blob: 5cab02d92489c9e6b3fe9a8f80890e42b2a3b9ee [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"
39#include "llvm/Support/SourceMgr.h"
40#include "llvm/Support/TargetRegistry.h"
41#include "llvm/Support/TargetSelect.h"
42using namespace clang;
43using namespace sema;
44
45/// CheckAsmLValue - GNU C has an extremely ugly extension whereby they silently
46/// ignore "noop" casts in places where an lvalue is required by an inline asm.
47/// We emulate this behavior when -fheinous-gnu-extensions is specified, but
48/// provide a strong guidance to not use it.
49///
50/// This method checks to see if the argument is an acceptable l-value and
51/// returns false if it is a case we can handle.
52static bool CheckAsmLValue(const Expr *E, Sema &S) {
53 // Type dependent expressions will be checked during instantiation.
54 if (E->isTypeDependent())
55 return false;
56
57 if (E->isLValue())
58 return false; // Cool, this is an lvalue.
59
60 // Okay, this is not an lvalue, but perhaps it is the result of a cast that we
61 // are supposed to allow.
62 const Expr *E2 = E->IgnoreParenNoopCasts(S.Context);
63 if (E != E2 && E2->isLValue()) {
64 if (!S.getLangOpts().HeinousExtensions)
65 S.Diag(E2->getLocStart(), diag::err_invalid_asm_cast_lvalue)
66 << E->getSourceRange();
67 else
68 S.Diag(E2->getLocStart(), diag::warn_invalid_asm_cast_lvalue)
69 << E->getSourceRange();
70 // Accept, even if we emitted an error diagnostic.
71 return false;
72 }
73
74 // None of the above, just randomly invalid non-lvalue.
75 return true;
76}
77
78/// isOperandMentioned - Return true if the specified operand # is mentioned
79/// anywhere in the decomposed asm string.
80static bool isOperandMentioned(unsigned OpNo,
Chad Rosierdf5faf52012-08-25 00:11:56 +000081 ArrayRef<GCCAsmStmt::AsmStringPiece> AsmStrPieces) {
Chad Rosier4b5e48d2012-08-17 21:19:40 +000082 for (unsigned p = 0, e = AsmStrPieces.size(); p != e; ++p) {
Chad Rosierdf5faf52012-08-25 00:11:56 +000083 const GCCAsmStmt::AsmStringPiece &Piece = AsmStrPieces[p];
Chad Rosier4b5e48d2012-08-17 21:19:40 +000084 if (!Piece.isOperand()) continue;
85
86 // If this is a reference to the input and if the input was the smaller
87 // one, then we have to reject this asm.
88 if (Piece.getOperandNo() == OpNo)
89 return true;
90 }
91 return false;
92}
93
Chad Rosierdf5faf52012-08-25 00:11:56 +000094StmtResult Sema::ActOnGCCAsmStmt(SourceLocation AsmLoc, bool IsSimple,
95 bool IsVolatile, unsigned NumOutputs,
96 unsigned NumInputs, IdentifierInfo **Names,
97 MultiExprArg constraints, MultiExprArg exprs,
98 Expr *asmString, MultiExprArg clobbers,
99 SourceLocation RParenLoc) {
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000100 unsigned NumClobbers = clobbers.size();
101 StringLiteral **Constraints =
Benjamin Kramer5354e772012-08-23 23:38:35 +0000102 reinterpret_cast<StringLiteral**>(constraints.data());
103 Expr **Exprs = exprs.data();
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000104 StringLiteral *AsmString = cast<StringLiteral>(asmString);
Benjamin Kramer5354e772012-08-23 23:38:35 +0000105 StringLiteral **Clobbers = reinterpret_cast<StringLiteral**>(clobbers.data());
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000106
107 SmallVector<TargetInfo::ConstraintInfo, 4> OutputConstraintInfos;
108
109 // The parser verifies that there is a string literal here.
110 if (!AsmString->isAscii())
111 return StmtError(Diag(AsmString->getLocStart(),diag::err_asm_wide_character)
112 << AsmString->getSourceRange());
113
114 for (unsigned i = 0; i != NumOutputs; i++) {
115 StringLiteral *Literal = Constraints[i];
116 if (!Literal->isAscii())
117 return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
118 << Literal->getSourceRange());
119
120 StringRef OutputName;
121 if (Names[i])
122 OutputName = Names[i]->getName();
123
124 TargetInfo::ConstraintInfo Info(Literal->getString(), OutputName);
125 if (!Context.getTargetInfo().validateOutputConstraint(Info))
126 return StmtError(Diag(Literal->getLocStart(),
127 diag::err_asm_invalid_output_constraint)
128 << Info.getConstraintStr());
129
130 // Check that the output exprs are valid lvalues.
