blob: 612c4dc4cbfa3fa1382f9d00fdf10db809eca2e0 [file] [log] [blame]
Chris Lattner59907c42007-08-10 20:18:51 +00001//===--- SemaChecking.cpp - Extra Semantic Checking -----------------------===//
2//
3// The LLVM Compiler Infrastructure
4//
Chris Lattner0bc735f2007-12-29 19:59:25 +00005// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
Chris Lattner59907c42007-08-10 20:18:51 +00007//
8//===----------------------------------------------------------------------===//
9//
10// This file implements extra semantic analysis beyond what is enforced
11// by the C type system.
12//
13//===----------------------------------------------------------------------===//
14
15#include "Sema.h"
16#include "clang/AST/ASTContext.h"
Daniel Dunbarc4a1dea2008-08-11 05:35:13 +000017#include "clang/AST/DeclObjC.h"
Ted Kremenek23245122007-08-20 16:18:38 +000018#include "clang/AST/ExprCXX.h"
Ted Kremenek7ff22b22008-06-16 18:00:42 +000019#include "clang/AST/ExprObjC.h"
Chris Lattner719e6152009-02-18 19:21:10 +000020#include "clang/Lex/LiteralSupport.h"
Chris Lattner59907c42007-08-10 20:18:51 +000021#include "clang/Lex/Preprocessor.h"
Zhongxing Xua1f3dba2009-05-20 01:55:10 +000022#include <limits>
Chris Lattner59907c42007-08-10 20:18:51 +000023using namespace clang;
24
Chris Lattner60800082009-02-18 17:49:48 +000025/// getLocationOfStringLiteralByte - Return a source location that points to the
26/// specified byte of the specified string literal.
27///
28/// Strings are amazingly complex. They can be formed from multiple tokens and
29/// can have escape sequences in them in addition to the usual trigraph and
30/// escaped newline business. This routine handles this complexity.
31///
32SourceLocation Sema::getLocationOfStringLiteralByte(const StringLiteral *SL,
33 unsigned ByteNo) const {
34 assert(!SL->isWide() && "This doesn't work for wide strings yet");
35
36 // Loop over all of the tokens in this string until we find the one that
37 // contains the byte we're looking for.
38 unsigned TokNo = 0;
39 while (1) {
40 assert(TokNo < SL->getNumConcatenated() && "Invalid byte number!");
41 SourceLocation StrTokLoc = SL->getStrTokenLoc(TokNo);
42
43 // Get the spelling of the string so that we can get the data that makes up
44 // the string literal, not the identifier for the macro it is potentially
45 // expanded through.
46 SourceLocation StrTokSpellingLoc = SourceMgr.getSpellingLoc(StrTokLoc);
47
48 // Re-lex the token to get its length and original spelling.
49 std::pair<FileID, unsigned> LocInfo =
50 SourceMgr.getDecomposedLoc(StrTokSpellingLoc);
51 std::pair<const char *,const char *> Buffer =
52 SourceMgr.getBufferData(LocInfo.first);
53 const char *StrData = Buffer.first+LocInfo.second;
54
55 // Create a langops struct and enable trigraphs. This is sufficient for
56 // relexing tokens.
57 LangOptions LangOpts;
58 LangOpts.Trigraphs = true;
59
60 // Create a lexer starting at the beginning of this token.
61 Lexer TheLexer(StrTokSpellingLoc, LangOpts, Buffer.first, StrData,
62 Buffer.second);
63 Token TheTok;
64 TheLexer.LexFromRawLexer(TheTok);
65
Chris Lattner443e53c2009-02-18 19:26:42 +000066 // Use the StringLiteralParser to compute the length of the string in bytes.
67 StringLiteralParser SLP(&TheTok, 1, PP);
68 unsigned TokNumBytes = SLP.GetStringLength();
Chris Lattnerd0d082f2009-02-18 18:34:12 +000069
Chris Lattner2197c962009-02-18 18:52:52 +000070 // If the byte is in this token, return the location of the byte.
Chris Lattner60800082009-02-18 17:49:48 +000071 if (ByteNo < TokNumBytes ||
72 (ByteNo == TokNumBytes && TokNo == SL->getNumConcatenated())) {
Chris Lattner719e6152009-02-18 19:21:10 +000073 unsigned Offset =
74 StringLiteralParser::getOffsetOfStringByte(TheTok, ByteNo, PP);
75
76 // Now that we know the offset of the token in the spelling, use the
77 // preprocessor to get the offset in the original source.
78 return PP.AdvanceToTokenCharacter(StrTokLoc, Offset);
Chris Lattner60800082009-02-18 17:49:48 +000079 }
80
81 // Move to the next string token.
82 ++TokNo;
83 ByteNo -= TokNumBytes;
84 }
85}
86
Ryan Flynn4403a5e2009-08-06 03:00:50 +000087/// CheckablePrintfAttr - does a function call have a "printf" attribute
88/// and arguments that merit checking?
89bool Sema::CheckablePrintfAttr(const FormatAttr *Format, CallExpr *TheCall) {
90 if (Format->getType() == "printf") return true;
91 if (Format->getType() == "printf0") {
92 // printf0 allows null "format" string; if so don't check format/args
93 unsigned format_idx = Format->getFormatIdx() - 1;
94 if (format_idx < TheCall->getNumArgs()) {
95 Expr *Format = TheCall->getArg(format_idx)->IgnoreParenCasts();
96 if (!Format->isNullPointerConstant(Context))
97 return true;
98 }
99 }
100 return false;
101}
Chris Lattner60800082009-02-18 17:49:48 +0000102
Sebastian Redl0eb23302009-01-19 00:08:26 +0000103Action::OwningExprResult
Anders Carlssond406bf02009-08-16 01:56:34 +0000104Sema::CheckBuiltinFunctionCall(unsigned BuiltinID, CallExpr *TheCall) {
Sebastian Redl0eb23302009-01-19 00:08:26 +0000105 OwningExprResult TheCallResult(Owned(TheCall));
Douglas Gregor2def4832008-11-17 20:34:05 +0000106
Anders Carlssond406bf02009-08-16 01:56:34 +0000107 switch (BuiltinID) {
Chris Lattner30ce3442007-12-19 23:59:04 +0000108 case Builtin::BI__builtin___CFStringMakeConstantString:
Chris Lattner925e60d2007-12-28 05:29:59 +0000109 assert(TheCall->getNumArgs() == 1 &&
Chris Lattner1b9a0792007-12-20 00:26:33 +0000110 "Wrong # arguments to builtin CFStringMakeConstantString");
Chris Lattner69039812009-02-18 06:01:06 +0000111 if (CheckObjCString(TheCall->getArg(0)))
Sebastian Redl0eb23302009-01-19 00:08:26 +0000112 return ExprError();
Anders Carlssond406bf02009-08-16 01:56:34 +0000113 break;
Ted Kremenek49ff7a12008-07-09 17:58:53 +0000114 case Builtin::BI__builtin_stdarg_start:
Chris Lattner30ce3442007-12-19 23:59:04 +0000115 case Builtin::BI__builtin_va_start:
Sebastian Redl0eb23302009-01-19 00:08:26 +0000116 if (SemaBuiltinVAStart(TheCall))
117 return ExprError();
Anders Carlssond406bf02009-08-16 01:56:34 +0000118 break;
Chris Lattner1b9a0792007-12-20 00:26:33 +0000119 case Builtin::BI__builtin_isgreater:
120 case Builtin::BI__builtin_isgreaterequal:
121 case Builtin::BI__builtin_isless:
122 case Builtin::BI__builtin_islessequal:
123 case Builtin::BI__builtin_islessgreater:
124 case Builtin::BI__builtin_isunordered:
Sebastian Redl0eb23302009-01-19 00:08:26 +0000125 if (SemaBuiltinUnorderedCompare(TheCall))
126 return ExprError();
Anders Carlssond406bf02009-08-16 01:56:34 +0000127 break;
Eli Friedman6cfda232008-05-20 08:23:37 +0000128 case Builtin::BI__builtin_return_address:
129 case Builtin::BI__builtin_frame_address:
Sebastian Redl0eb23302009-01-19 00:08:26 +0000130 if (SemaBuiltinStackAddress(TheCall))
131 return ExprError();
Anders Carlssond406bf02009-08-16 01:56:34 +0000132 break;
Eli Friedmand38617c2008-05-14 19:38:39 +0000133 case Builtin::BI__builtin_shufflevector:
Sebastian Redl0eb23302009-01-19 00:08:26 +0000134 return SemaBuiltinShuffleVector(TheCall);
135 // TheCall will be freed by the smart pointer here, but that's fine, since
136 // SemaBuiltinShuffleVector guts it, but then doesn't release it.
