 return StmtError();
318 }
319
320 return Owned(NS);
321}
322
Chad Rosier358ab762012-08-22 21:08:06 +0000323// getSpelling - Get the spelling of the AsmTok token.
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000324static StringRef getSpelling(Sema &SemaRef, Token AsmTok) {
325 StringRef Asm;
326 SmallString<512> TokenBuf;
327 TokenBuf.resize(512);
328 bool StringInvalid = false;
329 Asm = SemaRef.PP.getSpelling(AsmTok, TokenBuf, &StringInvalid);
330 assert (!StringInvalid && "Expected valid string!");
331 return Asm;
332}
333
Bob Wilson40d39e32012-09-24 19:57:55 +0000334// Break the AsmString into pieces (i.e., mnemonic and operands).
Chad Rosier38c71d32012-08-21 21:56:39 +0000335static void buildMSAsmPieces(StringRef Asm, std::vector<StringRef> &Pieces) {
336 std::pair<StringRef,StringRef> Split = Asm.split(' ');
337
338 // Mnemonic
339 Pieces.push_back(Split.first);
340 Asm = Split.second;
341
342 // Operands
343 while (!Asm.empty()) {
344 Split = Asm.split(", ");
345 Pieces.push_back(Split.first);
346 Asm = Split.second;
347 }
348}
349
Chad Rosierf0fbd772012-08-22 21:04:07 +0000350static void buildMSAsmPieces(std::vector<std::string> &AsmStrings,
351 std::vector<std::vector<StringRef> > &Pieces) {
352 for (unsigned i = 0, e = AsmStrings.size(); i != e; ++i)
353 buildMSAsmPieces(AsmStrings[i], Pieces[i]);
354}
355
Chad Rosierb55e6022012-09-13 00:06:55 +0000356// Build the individual assembly instruction(s) and place them in the AsmStrings
Bob Wilsonb0f6b9c2012-09-24 19:57:59 +0000357// vector. These strings are fed to the AsmParser. Returns true on error.
358static bool buildMSAsmStrings(Sema &SemaRef,
359 SourceLocation AsmLoc,
360 ArrayRef<Token> AsmToks,
Chad Rosier6ec69c32012-10-12 00:02:48 +0000361 std::vector<std::string> &AsmStrings) {
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000362 assert (!AsmToks.empty() && "Didn't expect an empty AsmToks!");
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000363
364 SmallString<512> Asm;
365 for (unsigned i = 0, e = AsmToks.size(); i < e; ++i) {
Bob Wilson40d39e32012-09-24 19:57:55 +0000366 bool isNewAsm = ((i == 0) ||
367 AsmToks[i].isAtStartOfLine() ||
368 AsmToks[i].is(tok::kw_asm));
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000369
370 if (isNewAsm) {
Chad Rosier38c71d32012-08-21 21:56:39 +0000371 if (i) {
Benjamin Kramer32f3acc2012-08-24 20:43:21 +0000372 AsmStrings.push_back(Asm.str());
Chad Rosier38c71d32012-08-21 21:56:39 +0000373 Asm.clear();
Chad Rosier38c71d32012-08-21 21:56:39 +0000374 }
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000375 if (AsmToks[i].is(tok::kw_asm)) {
376 i++; // Skip __asm
Bob Wilsonb0f6b9c2012-09-24 19:57:59 +0000377 if (i == e) {
378 SemaRef.Diag(AsmLoc, diag::err_asm_empty);
379 return true;
380 }
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000381 }
382 }
383
Chad Rosierb55e6022012-09-13 00:06:55 +0000384 if (i && AsmToks[i].hasLeadingSpace() && !isNewAsm)
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000385 Asm += ' ';