131 Expr *OutputExpr = Exprs[i];
132 if (CheckAsmLValue(OutputExpr, *this)) {
133 return StmtError(Diag(OutputExpr->getLocStart(),
134 diag::err_asm_invalid_lvalue_in_output)
135 << OutputExpr->getSourceRange());
136 }
137
138 OutputConstraintInfos.push_back(Info);
139 }
140
141 SmallVector<TargetInfo::ConstraintInfo, 4> InputConstraintInfos;
142
143 for (unsigned i = NumOutputs, e = NumOutputs + NumInputs; i != e; i++) {
144 StringLiteral *Literal = Constraints[i];
145 if (!Literal->isAscii())
146 return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
147 << Literal->getSourceRange());
148
149 StringRef InputName;
150 if (Names[i])
151 InputName = Names[i]->getName();
152
153 TargetInfo::ConstraintInfo Info(Literal->getString(), InputName);
154 if (!Context.getTargetInfo().validateInputConstraint(OutputConstraintInfos.data(),
155 NumOutputs, Info)) {
156 return StmtError(Diag(Literal->getLocStart(),
157 diag::err_asm_invalid_input_constraint)
158 << Info.getConstraintStr());
159 }
160
161 Expr *InputExpr = Exprs[i];
162
163 // Only allow void types for memory constraints.
164 if (Info.allowsMemory() && !Info.allowsRegister()) {
165 if (CheckAsmLValue(InputExpr, *this))
166 return StmtError(Diag(InputExpr->getLocStart(),
167 diag::err_asm_invalid_lvalue_in_input)
168 << Info.getConstraintStr()
169 << InputExpr->getSourceRange());
170 }
171
172 if (Info.allowsRegister()) {
173 if (InputExpr->getType()->isVoidType()) {
174 return StmtError(Diag(InputExpr->getLocStart(),
175 diag::err_asm_invalid_type_in_input)
176 << InputExpr->getType() << Info.getConstraintStr()
177 << InputExpr->getSourceRange());
178 }
179 }
180
181 ExprResult Result = DefaultFunctionArrayLvalueConversion(Exprs[i]);
182 if (Result.isInvalid())
183 return StmtError();
184
185 Exprs[i] = Result.take();
186 InputConstraintInfos.push_back(Info);
187 }
188
189 // Check that the clobbers are valid.
190 for (unsigned i = 0; i != NumClobbers; i++) {
191 StringLiteral *Literal = Clobbers[i];
192 if (!Literal->isAscii())
193 return StmtError(Diag(Literal->getLocStart(),diag::err_asm_wide_character)
194 << Literal->getSourceRange());
195
196 StringRef Clobber = Literal->getString();
197
198 if (!Context.getTargetInfo().isValidClobber(Clobber))
199 return StmtError(Diag(Literal->getLocStart(),
200 diag::err_asm_unknown_register_name) << Clobber);
201 }
202
Chad Rosierdf5faf52012-08-25 00:11:56 +0000203 GCCAsmStmt *NS =
204 new (Context) GCCAsmStmt(Context, AsmLoc, IsSimple, IsVolatile, NumOutputs,
205 NumInputs, Names, Constraints, Exprs, AsmString,
206 NumClobbers, Clobbers, RParenLoc);
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000207 // Validate the asm string, ensuring it makes sense given the operands we
208 // have.
Chad Rosierdf5faf52012-08-25 00:11:56 +0000209 SmallVector<GCCAsmStmt::AsmStringPiece, 8> Pieces;
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000210 unsigned DiagOffs;
211 if (unsigned DiagID = NS->AnalyzeAsmString(Pieces, Context, DiagOffs)) {
212 Diag(getLocationOfStringLiteralByte(AsmString, DiagOffs), DiagID)
213 << AsmString->getSourceRange();
214 return StmtError();
215 }
216
217 // Validate tied input operands for type mismatches.
218 for (unsigned i = 0, e = InputConstraintInfos.size(); i != e; ++i) {
219 TargetInfo::ConstraintInfo &Info = InputConstraintInfos[i];
220
221 // If this is a tied constraint, verify that the output and input have
222 // either exactly the same type, or that they are int/ptr operands with the
223 // same size (int/long, int*/long, are ok etc).