Daniel Dunbar4493f792008-07-21 22:59:13 +0000137 case Builtin::BI__builtin_prefetch:
Sebastian Redl0eb23302009-01-19 00:08:26 +0000138 if (SemaBuiltinPrefetch(TheCall))
139 return ExprError();
Anders Carlssond406bf02009-08-16 01:56:34 +0000140 break;
Daniel Dunbard5f8a4f2008-09-03 21:13:56 +0000141 case Builtin::BI__builtin_object_size:
Sebastian Redl0eb23302009-01-19 00:08:26 +0000142 if (SemaBuiltinObjectSize(TheCall))
143 return ExprError();
Anders Carlssond406bf02009-08-16 01:56:34 +0000144 break;
Eli Friedmand875fed2009-05-03 04:46:36 +0000145 case Builtin::BI__builtin_longjmp:
146 if (SemaBuiltinLongjmp(TheCall))
147 return ExprError();
Anders Carlssond406bf02009-08-16 01:56:34 +0000148 break;
Chris Lattner5caa3702009-05-08 06:58:22 +0000149 case Builtin::BI__sync_fetch_and_add:
150 case Builtin::BI__sync_fetch_and_sub:
151 case Builtin::BI__sync_fetch_and_or:
152 case Builtin::BI__sync_fetch_and_and:
153 case Builtin::BI__sync_fetch_and_xor:
Chris Lattnereebd9d22009-05-13 04:37:52 +0000154 case Builtin::BI__sync_fetch_and_nand:
Chris Lattner5caa3702009-05-08 06:58:22 +0000155 case Builtin::BI__sync_add_and_fetch:
156 case Builtin::BI__sync_sub_and_fetch:
157 case Builtin::BI__sync_and_and_fetch:
158 case Builtin::BI__sync_or_and_fetch:
159 case Builtin::BI__sync_xor_and_fetch:
Chris Lattnereebd9d22009-05-13 04:37:52 +0000160 case Builtin::BI__sync_nand_and_fetch:
Chris Lattner5caa3702009-05-08 06:58:22 +0000161 case Builtin::BI__sync_val_compare_and_swap:
162 case Builtin::BI__sync_bool_compare_and_swap:
163 case Builtin::BI__sync_lock_test_and_set:
164 case Builtin::BI__sync_lock_release:
165 if (SemaBuiltinAtomicOverloaded(TheCall))
166 return ExprError();
Anders Carlssond406bf02009-08-16 01:56:34 +0000167 break;
Anders Carlsson71993dd2007-08-17 05:31:46 +0000168 }
Anders Carlssond406bf02009-08-16 01:56:34 +0000169
170 return move(TheCallResult);
171}
Daniel Dunbarde454282008-10-02 18:44:07 +0000172
Anders Carlssond406bf02009-08-16 01:56:34 +0000173/// CheckFunctionCall - Check a direct function call for various correctness
174/// and safety properties not strictly enforced by the C type system.
175bool Sema::CheckFunctionCall(FunctionDecl *FDecl, CallExpr *TheCall) {
176 // Get the IdentifierInfo* for the called function.
177 IdentifierInfo *FnInfo = FDecl->getIdentifier();
178
179 // None of the checks below are needed for functions that don't have
180 // simple names (e.g., C++ conversion functions).
181 if (!FnInfo)
182 return false;
183
Daniel Dunbarde454282008-10-02 18:44:07 +0000184 // FIXME: This mechanism should be abstracted to be less fragile and
185 // more efficient. For example, just map function ids to custom
186 // handlers.
187
Chris Lattner59907c42007-08-10 20:18:51 +0000188 // Printf checking.
Argyrios Kyrtzidis40b598e2009-06-30 02:34:44 +0000189 if (const FormatAttr *Format = FDecl->getAttr<FormatAttr>()) {
Ryan Flynn4403a5e2009-08-06 03:00:50 +0000190 if (CheckablePrintfAttr(Format, TheCall)) {
Ted Kremenek3d692df2009-02-27 17:58:43 +0000191 bool HasVAListArg = Format->getFirstArg() == 0;
192 if (!HasVAListArg) {
193 if (const FunctionProtoType *Proto
194 = FDecl->getType()->getAsFunctionProtoType())
Douglas Gregor3c385e52009-02-14 18:57:46 +0000195 HasVAListArg = !Proto->isVariadic();
Ted Kremenek3d692df2009-02-27 17:58:43 +0000196 }
Douglas Gregor3c385e52009-02-14 18:57:46 +0000197 CheckPrintfArguments(TheCall, HasVAListArg, Format->getFormatIdx() - 1,
Ted Kremenek3d692df2009-02-27 17:58:43 +0000198 HasVAListArg ? 0 : Format->getFirstArg() - 1);
Douglas Gregor3c385e52009-02-14 18:57:46 +0000199 }
Chris Lattner59907c42007-08-10 20:18:51 +0000200 }
Anders Carlssond406bf02009-08-16 01:56:34 +0000201
202 for (const NonNullAttr *NonNull = FDecl->getAttr<NonNullAttr>(); NonNull;
203 NonNull = NonNull->getNext<NonNullAttr>())
204 CheckNonNullArguments(NonNull, TheCall);
Sebastian Redl0eb23302009-01-19 00:08:26 +0000205
Anders Carlssond406bf02009-08-16 01:56:34 +0000206 return false;
Anders Carlsson71993dd2007-08-17 05:31:46 +0000207}
208
Anders Carlssond406bf02009-08-16 01:56:34 +0000209bool Sema::CheckBlockCall(NamedDecl *NDecl, CallExpr *TheCall) {
Fariborz Jahanian725165f2009-05-18 21:05:18 +0000210 // Printf checking.
Argyrios Kyrtzidis40b598e2009-06-30 02:34:44 +0000211 const FormatAttr *Format = NDecl->getAttr<FormatAttr>();
Fariborz Jahanian725165f2009-05-18 21:05:18 +0000212 if (!Format)
Anders Carlssond406bf02009-08-16 01:56:34 +0000213 return false;
214
Fariborz Jahanian725165f2009-05-18 21:05:18 +0000215 const VarDecl *V = dyn_cast<VarDecl>(NDecl);
216 if (!V)
Anders Carlssond406bf02009-08-16 01:56:34 +0000217 return false;
218
Fariborz Jahanian725165f2009-05-18 21:05:18 +0000219 QualType Ty = V->getType();
220 if (!Ty->isBlockPointerType())
Anders Carlssond406bf02009-08-16 01:56:34 +0000221 return false;
222
223 if (!CheckablePrintfAttr(Format, TheCall))
224 return false;
225
226 bool HasVAListArg = Format->getFirstArg() == 0;
227 if (!HasVAListArg) {
228 const FunctionType *FT =
229 Ty->getAs<BlockPointerType>()->getPointeeType()->getAsFunctionType();
230 if (const FunctionProtoType *Proto = dyn_cast<FunctionProtoType>(FT))
231 HasVAListArg = !Proto->isVariadic();
Fariborz Jahanian725165f2009-05-18 21:05:18 +0000232 }
Anders Carlssond406bf02009-08-16 01:56:34 +0000233 CheckPrintfArguments(TheCall, HasVAListArg, Format->getFormatIdx() - 1,
234 HasVAListArg ? 0 : Format->getFirstArg() - 1);
235
236 return false;
Fariborz Jahanian725165f2009-05-18 21:05:18 +0000237}
238
Chris Lattner5caa3702009-05-08 06:58:22 +0000239/// SemaBuiltinAtomicOverloaded - We have a call to a function like
240/// __sync_fetch_and_add, which is an overloaded function based on the pointer
241/// type of its first argument. The main ActOnCallExpr routines have already
242/// promoted the types of arguments because all of these calls are prototyped as
243/// void(...).
244///
245/// This function goes through and does final semantic checking for these
246/// builtins,
247bool Sema::SemaBuiltinAtomicOverloaded(CallExpr *TheCall) {
248 DeclRefExpr *DRE =cast<DeclRefExpr>(TheCall->getCallee()->IgnoreParenCasts());
249 FunctionDecl *FDecl = cast<FunctionDecl>(DRE->getDecl());
250
251 // Ensure that we have at least one argument to do type inference from.
252 if (TheCall->getNumArgs() < 1)
253 return Diag(TheCall->getLocEnd(), diag::err_typecheck_call_too_few_args)
254 << 0 << TheCall->getCallee()->getSourceRange();
255
256 // Inspect the first argument of the atomic builtin. This should always be
257 // a pointer type, whose element is an integral scalar or pointer type.
258 // Because it is a pointer type, we don't have to worry about any implicit
259 // casts here.
260 Expr *FirstArg = TheCall->getArg(0);
261 if (!FirstArg->getType()->isPointerType())
262 return Diag(DRE->getLocStart(), diag::err_atomic_builtin_must_be_pointer)
263 << FirstArg->getType() << FirstArg->getSourceRange();
264
Ted Kremenek6217b802009-07-29 21:53:49 +0000265 QualType ValType = FirstArg->getType()->getAs<PointerType>()->getPointeeType();
Chris Lattner5caa3702009-05-08 06:58:22 +0000266 if (!ValType->isIntegerType() && !ValType->isPointerType() &&
267 !ValType->isBlockPointerType())
268 return Diag(DRE->getLocStart(),
269 diag::err_atomic_builtin_must_be_pointer_intptr)
270 << FirstArg->getType() << FirstArg->getSourceRange();
271
272 // We need to figure out which concrete builtin this maps onto. For example,
273 // __sync_fetch_and_add with a 2 byte object turns into
274 // __sync_fetch_and_add_2.