Chad Rosier4de97162012-09-11 00:51:28 +0000386
387 StringRef Spelling = getSpelling(SemaRef, AsmToks[i]);
388 Asm += Spelling;
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000389 }
Benjamin Kramer32f3acc2012-08-24 20:43:21 +0000390 AsmStrings.push_back(Asm.str());
Bob Wilsonb0f6b9c2012-09-24 19:57:59 +0000391
392 return false;
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000393}
394
Chad Rosierc6916492012-09-06 19:56:25 +0000395#define DEF_SIMPLE_MSASM(STR) \
Chad Rosier89fb6d72012-08-28 20:28:20 +0000396 MSAsmStmt *NS = \
397 new (Context) MSAsmStmt(Context, AsmLoc, LBraceLoc, /*IsSimple*/ true, \
398 /*IsVolatile*/ true, AsmToks, Inputs, Outputs, \
Chad Rosierc6916492012-09-06 19:56:25 +0000399 InputExprs, OutputExprs, STR, Constraints, \
Chad Rosier89fb6d72012-08-28 20:28:20 +0000400 Clobbers, EndLoc);
Chad Rosier2735df22012-08-22 19:18:30 +0000401
Chad Rosierb55e6022012-09-13 00:06:55 +0000402StmtResult Sema::ActOnMSAsmStmt(SourceLocation AsmLoc, SourceLocation LBraceLoc,
403 ArrayRef<Token> AsmToks,SourceLocation EndLoc) {
Chad Rosier89fb6d72012-08-28 20:28:20 +0000404 SmallVector<StringRef, 4> Constraints;
405 std::vector<std::string> InputConstraints;
406 std::vector<std::string> OutputConstraints;
Chad Rosier4112a4c2012-08-28 20:35:06 +0000407 SmallVector<StringRef, 4> Clobbers;
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000408 std::set<std::string> ClobberRegs;
Chad Rosierb55e6022012-09-13 00:06:55 +0000409
410 // FIXME: Use a struct to hold the various expression information.
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000411 SmallVector<IdentifierInfo*, 4> Inputs;
412 SmallVector<IdentifierInfo*, 4> Outputs;
Chad Rosier633abb02012-08-24 00:07:09 +0000413 SmallVector<Expr*, 4> InputExprs;
414 SmallVector<Expr*, 4> OutputExprs;
Chad Rosieracc22b62012-09-06 19:35:00 +0000415 SmallVector<std::string, 4> InputExprNames;
416 SmallVector<std::string, 4> OutputExprNames;
Chad Rosierb55e6022012-09-13 00:06:55 +0000417 SmallVector<unsigned, 4> InputExprStrIdx;
418 SmallVector<unsigned, 4> OutputExprStrIdx;
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000419
420 // Empty asm statements don't need to instantiate the AsmParser, etc.
Chad Rosierc6916492012-09-06 19:56:25 +0000421 StringRef EmptyAsmStr;
422 if (AsmToks.empty()) { DEF_SIMPLE_MSASM(EmptyAsmStr); return Owned(NS); }
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000423
Chad Rosier38c71d32012-08-21 21:56:39 +0000424 std::vector<std::string> AsmStrings;
Chad Rosier6ec69c32012-10-12 00:02:48 +0000425 if (buildMSAsmStrings(*this, AsmLoc, AsmToks, AsmStrings))
Bob Wilsonb0f6b9c2012-09-24 19:57:59 +0000426 return StmtError();
Chad Rosier38c71d32012-08-21 21:56:39 +0000427
Chad Rosiere78460f2012-08-22 20:57:07 +0000428 std::vector<std::vector<StringRef> > Pieces(AsmStrings.size());
Chad Rosierf0fbd772012-08-22 21:04:07 +0000429 buildMSAsmPieces(AsmStrings, Pieces);
Chad Rosier153f8ec2012-08-22 19:50:28 +0000430
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000431 // Get the target specific parser.