224 if (!Info.hasTiedOperand()) continue;
225
226 unsigned TiedTo = Info.getTiedOperand();
227 unsigned InputOpNo = i+NumOutputs;
228 Expr *OutputExpr = Exprs[TiedTo];
229 Expr *InputExpr = Exprs[InputOpNo];
230
231 if (OutputExpr->isTypeDependent() || InputExpr->isTypeDependent())
232 continue;
233
234 QualType InTy = InputExpr->getType();
235 QualType OutTy = OutputExpr->getType();
236 if (Context.hasSameType(InTy, OutTy))
237 continue; // All types can be tied to themselves.
238
239 // Decide if the input and output are in the same domain (integer/ptr or
240 // floating point.
241 enum AsmDomain {
242 AD_Int, AD_FP, AD_Other
243 } InputDomain, OutputDomain;
244
245 if (InTy->isIntegerType() || InTy->isPointerType())
246 InputDomain = AD_Int;
247 else if (InTy->isRealFloatingType())
248 InputDomain = AD_FP;
249 else
250 InputDomain = AD_Other;
251
252 if (OutTy->isIntegerType() || OutTy->isPointerType())
253 OutputDomain = AD_Int;
254 else if (OutTy->isRealFloatingType())
255 OutputDomain = AD_FP;
256 else
257 OutputDomain = AD_Other;
258
259 // They are ok if they are the same size and in the same domain. This
260 // allows tying things like:
261 // void* to int*
262 // void* to int if they are the same size.
263 // double to long double if they are the same size.
264 //
265 uint64_t OutSize = Context.getTypeSize(OutTy);
266 uint64_t InSize = Context.getTypeSize(InTy);
267 if (OutSize == InSize && InputDomain == OutputDomain &&
268 InputDomain != AD_Other)
269 continue;
270
271 // If the smaller input/output operand is not mentioned in the asm string,
272 // then we can promote the smaller one to a larger input and the asm string
273 // won't notice.
274 bool SmallerValueMentioned = false;
275
276 // If this is a reference to the input and if the input was the smaller
277 // one, then we have to reject this asm.
278 if (isOperandMentioned(InputOpNo, Pieces)) {
279 // This is a use in the asm string of the smaller operand. Since we
280 // codegen this by promoting to a wider value, the asm will get printed
281 // "wrong".
282 SmallerValueMentioned |= InSize < OutSize;
283 }
284 if (isOperandMentioned(TiedTo, Pieces)) {
285 // If this is a reference to the output, and if the output is the larger
286 // value, then it's ok because we'll promote the input to the larger type.
287 SmallerValueMentioned |= OutSize < InSize;
288 }
289
290 // If the smaller value wasn't mentioned in the asm string, and if the
291 // output was a register, just extend the shorter one to the size of the
292 // larger one.
293 if (!SmallerValueMentioned && InputDomain != AD_Other &&
294 OutputConstraintInfos[TiedTo].allowsRegister())
295 continue;
296
297 // Either both of the operands were mentioned or the smaller one was
298 // mentioned. One more special case that we'll allow: if the tied input is
299 // integer, unmentioned, and is a constant, then we'll allow truncating it
300 // down to the size of the destination.
301 if (InputDomain == AD_Int && OutputDomain == AD_Int &&
302 !isOperandMentioned(InputOpNo, Pieces) &&
303 InputExpr->isEvaluatable(Context)) {
304 CastKind castKind =
305 (OutTy->isBooleanType() ? CK_IntegralToBoolean : CK_IntegralCast);
306 InputExpr = ImpCastExprToType(InputExpr, OutTy, castKind).take();
307 Exprs[InputOpNo] = InputExpr;
308 NS->setInputExpr(i, InputExpr);
309 continue;
310 }
311
312 Diag(InputExpr->getLocStart(),
313 diag::err_asm_tying_incompatible_types)
314 << InTy << OutTy << OutputExpr->getSourceRange()
315 << InputExpr->getSourceRange();
316 return StmtError();
317 }
318
319 return Owned(NS);
320}
321
322// isMSAsmKeyword - Return true if this is an MS-style inline asm keyword. These
323// require special handling.
324static bool isMSAsmKeyword(StringRef Name) {
325 bool Ret = llvm::StringSwitch<bool>(Name)
326 .Cases("EVEN", "ALIGN", true) // Alignment directives.
327 .Cases("LENGTH", "SIZE", "TYPE", true) // Type and variable sizes.
328 .Case("_emit", true) // _emit Pseudoinstruction.