275#define BUILTIN_ROW(x) \
276 { Builtin::BI##x##_1, Builtin::BI##x##_2, Builtin::BI##x##_4, \
277 Builtin::BI##x##_8, Builtin::BI##x##_16 }
278
279 static const unsigned BuiltinIndices[][5] = {
280 BUILTIN_ROW(__sync_fetch_and_add),
281 BUILTIN_ROW(__sync_fetch_and_sub),
282 BUILTIN_ROW(__sync_fetch_and_or),
283 BUILTIN_ROW(__sync_fetch_and_and),
284 BUILTIN_ROW(__sync_fetch_and_xor),
Chris Lattnereebd9d22009-05-13 04:37:52 +0000285 BUILTIN_ROW(__sync_fetch_and_nand),
Chris Lattner5caa3702009-05-08 06:58:22 +0000286
287 BUILTIN_ROW(__sync_add_and_fetch),
288 BUILTIN_ROW(__sync_sub_and_fetch),
289 BUILTIN_ROW(__sync_and_and_fetch),
290 BUILTIN_ROW(__sync_or_and_fetch),
291 BUILTIN_ROW(__sync_xor_and_fetch),
Chris Lattnereebd9d22009-05-13 04:37:52 +0000292 BUILTIN_ROW(__sync_nand_and_fetch),
Chris Lattner5caa3702009-05-08 06:58:22 +0000293
294 BUILTIN_ROW(__sync_val_compare_and_swap),
295 BUILTIN_ROW(__sync_bool_compare_and_swap),
296 BUILTIN_ROW(__sync_lock_test_and_set),
297 BUILTIN_ROW(__sync_lock_release)
298 };
299#undef BUILTIN_ROW
300
301 // Determine the index of the size.
302 unsigned SizeIndex;
303 switch (Context.getTypeSize(ValType)/8) {
304 case 1: SizeIndex = 0; break;
305 case 2: SizeIndex = 1; break;
306 case 4: SizeIndex = 2; break;
307 case 8: SizeIndex = 3; break;
308 case 16: SizeIndex = 4; break;
309 default:
310 return Diag(DRE->getLocStart(), diag::err_atomic_builtin_pointer_size)
311 << FirstArg->getType() << FirstArg->getSourceRange();
312 }
313
314 // Each of these builtins has one pointer argument, followed by some number of
315 // values (0, 1 or 2) followed by a potentially empty varags list of stuff
316 // that we ignore. Find out which row of BuiltinIndices to read from as well
317 // as the number of fixed args.
318 unsigned BuiltinID = FDecl->getBuiltinID(Context);
319 unsigned BuiltinIndex, NumFixed = 1;
320 switch (BuiltinID) {
321 default: assert(0 && "Unknown overloaded atomic builtin!");
322 case Builtin::BI__sync_fetch_and_add: BuiltinIndex = 0; break;
323 case Builtin::BI__sync_fetch_and_sub: BuiltinIndex = 1; break;
324 case Builtin::BI__sync_fetch_and_or: BuiltinIndex = 2; break;
325 case Builtin::BI__sync_fetch_and_and: BuiltinIndex = 3; break;
326 case Builtin::BI__sync_fetch_and_xor: BuiltinIndex = 4; break;
Chris Lattnereebd9d22009-05-13 04:37:52 +0000327 case Builtin::BI__sync_fetch_and_nand:BuiltinIndex = 5; break;
Chris Lattner5caa3702009-05-08 06:58:22 +0000328
Chris Lattnereebd9d22009-05-13 04:37:52 +0000329 case Builtin::BI__sync_add_and_fetch: BuiltinIndex = 6; break;
330 case Builtin::BI__sync_sub_and_fetch: BuiltinIndex = 7; break;
331 case Builtin::BI__sync_and_and_fetch: BuiltinIndex = 8; break;
332 case Builtin::BI__sync_or_and_fetch: BuiltinIndex = 9; break;
333 case Builtin::BI__sync_xor_and_fetch: BuiltinIndex =10; break;
334 case Builtin::BI__sync_nand_and_fetch:BuiltinIndex =11; break;
Chris Lattner5caa3702009-05-08 06:58:22 +0000335
336 case Builtin::BI__sync_val_compare_and_swap:
Chris Lattnereebd9d22009-05-13 04:37:52 +0000337 BuiltinIndex = 12;
Chris Lattner5caa3702009-05-08 06:58:22 +0000338 NumFixed = 2;
339 break;
340 case Builtin::BI__sync_bool_compare_and_swap:
Chris Lattnereebd9d22009-05-13 04:37:52 +0000341 BuiltinIndex = 13;
Chris Lattner5caa3702009-05-08 06:58:22 +0000342 NumFixed = 2;
343 break;
Chris Lattnereebd9d22009-05-13 04:37:52 +0000344 case Builtin::BI__sync_lock_test_and_set: BuiltinIndex = 14; break;
Chris Lattner5caa3702009-05-08 06:58:22 +0000345 case Builtin::BI__sync_lock_release:
Chris Lattnereebd9d22009-05-13 04:37:52 +0000346 BuiltinIndex = 15;
Chris Lattner5caa3702009-05-08 06:58:22 +0000347 NumFixed = 0;
348 break;
349 }
350
351 // Now that we know how many fixed arguments we expect, first check that we
352 // have at least that many.
353 if (TheCall->getNumArgs() < 1+NumFixed)
354 return Diag(TheCall->getLocEnd(), diag::err_typecheck_call_too_few_args)
355 << 0 << TheCall->getCallee()->getSourceRange();
356
Chris Lattnere7ac0a92009-05-08 15:36:58 +0000357
358 // Get the decl for the concrete builtin from this, we can tell what the
359 // concrete integer type we should convert to is.
360 unsigned NewBuiltinID = BuiltinIndices[BuiltinIndex][SizeIndex];
361 const char *NewBuiltinName = Context.BuiltinInfo.GetName(NewBuiltinID);
362 IdentifierInfo *NewBuiltinII = PP.getIdentifierInfo(NewBuiltinName);
363 FunctionDecl *NewBuiltinDecl =
364 cast<FunctionDecl>(LazilyCreateBuiltin(NewBuiltinII, NewBuiltinID,
365 TUScope, false, DRE->getLocStart()));
366 const FunctionProtoType *BuiltinFT =
367 NewBuiltinDecl->getType()->getAsFunctionProtoType();
Ted Kremenek6217b802009-07-29 21:53:49 +0000368 ValType = BuiltinFT->getArgType(0)->getAs<PointerType>()->getPointeeType();
Chris Lattnere7ac0a92009-05-08 15:36:58 +0000369
370 // If the first type needs to be converted (e.g. void** -> int*), do it now.
371 if (BuiltinFT->getArgType(0) != FirstArg->getType()) {
Anders Carlsson3503d042009-07-31 01:23:52 +0000372 ImpCastExprToType(FirstArg, BuiltinFT->getArgType(0), CastExpr::CK_Unknown,
373 /*isLvalue=*/false);
Chris Lattnere7ac0a92009-05-08 15:36:58 +0000374 TheCall->setArg(0, FirstArg);
375 }
376
Chris Lattner5caa3702009-05-08 06:58:22 +0000377 // Next, walk the valid ones promoting to the right type.
378 for (unsigned i = 0; i != NumFixed; ++i) {
379 Expr *Arg = TheCall->getArg(i+1);
380
381 // If the argument is an implicit cast, then there was a promotion due to
382 // "...", just remove it now.
383 if (ImplicitCastExpr *ICE = dyn_cast<ImplicitCastExpr>(Arg)) {
384 Arg = ICE->getSubExpr();
385 ICE->setSubExpr(0);
386 ICE->Destroy(Context);
387 TheCall->setArg(i+1, Arg);
388 }
389
390 // GCC does an implicit conversion to the pointer or integer ValType. This
391 // can fail in some cases (1i -> int**), check for this error case now.
Anders Carlssoncdb61972009-08-07 22:21:05 +0000392 CastExpr::CastKind Kind = CastExpr::CK_Unknown;
393 if (CheckCastTypes(Arg->getSourceRange(), ValType, Arg, Kind))
Chris Lattner5caa3702009-05-08 06:58:22 +0000394 return true;
395
396 // Okay, we have something that *can* be converted to the right type. Check
397 // to see if there is a potentially weird extension going on here. This can
398 // happen when you do an atomic operation on something like an char* and
399 // pass in 42. The 42 gets converted to char. This is even more strange
400 // for things like 45.123 -> char, etc.
401 // FIXME: Do this check.
Anders Carlssoncdb61972009-08-07 22:21:05 +0000402 ImpCastExprToType(Arg, ValType, Kind, /*isLvalue=*/false);
Chris Lattner5caa3702009-05-08 06:58:22 +0000403 TheCall->setArg(i+1, Arg);
404 }
405
Chris Lattner5caa3702009-05-08 06:58:22 +0000406 // Switch the DeclRefExpr to refer to the new decl.