432 std::string Error;
433 const std::string &TT = Context.getTargetInfo().getTriple().getTriple();
434 const llvm::Target *TheTarget(llvm::TargetRegistry::lookupTarget(TT, Error));
435
436 OwningPtr<llvm::MCAsmInfo> MAI(TheTarget->createMCAsmInfo(TT));
437 OwningPtr<llvm::MCRegisterInfo> MRI(TheTarget->createMCRegInfo(TT));
438 OwningPtr<llvm::MCObjectFileInfo> MOFI(new llvm::MCObjectFileInfo());
439 OwningPtr<llvm::MCSubtargetInfo>
440 STI(TheTarget->createMCSubtargetInfo(TT, "", ""));
441
Chad Rosier25bd2982012-08-23 15:44:35 +0000442 for (unsigned StrIdx = 0, e = AsmStrings.size(); StrIdx != e; ++StrIdx) {
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000443 llvm::SourceMgr SrcMgr;
444 llvm::MCContext Ctx(*MAI, *MRI, MOFI.get(), &SrcMgr);
445 llvm::MemoryBuffer *Buffer =
Chad Rosier25bd2982012-08-23 15:44:35 +0000446 llvm::MemoryBuffer::getMemBuffer(AsmStrings[StrIdx], "<inline asm>");
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000447
448 // Tell SrcMgr about this buffer, which is what the parser will pick up.
449 SrcMgr.AddNewSourceBuffer(Buffer, llvm::SMLoc());
450
451 OwningPtr<llvm::MCStreamer> Str(createNullStreamer(Ctx));
452 OwningPtr<llvm::MCAsmParser>
453 Parser(createMCAsmParser(SrcMgr, Ctx, *Str.get(), *MAI));
454 OwningPtr<llvm::MCTargetAsmParser>
455 TargetParser(TheTarget->createMCAsmParser(*STI, *Parser));
456 // Change to the Intel dialect.
457 Parser->setAssemblerDialect(1);
458 Parser->setTargetParser(*TargetParser.get());
Chad Rosier67bd78f2012-10-13 00:26:22 +0000459 Parser->setParsingInlineAsm(true);
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000460
461 // Prime the lexer.
462 Parser->Lex();
463
464 // Parse the opcode.
465 StringRef IDVal;
466 Parser->ParseIdentifier(IDVal);
467
468 // Canonicalize the opcode to lower case.
Chad Rosierbe5c3fb2012-09-03 02:30:13 +0000469 SmallString<128> OpcodeStr;
Chad Rosierb706d902012-08-28 22:08:58 +0000470 for (unsigned i = 0, e = IDVal.size(); i != e; ++i)
Chad Rosierbe5c3fb2012-09-03 02:30:13 +0000471 OpcodeStr.push_back(tolower(IDVal[i]));
Chad Rosier3d442192012-09-21 22:22:39 +0000472 // FIXME: Convert to a StmtError.
473 assert(TargetParser->mnemonicIsValid(OpcodeStr) && "Invalid mnemonic!");
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000474
475 // Parse the operands.
476 llvm::SMLoc IDLoc;
477 SmallVector<llvm::MCParsedAsmOperand*, 8> Operands;
Chad Rosierbe5c3fb2012-09-03 02:30:13 +0000478 bool HadError = TargetParser->ParseInstruction(OpcodeStr.str(), IDLoc,
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000479 Operands);
Chad Rosier2735df22012-08-22 19:18:30 +0000480 // If we had an error parsing the operands, fail gracefully.
Chad Rosierc6916492012-09-06 19:56:25 +0000481 if (HadError) { DEF_SIMPLE_MSASM(EmptyAsmStr); return Owned(NS); }
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000482
483 // Match the MCInstr.
Chad Rosier3034d882012-10-01 23:45:59 +0000484 unsigned Opcode;
Chad Rosier83591b62012-08-21 18:15:08 +0000485 unsigned ErrorInfo;
Chad Rosier67bd78f2012-10-13 00:26:22 +0000486 HadError = TargetParser->MatchAndEmitInstruction(IDLoc, Opcode, Operands,
487 *Str.get(), ErrorInfo,
488 /*MatchingInlineAsm*/ true);
Chad Rosier2735df22012-08-22 19:18:30 +0000489 // If we had an error parsing the operands, fail gracefully.
Chad Rosierc6916492012-09-06 19:56:25 +0000490 if (HadError) { DEF_SIMPLE_MSASM(EmptyAsmStr); return Owned(NS); }
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000491
492 // Get the instruction descriptor.