329 .Default(false);
330 return Ret;
331}
332
Chad Rosier6e97be72012-08-22 23:42:09 +0000333// getIdentifierInfo - Given a Name and a range of tokens, find the associated
334// IdentifierInfo*.
335static IdentifierInfo *getIdentifierInfo(StringRef Name,
336 ArrayRef<Token> AsmToks,
337 unsigned Begin, unsigned End) {
338 for (unsigned i = Begin; i <= End; ++i) {
339 IdentifierInfo *II = AsmToks[i].getIdentifierInfo();
340 if (II && II->getName() == Name)
341 return II;
342 }
343 return 0;
344}
345
Chad Rosier358ab762012-08-22 21:08:06 +0000346// getSpelling - Get the spelling of the AsmTok token.
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000347static StringRef getSpelling(Sema &SemaRef, Token AsmTok) {
348 StringRef Asm;
349 SmallString<512> TokenBuf;
350 TokenBuf.resize(512);
351 bool StringInvalid = false;
352 Asm = SemaRef.PP.getSpelling(AsmTok, TokenBuf, &StringInvalid);
353 assert (!StringInvalid && "Expected valid string!");
354 return Asm;
355}
356
Chad Rosier358ab762012-08-22 21:08:06 +0000357// Determine if we should bail on this MSAsm instruction.
Chad Rosier2735df22012-08-22 19:18:30 +0000358static bool bailOnMSAsm(std::vector<StringRef> Piece) {
359 for (unsigned i = 0, e = Piece.size(); i != e; ++i)
360 if (isMSAsmKeyword(Piece[i]))
361 return true;
362 return false;
363}
364
Chad Rosier358ab762012-08-22 21:08:06 +0000365// Determine if we should bail on this MSAsm block.
Chad Rosier2735df22012-08-22 19:18:30 +0000366static bool bailOnMSAsm(std::vector<std::vector<StringRef> > Pieces) {
367 for (unsigned i = 0, e = Pieces.size(); i != e; ++i)
368 if (bailOnMSAsm(Pieces[i]))
369 return true;
370 return false;
371}
372
Chad Rosier358ab762012-08-22 21:08:06 +0000373// Determine if this is a simple MSAsm instruction.
Chad Rosier98ac6082012-08-21 23:09:21 +0000374static bool isSimpleMSAsm(std::vector<StringRef> &Pieces,
375 const TargetInfo &TI) {
376 if (isMSAsmKeyword(Pieces[0]))
377 return false;
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000378
Chad Rosier98ac6082012-08-21 23:09:21 +0000379 for (unsigned i = 1, e = Pieces.size(); i != e; ++i)
380 if (!TI.isValidGCCRegisterName(Pieces[i]))
381 return false;
Chad Rosier153f8ec2012-08-22 19:50:28 +0000382 return true;
383}
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000384
Chad Rosier358ab762012-08-22 21:08:06 +0000385// Determine if this is a simple MSAsm block.
Chad Rosier153f8ec2012-08-22 19:50:28 +0000386static bool isSimpleMSAsm(std::vector<std::vector<StringRef> > Pieces,
387 const TargetInfo &TI) {
388 for (unsigned i = 0, e = Pieces.size(); i != e; ++i)
389 if (!isSimpleMSAsm(Pieces[i], TI))
390 return false;
Chad Rosier98ac6082012-08-21 23:09:21 +0000391 return true;
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000392}
393
Chad Rosier358ab762012-08-22 21:08:06 +0000394// Break the AsmSting into pieces (i.e., mnemonic and operands).
Chad Rosier38c71d32012-08-21 21:56:39 +0000395static void buildMSAsmPieces(StringRef Asm, std::vector<StringRef> &Pieces) {
396 std::pair<StringRef,StringRef> Split = Asm.split(' ');
397
398 // Mnemonic
399 Pieces.push_back(Split.first);
400 Asm = Split.second;
401
402 // Operands
403 while (!Asm.empty()) {
404 Split = Asm.split(", ");
405 Pieces.push_back(Split.first);
406 Asm = Split.second;
407 }
408}
409
Chad Rosierf0fbd772012-08-22 21:04:07 +0000410static void buildMSAsmPieces(std::vector<std::string> &AsmStrings,
411 std::vector<std::vector<StringRef> > &Pieces) {
412 for (unsigned i = 0, e = AsmStrings.size(); i != e; ++i)
413 buildMSAsmPieces(AsmStrings[i], Pieces[i]);
414}
415
Chad Rosier682ad162012-08-22 21:12:19 +0000416// Build the unmodified AsmString used by the IR. Also build the individual
417// asm instruction(s) and place them in the AsmStrings vector; these are fed
418// to the AsmParser.