407 DRE->setDecl(NewBuiltinDecl);
408 DRE->setType(NewBuiltinDecl->getType());
409
410 // Set the callee in the CallExpr.
411 // FIXME: This leaks the original parens and implicit casts.
412 Expr *PromotedCall = DRE;
413 UsualUnaryConversions(PromotedCall);
414 TheCall->setCallee(PromotedCall);
415
416
417 // Change the result type of the call to match the result type of the decl.
418 TheCall->setType(NewBuiltinDecl->getResultType());
419 return false;
420}
421
422
Chris Lattner69039812009-02-18 06:01:06 +0000423/// CheckObjCString - Checks that the argument to the builtin
Anders Carlsson71993dd2007-08-17 05:31:46 +0000424/// CFString constructor is correct
Steve Narofffd942622009-04-13 20:26:29 +0000425/// FIXME: GCC currently emits the following warning:
426/// "warning: input conversion stopped due to an input byte that does not
427/// belong to the input codeset UTF-8"
428/// Note: It might also make sense to do the UTF-16 conversion here (would
429/// simplify the backend).
Chris Lattner69039812009-02-18 06:01:06 +0000430bool Sema::CheckObjCString(Expr *Arg) {
Chris Lattner56f34942008-02-13 01:02:39 +0000431 Arg = Arg->IgnoreParenCasts();
Anders Carlsson71993dd2007-08-17 05:31:46 +0000432 StringLiteral *Literal = dyn_cast<StringLiteral>(Arg);
433
434 if (!Literal || Literal->isWide()) {
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000435 Diag(Arg->getLocStart(), diag::err_cfstring_literal_not_string_constant)
436 << Arg->getSourceRange();
Anders Carlsson9cdc4d32007-08-17 15:44:17 +0000437 return true;
Anders Carlsson71993dd2007-08-17 05:31:46 +0000438 }
439
440 const char *Data = Literal->getStrData();
441 unsigned Length = Literal->getByteLength();
442
443 for (unsigned i = 0; i < Length; ++i) {
Anders Carlsson71993dd2007-08-17 05:31:46 +0000444 if (!Data[i]) {
Chris Lattner60800082009-02-18 17:49:48 +0000445 Diag(getLocationOfStringLiteralByte(Literal, i),
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000446 diag::warn_cfstring_literal_contains_nul_character)
447 << Arg->getSourceRange();
Anders Carlsson71993dd2007-08-17 05:31:46 +0000448 break;
449 }
450 }
451
Anders Carlsson9cdc4d32007-08-17 15:44:17 +0000452 return false;
Chris Lattner59907c42007-08-10 20:18:51 +0000453}
454
Chris Lattnerc27c6652007-12-20 00:05:45 +0000455/// SemaBuiltinVAStart - Check the arguments to __builtin_va_start for validity.
456/// Emit an error and return true on failure, return false on success.
Chris Lattner925e60d2007-12-28 05:29:59 +0000457bool Sema::SemaBuiltinVAStart(CallExpr *TheCall) {
458 Expr *Fn = TheCall->getCallee();
459 if (TheCall->getNumArgs() > 2) {
Chris Lattner2c21a072008-11-21 18:44:24 +0000460 Diag(TheCall->getArg(2)->getLocStart(),
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000461 diag::err_typecheck_call_too_many_args)
Chris Lattner2c21a072008-11-21 18:44:24 +0000462 << 0 /*function call*/ << Fn->getSourceRange()
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000463 << SourceRange(TheCall->getArg(2)->getLocStart(),
464 (*(TheCall->arg_end()-1))->getLocEnd());
Chris Lattner30ce3442007-12-19 23:59:04 +0000465 return true;
466 }
Eli Friedman56f20ae2008-12-15 22:05:35 +0000467
468 if (TheCall->getNumArgs() < 2) {
469 return Diag(TheCall->getLocEnd(), diag::err_typecheck_call_too_few_args)
470 << 0 /*function call*/;
471 }
472
Chris Lattnerc27c6652007-12-20 00:05:45 +0000473 // Determine whether the current function is variadic or not.
474 bool isVariadic;
Steve Naroffcd9c5142009-04-15 19:33:47 +0000475 if (CurBlock)
476 isVariadic = CurBlock->isVariadic;
477 else if (getCurFunctionDecl()) {
Douglas Gregor72564e72009-02-26 23:50:07 +0000478 if (FunctionProtoType* FTP =
479 dyn_cast<FunctionProtoType>(getCurFunctionDecl()->getType()))
Eli Friedman56f20ae2008-12-15 22:05:35 +0000480 isVariadic = FTP->isVariadic();
481 else
482 isVariadic = false;
483 } else {
Argyrios Kyrtzidis53d0ea52008-06-28 06:07:14 +0000484 isVariadic = getCurMethodDecl()->isVariadic();
Eli Friedman56f20ae2008-12-15 22:05:35 +0000485 }
Chris Lattner30ce3442007-12-19 23:59:04 +0000486
Chris Lattnerc27c6652007-12-20 00:05:45 +0000487 if (!isVariadic) {
Chris Lattner30ce3442007-12-19 23:59:04 +0000488 Diag(Fn->getLocStart(), diag::err_va_start_used_in_non_variadic_function);
489 return true;
490 }
491
492 // Verify that the second argument to the builtin is the last argument of the
493 // current function or method.
494 bool SecondArgIsLastNamedArgument = false;
Anders Carlssone2c14102008-02-13 01:22:59 +0000495 const Expr *Arg = TheCall->getArg(1)->IgnoreParenCasts();
Anders Carlsson88cf2262008-02-11 04:20:54 +0000496
497 if (const DeclRefExpr *DR = dyn_cast<DeclRefExpr>(Arg)) {
498 if (const ParmVarDecl *PV = dyn_cast<ParmVarDecl>(DR->getDecl())) {
Chris Lattner30ce3442007-12-19 23:59:04 +0000499 // FIXME: This isn't correct for methods (results in bogus warning).
500 // Get the last formal in the current function.
Anders Carlsson88cf2262008-02-11 04:20:54 +0000501 const ParmVarDecl *LastArg;
Steve Naroffcd9c5142009-04-15 19:33:47 +0000502 if (CurBlock)
503 LastArg = *(CurBlock->TheDecl->param_end()-1);
504 else if (FunctionDecl *FD = getCurFunctionDecl())
Chris Lattner371f2582008-12-04 23:50:19 +0000505 LastArg = *(FD->param_end()-1);
Chris Lattner30ce3442007-12-19 23:59:04 +0000506 else
Argyrios Kyrtzidis53d0ea52008-06-28 06:07:14 +0000507 LastArg = *(getCurMethodDecl()->param_end()-1);
Chris Lattner30ce3442007-12-19 23:59:04 +0000508 SecondArgIsLastNamedArgument = PV == LastArg;
509 }
510 }
511
512 if (!SecondArgIsLastNamedArgument)
Chris Lattner925e60d2007-12-28 05:29:59 +0000513 Diag(TheCall->getArg(1)->getLocStart(),
Chris Lattner30ce3442007-12-19 23:59:04 +0000514 diag::warn_second_parameter_of_va_start_not_last_named_argument);
515 return false;
Eli Friedman6cfda232008-05-20 08:23:37 +0000516}
Chris Lattner30ce3442007-12-19 23:59:04 +0000517
Chris Lattner1b9a0792007-12-20 00:26:33 +0000518/// SemaBuiltinUnorderedCompare - Handle functions like __builtin_isgreater and
519/// friends. This is declared to take (...), so we have to check everything.
Chris Lattner925e60d2007-12-28 05:29:59 +0000520bool Sema::SemaBuiltinUnorderedCompare(CallExpr *TheCall) {
521 if (TheCall->getNumArgs() < 2)
Chris Lattner2c21a072008-11-21 18:44:24 +0000522 return Diag(TheCall->getLocEnd(), diag::err_typecheck_call_too_few_args)
523 << 0 /*function call*/;
Chris Lattner925e60d2007-12-28 05:29:59 +0000524 if (TheCall->getNumArgs() > 2)
525 return Diag(TheCall->getArg(2)->getLocStart(),
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000526 diag::err_typecheck_call_too_many_args)
Chris Lattner2c21a072008-11-21 18:44:24 +0000527 << 0 /*function call*/
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000528 << SourceRange(TheCall->getArg(2)->getLocStart(),
529 (*(TheCall->arg_end()-1))->getLocEnd());
Chris Lattner1b9a0792007-12-20 00:26:33 +0000530
Chris Lattner925e60d2007-12-28 05:29:59 +0000531 Expr *OrigArg0 = TheCall->getArg(0);
532 Expr *OrigArg1 = TheCall->getArg(1);
Douglas Gregorcde01732009-05-19 22:10:17 +0000533
Chris Lattner1b9a0792007-12-20 00:26:33 +0000534 // Do standard promotions between the two arguments, returning their common
535 // type.
Chris Lattner925e60d2007-12-28 05:29:59 +0000536 QualType Res = UsualArithmeticConversions(OrigArg0, OrigArg1, false);
Daniel Dunbar403bc2b2009-02-19 19:28:43 +0000537
538 // Make sure any conversions are pushed back into the call; this is
539 // type safe since unordered compare builtins are declared as "_Bool
540 // foo(...)".