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000493 const llvm::MCInstrInfo *MII = TheTarget->createMCInstrInfo();
Chad Rosier3034d882012-10-01 23:45:59 +0000494 const llvm::MCInstrDesc &Desc = MII->get(Opcode);
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000495 llvm::MCInstPrinter *IP =
496 TheTarget->createMCInstPrinter(1, *MAI, *MII, *MRI, *STI);
497
Chad Rosier6e97be72012-08-22 23:42:09 +0000498 // Build the list of clobbers, outputs and inputs.
Chad Rosierfd5e56e2012-08-22 22:10:51 +0000499 unsigned NumDefs = Desc.getNumDefs();
Chad Rosier1b497f22012-09-03 20:40:52 +0000500 for (unsigned i = 1, e = Operands.size(); i != e; ++i) {
Chad Rosier3034d882012-10-01 23:45:59 +0000501 // Skip immediates.
502 if (Operands[i]->isImm())
Chad Rosierfb700262012-09-11 23:13:15 +0000503 continue;
504
Chad Rosier3034d882012-10-01 23:45:59 +0000505 // Register.
506 if (Operands[i]->isReg()) {
507 // Clobber.
Chad Rosier106ddc12012-10-12 22:53:52 +0000508 if (NumDefs && (Operands[i]->getMCOperandNum() < NumDefs)) {
Chad Rosier7f9678b2012-09-12 18:14:25 +0000509 std::string Reg;
510 llvm::raw_string_ostream OS(Reg);
Chad Rosier3034d882012-10-01 23:45:59 +0000511 IP->printRegName(OS, Operands[i]->getReg());
Chad Rosier7f9678b2012-09-12 18:14:25 +0000512 StringRef Clobber(OS.str());
513 if (!Context.getTargetInfo().isValidClobber(Clobber))
514 return StmtError(
515 Diag(AsmLoc, diag::err_asm_unknown_register_name) << Clobber);
516 ClobberRegs.insert(Reg);
Chad Rosier1b497f22012-09-03 20:40:52 +0000517 }
Chad Rosier3034d882012-10-01 23:45:59 +0000518 continue;
519 }
520
521 // Expr/Input or Output.
Chad Rosier90a76e52012-10-11 21:44:41 +0000522 StringRef Name = Pieces[StrIdx][i];
Chad Rosier34404152012-10-11 21:28:29 +0000523 if (IdentifierInfo *II = &Context.Idents.get(Name)) {
Chad Rosier3034d882012-10-01 23:45:59 +0000524 CXXScopeSpec SS;
525 UnqualifiedId Id;
526 SourceLocation Loc;
527 Id.setIdentifier(II, AsmLoc);
528 ExprResult Result = ActOnIdExpression(getCurScope(), SS, Loc, Id,
529 false, false);
530 if (!Result.isInvalid()) {
Chad Rosier3034d882012-10-01 23:45:59 +0000531 bool isMemDef = (i == 1) && Desc.mayStore();
532 if (isMemDef) {
533 Outputs.push_back(II);
534 OutputExprs.push_back(Result.take());
535 OutputExprNames.push_back(Name.str());
536 OutputExprStrIdx.push_back(StrIdx);
537
Chad Rosier106ddc12012-10-12 22:53:52 +0000538 std::string Constraint = "=" + Operands[i]->getConstraint().str();
Chad Rosier3034d882012-10-01 23:45:59 +0000539 OutputConstraints.push_back(Constraint);
540 } else {
541 Inputs.push_back(II);
542 InputExprs.push_back(Result.take());
543 InputExprNames.push_back(Name.str());
544 InputExprStrIdx.push_back(StrIdx);
Chad Rosier106ddc12012-10-12 22:53:52 +0000545 InputConstraints.push_back(Operands[i]->getConstraint());
Chad Rosier633abb02012-08-24 00:07:09 +0000546 }
Chad Rosier6e97be72012-08-22 23:42:09 +0000547 }
Chad Rosierfd5e56e2012-08-22 22:10:51 +0000548 }
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000549 }
550 }
551 for (std::set<std::string>::iterator I = ClobberRegs.begin(),
552 E = ClobberRegs.end(); I != E; ++I)
553 Clobbers.push_back(*I);
554
Chad Rosier89fb6d72012-08-28 20:28:20 +0000555 // Merge the output and input constraints. Output constraints are expected
556 // first.