Chad Rosier9072a022012-08-22 20:30:58 +0000419static std::string buildMSAsmString(Sema &SemaRef, ArrayRef<Token> AsmToks,
420 std::vector<std::string> &AsmStrings,
421 std::vector<std::pair<unsigned,unsigned> > &AsmTokRanges) {
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000422 assert (!AsmToks.empty() && "Didn't expect an empty AsmToks!");
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000423
Chad Rosier38c71d32012-08-21 21:56:39 +0000424 SmallString<512> Res;
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000425 SmallString<512> Asm;
Chad Rosier9072a022012-08-22 20:30:58 +0000426 unsigned startTok = 0;
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000427 for (unsigned i = 0, e = AsmToks.size(); i < e; ++i) {
428 bool isNewAsm = i == 0 || AsmToks[i].isAtStartOfLine() ||
429 AsmToks[i].is(tok::kw_asm);
430
431 if (isNewAsm) {
Chad Rosier38c71d32012-08-21 21:56:39 +0000432 if (i) {
Benjamin Kramer32f3acc2012-08-24 20:43:21 +0000433 AsmStrings.push_back(Asm.str());
Chad Rosier9072a022012-08-22 20:30:58 +0000434 AsmTokRanges.push_back(std::make_pair(startTok, i-1));
435 startTok = i;
Chad Rosier38c71d32012-08-21 21:56:39 +0000436 Res += Asm;
437 Asm.clear();
438 Res += '\n';
439 }
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000440 if (AsmToks[i].is(tok::kw_asm)) {
441 i++; // Skip __asm
442 assert (i != e && "Expected another token");
443 }
444 }
445
446 if (i && AsmToks[i].hasLeadingSpace() && !isNewAsm)
447 Asm += ' ';
448
449 Asm += getSpelling(SemaRef, AsmToks[i]);
450 }
Benjamin Kramer32f3acc2012-08-24 20:43:21 +0000451 AsmStrings.push_back(Asm.str());
Chad Rosier9072a022012-08-22 20:30:58 +0000452 AsmTokRanges.push_back(std::make_pair(startTok, AsmToks.size()-1));
Chad Rosier38c71d32012-08-21 21:56:39 +0000453 Res += Asm;
Benjamin Kramer32f3acc2012-08-24 20:43:21 +0000454 return Res.str();
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000455}
456
Chad Rosierc6916492012-09-06 19:56:25 +0000457#define DEF_SIMPLE_MSASM(STR) \
Chad Rosier89fb6d72012-08-28 20:28:20 +0000458 MSAsmStmt *NS = \
459 new (Context) MSAsmStmt(Context, AsmLoc, LBraceLoc, /*IsSimple*/ true, \
460 /*IsVolatile*/ true, AsmToks, Inputs, Outputs, \
Chad Rosierc6916492012-09-06 19:56:25 +0000461 InputExprs, OutputExprs, STR, Constraints, \
Chad Rosier89fb6d72012-08-28 20:28:20 +0000462 Clobbers, EndLoc);
Chad Rosier2735df22012-08-22 19:18:30 +0000463
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000464StmtResult Sema::ActOnMSAsmStmt(SourceLocation AsmLoc,
465 SourceLocation LBraceLoc,
466 ArrayRef<Token> AsmToks,
467 SourceLocation EndLoc) {
Chad Rosier89fb6d72012-08-28 20:28:20 +0000468 SmallVector<StringRef, 4> Constraints;
469 std::vector<std::string> InputConstraints;
470 std::vector<std::string> OutputConstraints;
Chad Rosier4112a4c2012-08-28 20:35:06 +0000471 SmallVector<StringRef, 4> Clobbers;
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000472 std::set<std::string> ClobberRegs;
473 SmallVector<IdentifierInfo*, 4> Inputs;
474 SmallVector<IdentifierInfo*, 4> Outputs;
Chad Rosier633abb02012-08-24 00:07:09 +0000475 SmallVector<Expr*, 4> InputExprs;
476 SmallVector<Expr*, 4> OutputExprs;
Chad Rosieracc22b62012-09-06 19:35:00 +0000477 SmallVector<std::string, 4> InputExprNames;
478 SmallVector<std::string, 4> OutputExprNames;
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000479
480 // Empty asm statements don't need to instantiate the AsmParser, etc.