541 TheCall->setArg(0, OrigArg0);
542 TheCall->setArg(1, OrigArg1);
Chris Lattner1b9a0792007-12-20 00:26:33 +0000543
Douglas Gregorcde01732009-05-19 22:10:17 +0000544 if (OrigArg0->isTypeDependent() || OrigArg1->isTypeDependent())
545 return false;
546
Chris Lattner1b9a0792007-12-20 00:26:33 +0000547 // If the common type isn't a real floating type, then the arguments were
548 // invalid for this operation.
549 if (!Res->isRealFloatingType())
Chris Lattner925e60d2007-12-28 05:29:59 +0000550 return Diag(OrigArg0->getLocStart(),
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000551 diag::err_typecheck_call_invalid_ordered_compare)
Chris Lattnerd1625842008-11-24 06:25:27 +0000552 << OrigArg0->getType() << OrigArg1->getType()
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000553 << SourceRange(OrigArg0->getLocStart(), OrigArg1->getLocEnd());
Chris Lattner1b9a0792007-12-20 00:26:33 +0000554
555 return false;
556}
557
Eli Friedman6cfda232008-05-20 08:23:37 +0000558bool Sema::SemaBuiltinStackAddress(CallExpr *TheCall) {
559 // The signature for these builtins is exact; the only thing we need
560 // to check is that the argument is a constant.
561 SourceLocation Loc;
Douglas Gregorcde01732009-05-19 22:10:17 +0000562 if (!TheCall->getArg(0)->isTypeDependent() &&
563 !TheCall->getArg(0)->isValueDependent() &&
564 !TheCall->getArg(0)->isIntegerConstantExpr(Context, &Loc))
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000565 return Diag(Loc, diag::err_stack_const_level) << TheCall->getSourceRange();
Chris Lattnerd1a0b6d2008-08-10 02:05:13 +0000566
Eli Friedman6cfda232008-05-20 08:23:37 +0000567 return false;
568}
569
Eli Friedmand38617c2008-05-14 19:38:39 +0000570/// SemaBuiltinShuffleVector - Handle __builtin_shufflevector.
571// This is declared to take (...), so we have to check everything.
Sebastian Redl0eb23302009-01-19 00:08:26 +0000572Action::OwningExprResult Sema::SemaBuiltinShuffleVector(CallExpr *TheCall) {
Eli Friedmand38617c2008-05-14 19:38:39 +0000573 if (TheCall->getNumArgs() < 3)
Sebastian Redl0eb23302009-01-19 00:08:26 +0000574 return ExprError(Diag(TheCall->getLocEnd(),
575 diag::err_typecheck_call_too_few_args)
576 << 0 /*function call*/ << TheCall->getSourceRange());
Eli Friedmand38617c2008-05-14 19:38:39 +0000577
Douglas Gregorcde01732009-05-19 22:10:17 +0000578 unsigned numElements = std::numeric_limits<unsigned>::max();
579 if (!TheCall->getArg(0)->isTypeDependent() &&
580 !TheCall->getArg(1)->isTypeDependent()) {
581 QualType FAType = TheCall->getArg(0)->getType();
582 QualType SAType = TheCall->getArg(1)->getType();
583
584 if (!FAType->isVectorType() || !SAType->isVectorType()) {
585 Diag(TheCall->getLocStart(), diag::err_shufflevector_non_vector)
586 << SourceRange(TheCall->getArg(0)->getLocStart(),
587 TheCall->getArg(1)->getLocEnd());
588 return ExprError();
589 }
590
591 if (Context.getCanonicalType(FAType).getUnqualifiedType() !=
592 Context.getCanonicalType(SAType).getUnqualifiedType()) {
593 Diag(TheCall->getLocStart(), diag::err_shufflevector_incompatible_vector)
594 << SourceRange(TheCall->getArg(0)->getLocStart(),
595 TheCall->getArg(1)->getLocEnd());
596 return ExprError();
597 }
Eli Friedmand38617c2008-05-14 19:38:39 +0000598
Douglas Gregorcde01732009-05-19 22:10:17 +0000599 numElements = FAType->getAsVectorType()->getNumElements();
600 if (TheCall->getNumArgs() != numElements+2) {
601 if (TheCall->getNumArgs() < numElements+2)
602 return ExprError(Diag(TheCall->getLocEnd(),
603 diag::err_typecheck_call_too_few_args)
604 << 0 /*function call*/ << TheCall->getSourceRange());
Sebastian Redl0eb23302009-01-19 00:08:26 +0000605 return ExprError(Diag(TheCall->getLocEnd(),
Douglas Gregorcde01732009-05-19 22:10:17 +0000606 diag::err_typecheck_call_too_many_args)
607 << 0 /*function call*/ << TheCall->getSourceRange());
608 }
Eli Friedmand38617c2008-05-14 19:38:39 +0000609 }
610
611 for (unsigned i = 2; i < TheCall->getNumArgs(); i++) {
Douglas Gregorcde01732009-05-19 22:10:17 +0000612 if (TheCall->getArg(i)->isTypeDependent() ||
613 TheCall->getArg(i)->isValueDependent())
614 continue;
615
Eli Friedmand38617c2008-05-14 19:38:39 +0000616 llvm::APSInt Result(32);
Chris Lattnerd1a0b6d2008-08-10 02:05:13 +0000617 if (!TheCall->getArg(i)->isIntegerConstantExpr(Result, Context))
Sebastian Redl0eb23302009-01-19 00:08:26 +0000618 return ExprError(Diag(TheCall->getLocStart(),
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000619 diag::err_shufflevector_nonconstant_argument)
Sebastian Redl0eb23302009-01-19 00:08:26 +0000620 << TheCall->getArg(i)->getSourceRange());
621
Chris Lattnerd1a0b6d2008-08-10 02:05:13 +0000622 if (Result.getActiveBits() > 64 || Result.getZExtValue() >= numElements*2)
Sebastian Redl0eb23302009-01-19 00:08:26 +0000623 return ExprError(Diag(TheCall->getLocStart(),
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000624 diag::err_shufflevector_argument_too_large)
Sebastian Redl0eb23302009-01-19 00:08:26 +0000625 << TheCall->getArg(i)->getSourceRange());
Eli Friedmand38617c2008-05-14 19:38:39 +0000626 }
627
628 llvm::SmallVector<Expr*, 32> exprs;
629
Chris Lattnerd1a0b6d2008-08-10 02:05:13 +0000630 for (unsigned i = 0, e = TheCall->getNumArgs(); i != e; i++) {
Eli Friedmand38617c2008-05-14 19:38:39 +0000631 exprs.push_back(TheCall->getArg(i));
632 TheCall->setArg(i, 0);
633 }
634
Nate Begemana88dc302009-08-12 02:10:25 +0000635 return Owned(new (Context) ShuffleVectorExpr(Context, exprs.begin(),
636 exprs.size(), exprs[0]->getType(),
Ted Kremenek8189cde2009-02-07 01:47:29 +0000637 TheCall->getCallee()->getLocStart(),
638 TheCall->getRParenLoc()));
Eli Friedmand38617c2008-05-14 19:38:39 +0000639}
Chris Lattner30ce3442007-12-19 23:59:04 +0000640
Daniel Dunbar4493f792008-07-21 22:59:13 +0000641/// SemaBuiltinPrefetch - Handle __builtin_prefetch.
642// This is declared to take (const void*, ...) and can take two
643// optional constant int args.
644bool Sema::SemaBuiltinPrefetch(CallExpr *TheCall) {
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000645 unsigned NumArgs = TheCall->getNumArgs();
Daniel Dunbar4493f792008-07-21 22:59:13 +0000646
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000647 if (NumArgs > 3)
648 return Diag(TheCall->getLocEnd(), diag::err_typecheck_call_too_many_args)
Chris Lattner2c21a072008-11-21 18:44:24 +0000649 << 0 /*function call*/ << TheCall->getSourceRange();
Daniel Dunbar4493f792008-07-21 22:59:13 +0000650
651 // Argument 0 is checked for us and the remaining arguments must be
652 // constant integers.
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000653 for (unsigned i = 1; i != NumArgs; ++i) {
Daniel Dunbar4493f792008-07-21 22:59:13 +0000654 Expr *Arg = TheCall->getArg(i);
Douglas Gregorcde01732009-05-19 22:10:17 +0000655 if (Arg->isTypeDependent())
656 continue;
657
Daniel Dunbar4493f792008-07-21 22:59:13 +0000658 QualType RWType = Arg->getType();
659
660 const BuiltinType *BT = RWType->getAsBuiltinType();
Daniel Dunbard5f8a4f2008-09-03 21:13:56 +0000661 llvm::APSInt Result;
Douglas Gregorcde01732009-05-19 22:10:17 +0000662 if (!BT || BT->getKind() != BuiltinType::Int)
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000663 return Diag(TheCall->getLocStart(), diag::err_prefetch_invalid_argument)
664 << SourceRange(Arg->getLocStart(), Arg->getLocEnd());
Douglas Gregorcde01732009-05-19 22:10:17 +0000665
666 if (Arg->isValueDependent())
667 continue;
668
669 if (!Arg->isIntegerConstantExpr(Result, Context))
670 return Diag(TheCall->getLocStart(), diag::err_prefetch_invalid_argument)
671 << SourceRange(Arg->getLocStart(), Arg->getLocEnd());
Daniel Dunbar4493f792008-07-21 22:59:13 +0000672
673 // FIXME: gcc issues a warning and rewrites these to 0. These
674 // seems especially odd for the third argument since the default
675 // is 3.