557 for (std::vector<std::string>::iterator I = OutputConstraints.begin(),
558 E = OutputConstraints.end(); I != E; ++I)
559 Constraints.push_back(*I);
560
561 for (std::vector<std::string>::iterator I = InputConstraints.begin(),
562 E = InputConstraints.end(); I != E; ++I)
563 Constraints.push_back(*I);
564
Chad Rosieracc22b62012-09-06 19:35:00 +0000565 // Enumerate the AsmString expressions.
Chad Rosieracc22b62012-09-06 19:35:00 +0000566 unsigned OpNum = 0;
567 for (unsigned i = 0, e = OutputExprNames.size(); i != e; ++i, ++OpNum) {
Chad Rosierb55e6022012-09-13 00:06:55 +0000568 unsigned StrIdx = OutputExprStrIdx[i];
569 // Iterate over the assembly instruction pieces, skipping the mnemonic.
570 for (unsigned j = 1, f = Pieces[StrIdx].size(); j != f; ++j) {
571 // If the operand and the expression name match, then rewrite the operand.
572 if (OutputExprNames[i] == Pieces[StrIdx][j]) {
573 SmallString<32> Res;
574 llvm::raw_svector_ostream OS(Res);
575 OS << '$' << OpNum;
576 OutputExprNames[i] = OS.str();
577 Pieces[StrIdx][j] = OutputExprNames[i];
578 break;
579 }
Chad Rosierb55e6022012-09-13 00:06:55 +0000580 }
Chad Rosieracc22b62012-09-06 19:35:00 +0000581 }
582 for (unsigned i = 0, e = InputExprNames.size(); i != e; ++i, ++OpNum) {
Chad Rosierb55e6022012-09-13 00:06:55 +0000583 unsigned StrIdx = InputExprStrIdx[i];
584 // Iterate over the assembly instruction pieces, skipping the mnemonic.
585 for (unsigned j = 1, f = Pieces[StrIdx].size(); j != f; ++j) {
586 // If the operand and the expression name match, then rewrite the operand.
587 if (InputExprNames[i] == Pieces[StrIdx][j]) {
588 SmallString<32> Res;
589 llvm::raw_svector_ostream OS(Res);
590 OS << '$' << OpNum;
591 InputExprNames[i] = OS.str();
592 Pieces[StrIdx][j] = InputExprNames[i];
593 break;
594 }
Chad Rosierb55e6022012-09-13 00:06:55 +0000595 }
596 }
597
598 // Emit the IR assembly string.
599 std::string AsmString;
600 for (unsigned i = 0, e = Pieces.size(); i != e; ++i) {
601 // Skip empty asm stmts.
602 if (Pieces[i].empty()) continue;
603
604 if (i > 0)
605 AsmString += "\n\t";
606
607 // Emit the mnemonic.
608 AsmString += Pieces[i][0];
609 if (Pieces[i].size() > 1)
610 AsmString += ' ';
611
612 // Emit the operands adding $$ to constants.
613 for (unsigned j = 1, f = Pieces[i].size(); j != f; ++j) {
614 if (j > 1) AsmString += ", ";
615 unsigned Val;
616 if (!Pieces[i][j].getAsInteger(0, Val))
617 AsmString += "$$";
618
619 AsmString += Pieces[i][j];
620 }
Chad Rosieracc22b62012-09-06 19:35:00 +0000621 }
622
Chad Rosierde31b282012-10-12 23:32:10 +0000623 bool IsSimple = Inputs.size() != 0 || Outputs.size() != 0;
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000624 MSAsmStmt *NS =
625 new (Context) MSAsmStmt(Context, AsmLoc, LBraceLoc, IsSimple,
626 /*IsVolatile*/ true, AsmToks, Inputs, Outputs,
Chad Rosier89fb6d72012-08-28 20:28:20 +0000627 InputExprs, OutputExprs, AsmString, Constraints,
628 Clobbers, EndLoc);
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000629 return Owned(NS);
630}