Chad Rosierc6916492012-09-06 19:56:25 +0000481 StringRef EmptyAsmStr;
482 if (AsmToks.empty()) { DEF_SIMPLE_MSASM(EmptyAsmStr); return Owned(NS); }
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000483
Chad Rosier38c71d32012-08-21 21:56:39 +0000484 std::vector<std::string> AsmStrings;
Chad Rosier9072a022012-08-22 20:30:58 +0000485 std::vector<std::pair<unsigned,unsigned> > AsmTokRanges;
Chad Rosier8a30e772012-08-24 15:51:10 +0000486 std::string AsmString = buildMSAsmString(*this, AsmToks, AsmStrings,
487 AsmTokRanges);
Chad Rosier38c71d32012-08-21 21:56:39 +0000488
Chad Rosiere78460f2012-08-22 20:57:07 +0000489 std::vector<std::vector<StringRef> > Pieces(AsmStrings.size());
Chad Rosierf0fbd772012-08-22 21:04:07 +0000490 buildMSAsmPieces(AsmStrings, Pieces);
Chad Rosier153f8ec2012-08-22 19:50:28 +0000491
492 bool IsSimple = isSimpleMSAsm(Pieces, Context.getTargetInfo());
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000493
Chad Rosier2735df22012-08-22 19:18:30 +0000494 // AsmParser doesn't fully support these asm statements.
Chad Rosierc6916492012-09-06 19:56:25 +0000495 if (bailOnMSAsm(Pieces)) { DEF_SIMPLE_MSASM(EmptyAsmStr); return Owned(NS); }
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000496
497 // Initialize targets and assembly printers/parsers.
498 llvm::InitializeAllTargetInfos();
499 llvm::InitializeAllTargetMCs();
500 llvm::InitializeAllAsmParsers();
501
502 // Get the target specific parser.
503 std::string Error;
504 const std::string &TT = Context.getTargetInfo().getTriple().getTriple();
505 const llvm::Target *TheTarget(llvm::TargetRegistry::lookupTarget(TT, Error));
506
507 OwningPtr<llvm::MCAsmInfo> MAI(TheTarget->createMCAsmInfo(TT));
508 OwningPtr<llvm::MCRegisterInfo> MRI(TheTarget->createMCRegInfo(TT));
509 OwningPtr<llvm::MCObjectFileInfo> MOFI(new llvm::MCObjectFileInfo());
510 OwningPtr<llvm::MCSubtargetInfo>
511 STI(TheTarget->createMCSubtargetInfo(TT, "", ""));
512
Chad Rosier25bd2982012-08-23 15:44:35 +0000513 for (unsigned StrIdx = 0, e = AsmStrings.size(); StrIdx != e; ++StrIdx) {
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000514 llvm::SourceMgr SrcMgr;
515 llvm::MCContext Ctx(*MAI, *MRI, MOFI.get(), &SrcMgr);
516 llvm::MemoryBuffer *Buffer =
Chad Rosier25bd2982012-08-23 15:44:35 +0000517 llvm::MemoryBuffer::getMemBuffer(AsmStrings[StrIdx], "<inline asm>");
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000518
519 // Tell SrcMgr about this buffer, which is what the parser will pick up.
520 SrcMgr.AddNewSourceBuffer(Buffer, llvm::SMLoc());
521
522 OwningPtr<llvm::MCStreamer> Str(createNullStreamer(Ctx));
523 OwningPtr<llvm::MCAsmParser>
524 Parser(createMCAsmParser(SrcMgr, Ctx, *Str.get(), *MAI));
525 OwningPtr<llvm::MCTargetAsmParser>
526 TargetParser(TheTarget->createMCAsmParser(*STI, *Parser));
527 // Change to the Intel dialect.
528 Parser->setAssemblerDialect(1);
529 Parser->setTargetParser(*TargetParser.get());
530
531 // Prime the lexer.
532 Parser->Lex();
533
534 // Parse the opcode.
535 StringRef IDVal;
536 Parser->ParseIdentifier(IDVal);
537
538 // Canonicalize the opcode to lower case.
Chad Rosierbe5c3fb2012-09-03 02:30:13 +0000539 SmallString<128> OpcodeStr;
Chad Rosierb706d902012-08-28 22:08:58 +0000540 for (unsigned i = 0, e = IDVal.size(); i != e; ++i)
Chad Rosierbe5c3fb2012-09-03 02:30:13 +0000541 OpcodeStr.push_back(tolower(IDVal[i]));
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000542
543 // Parse the operands.