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000676 if (i == 1) {
Daniel Dunbar4493f792008-07-21 22:59:13 +0000677 if (Result.getSExtValue() < 0 || Result.getSExtValue() > 1)
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000678 return Diag(TheCall->getLocStart(), diag::err_argument_invalid_range)
679 << "0" << "1" << SourceRange(Arg->getLocStart(), Arg->getLocEnd());
Daniel Dunbar4493f792008-07-21 22:59:13 +0000680 } else {
681 if (Result.getSExtValue() < 0 || Result.getSExtValue() > 3)
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000682 return Diag(TheCall->getLocStart(), diag::err_argument_invalid_range)
683 << "0" << "3" << SourceRange(Arg->getLocStart(), Arg->getLocEnd());
Daniel Dunbar4493f792008-07-21 22:59:13 +0000684 }
685 }
686
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000687 return false;
Daniel Dunbar4493f792008-07-21 22:59:13 +0000688}
689
Daniel Dunbard5f8a4f2008-09-03 21:13:56 +0000690/// SemaBuiltinObjectSize - Handle __builtin_object_size(void *ptr,
691/// int type). This simply type checks that type is one of the defined
692/// constants (0-3).
693bool Sema::SemaBuiltinObjectSize(CallExpr *TheCall) {
694 Expr *Arg = TheCall->getArg(1);
Douglas Gregorcde01732009-05-19 22:10:17 +0000695 if (Arg->isTypeDependent())
696 return false;
697
Daniel Dunbard5f8a4f2008-09-03 21:13:56 +0000698 QualType ArgType = Arg->getType();
699 const BuiltinType *BT = ArgType->getAsBuiltinType();
700 llvm::APSInt Result(32);
Douglas Gregorcde01732009-05-19 22:10:17 +0000701 if (!BT || BT->getKind() != BuiltinType::Int)
702 return Diag(TheCall->getLocStart(), diag::err_object_size_invalid_argument)
703 << SourceRange(Arg->getLocStart(), Arg->getLocEnd());
704
705 if (Arg->isValueDependent())
706 return false;
707
708 if (!Arg->isIntegerConstantExpr(Result, Context)) {
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000709 return Diag(TheCall->getLocStart(), diag::err_object_size_invalid_argument)
710 << SourceRange(Arg->getLocStart(), Arg->getLocEnd());
Daniel Dunbard5f8a4f2008-09-03 21:13:56 +0000711 }
712
713 if (Result.getSExtValue() < 0 || Result.getSExtValue() > 3) {
Chris Lattnerfa25bbb2008-11-19 05:08:23 +0000714 return Diag(TheCall->getLocStart(), diag::err_argument_invalid_range)
715 << "0" << "3" << SourceRange(Arg->getLocStart(), Arg->getLocEnd());
Daniel Dunbard5f8a4f2008-09-03 21:13:56 +0000716 }
717
718 return false;
719}
720
Eli Friedman586d6a82009-05-03 06:04:26 +0000721/// SemaBuiltinLongjmp - Handle __builtin_longjmp(void *env[5], int val).
Eli Friedmand875fed2009-05-03 04:46:36 +0000722/// This checks that val is a constant 1.
723bool Sema::SemaBuiltinLongjmp(CallExpr *TheCall) {
724 Expr *Arg = TheCall->getArg(1);
Douglas Gregorcde01732009-05-19 22:10:17 +0000725 if (Arg->isTypeDependent() || Arg->isValueDependent())
726 return false;
727
Eli Friedmand875fed2009-05-03 04:46:36 +0000728 llvm::APSInt Result(32);
729 if (!Arg->isIntegerConstantExpr(Result, Context) || Result != 1)
730 return Diag(TheCall->getLocStart(), diag::err_builtin_longjmp_invalid_val)
731 << SourceRange(Arg->getLocStart(), Arg->getLocEnd());
732
733 return false;
734}
735
Ted Kremenekd30ef872009-01-12 23:09:09 +0000736// Handle i > 1 ? "x" : "y", recursivelly
Ted Kremenek082d9362009-03-20 21:35:28 +0000737bool Sema::SemaCheckStringLiteral(const Expr *E, const CallExpr *TheCall,
738 bool HasVAListArg,
Douglas Gregor3c385e52009-02-14 18:57:46 +0000739 unsigned format_idx, unsigned firstDataArg) {
Douglas Gregorcde01732009-05-19 22:10:17 +0000740 if (E->isTypeDependent() || E->isValueDependent())
741 return false;
Ted Kremenekd30ef872009-01-12 23:09:09 +0000742
743 switch (E->getStmtClass()) {
744 case Stmt::ConditionalOperatorClass: {
Ted Kremenek082d9362009-03-20 21:35:28 +0000745 const ConditionalOperator *C = cast<ConditionalOperator>(E);
Ted Kremenekd30ef872009-01-12 23:09:09 +0000746 return SemaCheckStringLiteral(C->getLHS(), TheCall,
Douglas Gregor3c385e52009-02-14 18:57:46 +0000747 HasVAListArg, format_idx, firstDataArg)
Ted Kremenekd30ef872009-01-12 23:09:09 +0000748 && SemaCheckStringLiteral(C->getRHS(), TheCall,
Douglas Gregor3c385e52009-02-14 18:57:46 +0000749 HasVAListArg, format_idx, firstDataArg);
Ted Kremenekd30ef872009-01-12 23:09:09 +0000750 }
751
752 case Stmt::ImplicitCastExprClass: {
Ted Kremenek082d9362009-03-20 21:35:28 +0000753 const ImplicitCastExpr *Expr = cast<ImplicitCastExpr>(E);
Ted Kremenekd30ef872009-01-12 23:09:09 +0000754 return SemaCheckStringLiteral(Expr->getSubExpr(), TheCall, HasVAListArg,
Douglas Gregor3c385e52009-02-14 18:57:46 +0000755 format_idx, firstDataArg);
Ted Kremenekd30ef872009-01-12 23:09:09 +0000756 }
757
758 case Stmt::ParenExprClass: {
Ted Kremenek082d9362009-03-20 21:35:28 +0000759 const ParenExpr *Expr = cast<ParenExpr>(E);
Ted Kremenekd30ef872009-01-12 23:09:09 +0000760 return SemaCheckStringLiteral(Expr->getSubExpr(), TheCall, HasVAListArg,
Douglas Gregor3c385e52009-02-14 18:57:46 +0000761 format_idx, firstDataArg);
Ted Kremenekd30ef872009-01-12 23:09:09 +0000762 }
Ted Kremenek082d9362009-03-20 21:35:28 +0000763
764 case Stmt::DeclRefExprClass: {
765 const DeclRefExpr *DR = cast<DeclRefExpr>(E);
766
767 // As an exception, do not flag errors for variables binding to
768 // const string literals.
769 if (const VarDecl *VD = dyn_cast<VarDecl>(DR->getDecl())) {
770 bool isConstant = false;
771 QualType T = DR->getType();
Ted Kremenekd30ef872009-01-12 23:09:09 +0000772
Ted Kremenek082d9362009-03-20 21:35:28 +0000773 if (const ArrayType *AT = Context.getAsArrayType(T)) {
774 isConstant = AT->getElementType().isConstant(Context);
Mike Stumpac5fc7c2009-08-04 21:02:39 +0000775 } else if (const PointerType *PT = T->getAs<PointerType>()) {
Ted Kremenek082d9362009-03-20 21:35:28 +0000776 isConstant = T.isConstant(Context) &&
777 PT->getPointeeType().isConstant(Context);
778 }
779
780 if (isConstant) {
781 const VarDecl *Def = 0;
782 if (const Expr *Init = VD->getDefinition(Def))
783 return SemaCheckStringLiteral(Init, TheCall,
784 HasVAListArg, format_idx, firstDataArg);
785 }
Anders Carlssond966a552009-06-28 19:55:58 +0000786
787 // For vprintf* functions (i.e., HasVAListArg==true), we add a
788 // special check to see if the format string is a function parameter
789 // of the function calling the printf function. If the function
790 // has an attribute indicating it is a printf-like function, then we
791 // should suppress warnings concerning non-literals being used in a call
792 // to a vprintf function. For example:
793 //
794 // void
795 // logmessage(char const *fmt __attribute__ (format (printf, 1, 2)), ...){
796 // va_list ap;
797 // va_start(ap, fmt);
798 // vprintf(fmt, ap); // Do NOT emit a warning about "fmt".
799 // ...
800 //
801 //
802 // FIXME: We don't have full attribute support yet, so just check to see
803 // if the argument is a DeclRefExpr that references a parameter. We'll
804 // add proper support for checking the attribute later.
805 if (HasVAListArg)
806 if (isa<ParmVarDecl>(VD))
807 return true;
Ted Kremenek082d9362009-03-20 21:35:28 +0000808 }
809
810 return false;
811 }
Ted Kremenekd30ef872009-01-12 23:09:09 +0000812
Anders Carlsson8f031b32009-06-27 04:05:33 +0000813 case Stmt::CallExprClass: {
814 const CallExpr *CE = cast<CallExpr>(E);
815 if (const ImplicitCastExpr *ICE
816 = dyn_cast<ImplicitCastExpr>(CE->getCallee())) {
817 if (const DeclRefExpr *DRE = dyn_cast<DeclRefExpr>(ICE->getSubExpr())) {
818 if (const FunctionDecl *FD = dyn_cast<FunctionDecl>(DRE->getDecl())) {
Argyrios Kyrtzidis40b598e2009-06-30 02:34:44 +0000819 if (const FormatArgAttr *FA = FD->getAttr<FormatArgAttr>()) {
Anders Carlsson8f031b32009-06-27 04:05:33 +0000820 unsigned ArgIndex = FA->getFormatIdx();
821 const Expr *Arg = CE->getArg(ArgIndex - 1);
822
823 return SemaCheckStringLiteral(Arg, TheCall, HasVAListArg,
824 format_idx, firstDataArg);
825 }
826 }
827 }
828 }
829
830 return false;
831 }
Ted Kremenek082d9362009-03-20 21:35:28 +0000832 case Stmt::ObjCStringLiteralClass:
833 case Stmt::StringLiteralClass: {
834 const StringLiteral *StrE = NULL;
835
836 if (const ObjCStringLiteral *ObjCFExpr = dyn_cast<ObjCStringLiteral>(E))
Ted Kremenekd30ef872009-01-12 23:09:09 +0000837 StrE = ObjCFExpr->getString();
838 else
Ted Kremenek082d9362009-03-20 21:35:28 +0000839 StrE = cast<StringLiteral>(E);
840
Ted Kremenekd30ef872009-01-12 23:09:09 +0000841 if (StrE) {
Douglas Gregor3c385e52009-02-14 18:57:46 +0000842 CheckPrintfString(StrE, E, TheCall, HasVAListArg, format_idx,
843 firstDataArg);
Ted Kremenekd30ef872009-01-12 23:09:09 +0000844 return true;
845 }
846
847 return false;
848 }
Ted Kremenek082d9362009-03-20 21:35:28 +0000849
850 default:
851 return false;
Ted Kremenekd30ef872009-01-12 23:09:09 +0000852 }
853}
854
Fariborz Jahaniane898f8a2009-05-21 18:48:51 +0000855void
856Sema::CheckNonNullArguments(const NonNullAttr *NonNull, const CallExpr *TheCall)
857{
858 for (NonNullAttr::iterator i = NonNull->begin(), e = NonNull->end();
859 i != e; ++i) {
Chris Lattner12b97ff2009-05-25 18:23:36 +0000860 const Expr *ArgExpr = TheCall->getArg(*i);
Fariborz Jahaniane898f8a2009-05-21 18:48:51 +0000861 if (ArgExpr->isNullPointerConstant(Context))
Chris Lattner12b97ff2009-05-25 18:23:36 +0000862 Diag(TheCall->getCallee()->getLocStart(), diag::warn_null_arg)
863 << ArgExpr->getSourceRange();
Fariborz Jahaniane898f8a2009-05-21 18:48:51 +0000864 }
865}
Ted Kremenekd30ef872009-01-12 23:09:09 +0000866
Chris Lattner59907c42007-08-10 20:18:51 +0000867/// CheckPrintfArguments - Check calls to printf (and similar functions) for
Ted Kremenek71895b92007-08-14 17:39:48 +0000868/// correct use of format strings.
869///
870/// HasVAListArg - A predicate indicating whether the printf-like
871/// function is passed an explicit va_arg argument (e.g., vprintf)
872///
873/// format_idx - The index into Args for the format string.
874///
875/// Improper format strings to functions in the printf family can be
876/// the source of bizarre bugs and very serious security holes. A
877/// good source of information is available in the following paper
878/// (which includes additional references):
Chris Lattner59907c42007-08-10 20:18:51 +0000879///
880/// FormatGuard: Automatic Protection From printf Format String
881/// Vulnerabilities, Proceedings of the 10th USENIX Security Symposium, 2001.
Ted Kremenek71895b92007-08-14 17:39:48 +0000882///
883/// Functionality implemented:
884///
885/// We can statically check the following properties for string
886/// literal format strings for non v.*printf functions (where the
887/// arguments are passed directly):
888//
889/// (1) Are the number of format conversions equal to the number of
890/// data arguments?
891///
892/// (2) Does each format conversion correctly match the type of the
893/// corresponding data argument? (TODO)
894///
895/// Moreover, for all printf functions we can:
896///
897/// (3) Check for a missing format string (when not caught by type checking).
898///
899/// (4) Check for no-operation flags; e.g. using "#" with format
900/// conversion 'c' (TODO)
901///
902/// (5) Check the use of '%n', a major source of security holes.
903///
904/// (6) Check for malformed format conversions that don't specify anything.
905///
906/// (7) Check for empty format strings. e.g: printf("");
907///
908/// (8) Check that the format string is a wide literal.
909///
Ted Kremenek6d439592008-03-03 16:50:00 +0000910/// (9) Also check the arguments of functions with the __format__ attribute.
911/// (TODO).
912///
Ted Kremenek71895b92007-08-14 17:39:48 +0000913/// All of these checks can be done by parsing the format string.
914///
915/// For now, we ONLY do (1), (3), (5), (6), (7), and (8).
Chris Lattner59907c42007-08-10 20:18:51 +0000916void
Ted Kremenek082d9362009-03-20 21:35:28 +0000917Sema::CheckPrintfArguments(const CallExpr *TheCall, bool HasVAListArg,
Douglas Gregor3c385e52009-02-14 18:57:46 +0000918 unsigned format_idx, unsigned firstDataArg) {
Ted Kremenek082d9362009-03-20 21:35:28 +0000919 const Expr *Fn = TheCall->getCallee();
Chris Lattner925e60d2007-12-28 05:29:59 +0000920
Ted Kremenek71895b92007-08-14 17:39:48 +0000921 // CHECK: printf-like function is called with no format string.
Chris Lattner925e60d2007-12-28 05:29:59 +0000922 if (format_idx >= TheCall->getNumArgs()) {
Chris Lattnerdcd5ef12008-11-19 05:27:50 +0000923 Diag(TheCall->getRParenLoc(), diag::warn_printf_missing_format_string)
924 << Fn->getSourceRange();
Ted Kremenek71895b92007-08-14 17:39:48 +0000925 return;
926 }
927
Ted Kremenek082d9362009-03-20 21:35:28 +0000928 const Expr *OrigFormatExpr = TheCall->getArg(format_idx)->IgnoreParenCasts();
Chris Lattner459e8482007-08-25 05:36:18 +0000929
Chris Lattner59907c42007-08-10 20:18:51 +0000930 // CHECK: format string is not a string literal.
931 //
Ted Kremenek71895b92007-08-14 17:39:48 +0000932 // Dynamically generated format strings are difficult to
933 // automatically vet at compile time. Requiring that format strings
934 // are string literals: (1) permits the checking of format strings by
935 // the compiler and thereby (2) can practically remove the source of
936 // many format string exploits.
Ted Kremenek7ff22b22008-06-16 18:00:42 +0000937
938 // Format string can be either ObjC string (e.g. @"%d") or
939 // C string (e.g. "%d")
940 // ObjC string uses the same format specifiers as C string, so we can use
941 // the same format string checking logic for both ObjC and C strings.
Chris Lattner1cd3e1f2009-04-29 04:49:34 +0000942 if (SemaCheckStringLiteral(OrigFormatExpr, TheCall, HasVAListArg, format_idx,
943 firstDataArg))
944 return; // Literal format string found, check done!
Ted Kremenek7ff22b22008-06-16 18:00:42 +0000945
Chris Lattner655f1412009-04-29 04:59:47 +0000946 // If there are no arguments specified, warn with -Wformat-security, otherwise
947 // warn only with -Wformat-nonliteral.
948 if (TheCall->getNumArgs() == format_idx+1)
949 Diag(TheCall->getArg(format_idx)->getLocStart(),
950 diag::warn_printf_nonliteral_noargs)
951 << OrigFormatExpr->getSourceRange();
952 else
953 Diag(TheCall->getArg(format_idx)->getLocStart(),
954 diag::warn_printf_nonliteral)
955 << OrigFormatExpr->getSourceRange();
Ted Kremenekd30ef872009-01-12 23:09:09 +0000956}
Ted Kremenek71895b92007-08-14 17:39:48 +0000957
Ted Kremenek082d9362009-03-20 21:35:28 +0000958void Sema::CheckPrintfString(const StringLiteral *FExpr,
959 const Expr *OrigFormatExpr,
960 const CallExpr *TheCall, bool HasVAListArg,
961 unsigned format_idx, unsigned firstDataArg) {
Ted Kremenekd30ef872009-01-12 23:09:09 +0000962
Ted Kremenek082d9362009-03-20 21:35:28 +0000963 const ObjCStringLiteral *ObjCFExpr =
964 dyn_cast<ObjCStringLiteral>(OrigFormatExpr);
965
Ted Kremenek71895b92007-08-14 17:39:48 +0000966 // CHECK: is the format string a wide literal?