544 llvm::SMLoc IDLoc;
545 SmallVector<llvm::MCParsedAsmOperand*, 8> Operands;
Chad Rosierbe5c3fb2012-09-03 02:30:13 +0000546 bool HadError = TargetParser->ParseInstruction(OpcodeStr.str(), IDLoc,
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000547 Operands);
Chad Rosier2735df22012-08-22 19:18:30 +0000548 // If we had an error parsing the operands, fail gracefully.
Chad Rosierc6916492012-09-06 19:56:25 +0000549 if (HadError) { DEF_SIMPLE_MSASM(EmptyAsmStr); return Owned(NS); }
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000550
551 // Match the MCInstr.
Chad Rosierbe5c3fb2012-09-03 02:30:13 +0000552 unsigned Kind;
Chad Rosier83591b62012-08-21 18:15:08 +0000553 unsigned ErrorInfo;
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000554 SmallVector<llvm::MCInst, 2> Instrs;
Chad Rosier7065c522012-09-03 03:16:15 +0000555 HadError = TargetParser->MatchInstruction(IDLoc, Kind, Operands, Instrs,
556 ErrorInfo,
Chad Rosier51a6b3f2012-08-21 19:37:55 +0000557 /*matchingInlineAsm*/ true);
Chad Rosier2735df22012-08-22 19:18:30 +0000558 // If we had an error parsing the operands, fail gracefully.
Chad Rosierc6916492012-09-06 19:56:25 +0000559 if (HadError) { DEF_SIMPLE_MSASM(EmptyAsmStr); return Owned(NS); }
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000560
561 // Get the instruction descriptor.
562 llvm::MCInst Inst = Instrs[0];
563 const llvm::MCInstrInfo *MII = TheTarget->createMCInstrInfo();
564 const llvm::MCInstrDesc &Desc = MII->get(Inst.getOpcode());
565 llvm::MCInstPrinter *IP =
566 TheTarget->createMCInstPrinter(1, *MAI, *MII, *MRI, *STI);
567
Chad Rosier6e97be72012-08-22 23:42:09 +0000568 // Build the list of clobbers, outputs and inputs.
Chad Rosierfd5e56e2012-08-22 22:10:51 +0000569 unsigned NumDefs = Desc.getNumDefs();
Chad Rosier1b497f22012-09-03 20:40:52 +0000570 for (unsigned i = 1, e = Operands.size(); i != e; ++i) {
571 unsigned NumMCOperands;
Chad Rosier8cdd8a92012-09-05 01:16:06 +0000572 unsigned MCIdx = TargetParser->getMCInstOperandNum(Kind, Inst, Operands,
Chad Rosierd5eb5852012-09-04 15:58:44 +0000573 i, NumMCOperands);
Chad Rosier1b497f22012-09-03 20:40:52 +0000574 assert (NumMCOperands && "Expected at least 1 MCOperand!");
575 // If we have a one-to-many mapping, then search for the MCExpr.
576 if (NumMCOperands > 1) {
577 bool foundExpr = false;
578 for (unsigned j = MCIdx, e = MCIdx + NumMCOperands; j != e; ++j) {
579 if (Inst.getOperand(j).isExpr()) {
580 foundExpr = true;
581 MCIdx = j;
582 break;
583 }
584 }
585 assert (foundExpr && "Expected for find an expression!");
586 }
587
588 const llvm::MCOperand &Op = Inst.getOperand(MCIdx);
Chad Rosierfd5e56e2012-08-22 22:10:51 +0000589
590 // Immediate.
591 if (Op.isImm() || Op.isFPImm())
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000592 continue;
593
Chad Rosier1b497f22012-09-03 20:40:52 +0000594 bool isDef = NumDefs && (MCIdx < NumDefs);
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000595
Chad Rosierfd5e56e2012-08-22 22:10:51 +0000596 // Register/Clobber.
597 if (Op.isReg() && isDef) {
598 std::string Reg;
599 llvm::raw_string_ostream OS(Reg);
600 IP->printRegName(OS, Op.getReg());
601
602 StringRef Clobber(OS.str());
603 if (!Context.getTargetInfo().isValidClobber(Clobber))
604 return StmtError(Diag(AsmLoc, diag::err_asm_unknown_register_name) <<
605 Clobber);
606 ClobberRegs.insert(Reg);
Chad Rosier6e97be72012-08-22 23:42:09 +0000607 continue;
608 }
609 // Expr/Input or Output.
610 if (Op.isExpr()) {
611 const llvm::MCExpr *Expr = Op.getExpr();
612 const llvm::MCSymbolRefExpr *SymRef;
613 if ((SymRef = dyn_cast<llvm::MCSymbolRefExpr>(Expr))) {
614 StringRef Name = SymRef->getSymbol().getName();
Chad Rosier60ce5842012-08-23 00:12:51 +0000615 IdentifierInfo *II = getIdentifierInfo(Name, AsmToks,
Chad Rosier25bd2982012-08-23 15:44:35 +0000616 AsmTokRanges[StrIdx].first,
617 AsmTokRanges[StrIdx].second);
Chad Rosier633abb02012-08-24 00:07:09 +0000618 if (II) {
Chad Rosier1016bdf2012-08-24 16:38:58 +0000619 CXXScopeSpec SS;
620 UnqualifiedId Id;
621 SourceLocation Loc;
622 Id.setIdentifier(II, AsmLoc);
623 ExprResult Result = ActOnIdExpression(getCurScope(), SS, Loc, Id,
624 false, false);
625 if (!Result.isInvalid()) {
Chad Rosier21a37042012-09-04 16:39:38 +0000626 bool isMemDef = (i == 1) && Desc.mayStore();
627 if (isDef || isMemDef) {
Chad Rosier1016bdf2012-08-24 16:38:58 +0000628 Outputs.push_back(II);
629 OutputExprs.push_back(Result.take());
Chad Rosieracc22b62012-09-06 19:35:00 +0000630 OutputExprNames.push_back(Name.str());
Chad Rosier89fb6d72012-08-28 20:28:20 +0000631 OutputConstraints.push_back("=r");
Chad Rosier1016bdf2012-08-24 16:38:58 +0000632 } else {
633 Inputs.push_back(II);
634 InputExprs.push_back(Result.take());
Chad Rosieracc22b62012-09-06 19:35:00 +0000635 InputExprNames.push_back(Name.str());
Chad Rosier89fb6d72012-08-28 20:28:20 +0000636 InputConstraints.push_back("r");
Chad Rosier1016bdf2012-08-24 16:38:58 +0000637 }
Chad Rosier633abb02012-08-24 00:07:09 +0000638 }
639 }
Chad Rosier6e97be72012-08-22 23:42:09 +0000640 }
Chad Rosierfd5e56e2012-08-22 22:10:51 +0000641 }
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000642 }
643 }
644 for (std::set<std::string>::iterator I = ClobberRegs.begin(),
645 E = ClobberRegs.end(); I != E; ++I)
646 Clobbers.push_back(*I);
647
Chad Rosier89fb6d72012-08-28 20:28:20 +0000648 // Merge the output and input constraints. Output constraints are expected
649 // first.
650 for (std::vector<std::string>::iterator I = OutputConstraints.begin(),
651 E = OutputConstraints.end(); I != E; ++I)
652 Constraints.push_back(*I);
653
654 for (std::vector<std::string>::iterator I = InputConstraints.begin(),
655 E = InputConstraints.end(); I != E; ++I)
656 Constraints.push_back(*I);
657
Chad Rosieracc22b62012-09-06 19:35:00 +0000658 // Enumerate the AsmString expressions.
659 // FIXME: This isn't going to work if:
660 // 1. The symbol name and an opcode/reg share the same, or are a substring of
661 // the, name.
662 // 2. The symbol name appears more then once in the asm string.
663 unsigned OpNum = 0;
664 for (unsigned i = 0, e = OutputExprNames.size(); i != e; ++i, ++OpNum) {
665 size_t found = AsmString.find(OutputExprNames[i]);
666 SmallString<32> Res;
667 llvm::raw_svector_ostream OS(Res);
668 OS << '$' << OpNum;
669 AsmString.replace(found, OutputExprNames[i].size(), OS.str());
670 }
671 for (unsigned i = 0, e = InputExprNames.size(); i != e; ++i, ++OpNum) {
672 size_t found = AsmString.find(InputExprNames[i]);
673 SmallString<32> Res;
674 llvm::raw_svector_ostream OS(Res);
675 OS << '$' << OpNum;
676 AsmString.replace(found, InputExprNames[i].size(), OS.str());
677 }
678
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000679 MSAsmStmt *NS =
680 new (Context) MSAsmStmt(Context, AsmLoc, LBraceLoc, IsSimple,
681 /*IsVolatile*/ true, AsmToks, Inputs, Outputs,
Chad Rosier89fb6d72012-08-28 20:28:20 +0000682 InputExprs, OutputExprs, AsmString, Constraints,
683 Clobbers, EndLoc);
Chad Rosier4b5e48d2012-08-17 21:19:40 +0000684 return Owned(NS);
685}