967 if (FExpr->isWide()) {
Chris Lattner925e60d2007-12-28 05:29:59 +0000968 Diag(FExpr->getLocStart(),
Chris Lattnerdcd5ef12008-11-19 05:27:50 +0000969 diag::warn_printf_format_string_is_wide_literal)
970 << OrigFormatExpr->getSourceRange();
Ted Kremenek71895b92007-08-14 17:39:48 +0000971 return;
972 }
973
974 // Str - The format string. NOTE: this is NOT null-terminated!
Chris Lattnerb9fc8562009-04-29 04:12:34 +0000975 const char *Str = FExpr->getStrData();
Ted Kremenek71895b92007-08-14 17:39:48 +0000976
977 // CHECK: empty format string?
Chris Lattnerb9fc8562009-04-29 04:12:34 +0000978 unsigned StrLen = FExpr->getByteLength();
Ted Kremenek71895b92007-08-14 17:39:48 +0000979
980 if (StrLen == 0) {
Chris Lattnerdcd5ef12008-11-19 05:27:50 +0000981 Diag(FExpr->getLocStart(), diag::warn_printf_empty_format_string)
982 << OrigFormatExpr->getSourceRange();
Ted Kremenek71895b92007-08-14 17:39:48 +0000983 return;
984 }
985
986 // We process the format string using a binary state machine. The
987 // current state is stored in CurrentState.
988 enum {
989 state_OrdChr,
990 state_Conversion
991 } CurrentState = state_OrdChr;
992
993 // numConversions - The number of conversions seen so far. This is
994 // incremented as we traverse the format string.
995 unsigned numConversions = 0;
996
997 // numDataArgs - The number of data arguments after the format
998 // string. This can only be determined for non vprintf-like
999 // functions. For those functions, this value is 1 (the sole
1000 // va_arg argument).
Douglas Gregor3c385e52009-02-14 18:57:46 +00001001 unsigned numDataArgs = TheCall->getNumArgs()-firstDataArg;
Ted Kremenek71895b92007-08-14 17:39:48 +00001002
1003 // Inspect the format string.
1004 unsigned StrIdx = 0;
1005
1006 // LastConversionIdx - Index within the format string where we last saw
1007 // a '%' character that starts a new format conversion.
1008 unsigned LastConversionIdx = 0;
1009
Chris Lattner925e60d2007-12-28 05:29:59 +00001010 for (; StrIdx < StrLen; ++StrIdx) {
Chris Lattner998568f2007-12-28 05:38:24 +00001011
Ted Kremenek71895b92007-08-14 17:39:48 +00001012 // Is the number of detected conversion conversions greater than
1013 // the number of matching data arguments? If so, stop.
1014 if (!HasVAListArg && numConversions > numDataArgs) break;
1015
1016 // Handle "\0"
Chris Lattner925e60d2007-12-28 05:29:59 +00001017 if (Str[StrIdx] == '\0') {
Ted Kremenek71895b92007-08-14 17:39:48 +00001018 // The string returned by getStrData() is not null-terminated,
1019 // so the presence of a null character is likely an error.
Chris Lattner60800082009-02-18 17:49:48 +00001020 Diag(getLocationOfStringLiteralByte(FExpr, StrIdx),
Chris Lattnerdcd5ef12008-11-19 05:27:50 +00001021 diag::warn_printf_format_string_contains_null_char)
1022 << OrigFormatExpr->getSourceRange();
Ted Kremenek71895b92007-08-14 17:39:48 +00001023 return;
1024 }
1025
1026 // Ordinary characters (not processing a format conversion).
1027 if (CurrentState == state_OrdChr) {
1028 if (Str[StrIdx] == '%') {
1029 CurrentState = state_Conversion;
1030 LastConversionIdx = StrIdx;
1031 }
1032 continue;
1033 }
1034
1035 // Seen '%'. Now processing a format conversion.
1036 switch (Str[StrIdx]) {
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001037 // Handle dynamic precision or width specifier.
1038 case '*': {
1039 ++numConversions;
1040
Ted Kremenek42ae3e82009-05-13 16:06:05 +00001041 if (!HasVAListArg) {
1042 if (numConversions > numDataArgs) {
1043 SourceLocation Loc = getLocationOfStringLiteralByte(FExpr, StrIdx);
Ted Kremenek580b6642007-10-12 20:51:52 +00001044
Ted Kremenek42ae3e82009-05-13 16:06:05 +00001045 if (Str[StrIdx-1] == '.')
1046 Diag(Loc, diag::warn_printf_asterisk_precision_missing_arg)
1047 << OrigFormatExpr->getSourceRange();
1048 else
1049 Diag(Loc, diag::warn_printf_asterisk_width_missing_arg)
1050 << OrigFormatExpr->getSourceRange();
1051
1052 // Don't do any more checking. We'll just emit spurious errors.
1053 return;
1054 }
1055
1056 // Perform type checking on width/precision specifier.
1057 const Expr *E = TheCall->getArg(format_idx+numConversions);
1058 if (const BuiltinType *BT = E->getType()->getAsBuiltinType())
1059 if (BT->getKind() == BuiltinType::Int)
1060 break;
Ted Kremenek580b6642007-10-12 20:51:52 +00001061
Ted Kremenek42ae3e82009-05-13 16:06:05 +00001062 SourceLocation Loc = getLocationOfStringLiteralByte(FExpr, StrIdx);
1063
1064 if (Str[StrIdx-1] == '.')
1065 Diag(Loc, diag::warn_printf_asterisk_precision_wrong_type)
1066 << E->getType() << E->getSourceRange();
1067 else
1068 Diag(Loc, diag::warn_printf_asterisk_width_wrong_type)
1069 << E->getType() << E->getSourceRange();
1070
1071 break;
Ted Kremenek580b6642007-10-12 20:51:52 +00001072 }
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001073 }
1074
1075 // Characters which can terminate a format conversion
1076 // (e.g. "%d"). Characters that specify length modifiers or
1077 // other flags are handled by the default case below.
1078 //
1079 // FIXME: additional checks will go into the following cases.
1080 case 'i':
1081 case 'd':
1082 case 'o':
1083 case 'u':
1084 case 'x':
1085 case 'X':
1086 case 'D':
1087 case 'O':
1088 case 'U':
1089 case 'e':
1090 case 'E':
1091 case 'f':
1092 case 'F':
1093 case 'g':
1094 case 'G':
1095 case 'a':
1096 case 'A':
1097 case 'c':
1098 case 'C':
1099 case 'S':
1100 case 's':
1101 case 'p':
1102 ++numConversions;
1103 CurrentState = state_OrdChr;
1104 break;
1105
Eli Friedmanb92abb42009-06-02 08:36:19 +00001106 case 'm':
1107 // FIXME: Warn in situations where this isn't supported!
1108 CurrentState = state_OrdChr;
1109 break;
1110
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001111 // CHECK: Are we using "%n"? Issue a warning.
1112 case 'n': {
1113 ++numConversions;
1114 CurrentState = state_OrdChr;
Chris Lattner60800082009-02-18 17:49:48 +00001115 SourceLocation Loc = getLocationOfStringLiteralByte(FExpr,
1116 LastConversionIdx);
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001117
Chris Lattnerdcd5ef12008-11-19 05:27:50 +00001118 Diag(Loc, diag::warn_printf_write_back)<<OrigFormatExpr->getSourceRange();
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001119 break;
1120 }
Ted Kremenek7ff22b22008-06-16 18:00:42 +00001121
1122 // Handle "%@"
1123 case '@':
1124 // %@ is allowed in ObjC format strings only.
1125 if(ObjCFExpr != NULL)
1126 CurrentState = state_OrdChr;
1127 else {
1128 // Issue a warning: invalid format conversion.
Chris Lattner60800082009-02-18 17:49:48 +00001129 SourceLocation Loc =
1130 getLocationOfStringLiteralByte(FExpr, LastConversionIdx);
Ted Kremenek7ff22b22008-06-16 18:00:42 +00001131
Chris Lattnerd3a94e22008-11-20 06:06:08 +00001132 Diag(Loc, diag::warn_printf_invalid_conversion)
1133 << std::string(Str+LastConversionIdx,
1134 Str+std::min(LastConversionIdx+2, StrLen))
1135 << OrigFormatExpr->getSourceRange();
Ted Kremenek7ff22b22008-06-16 18:00:42 +00001136 }
1137 ++numConversions;
1138 break;
1139
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001140 // Handle "%%"
1141 case '%':
1142 // Sanity check: Was the first "%" character the previous one?
1143 // If not, we will assume that we have a malformed format
1144 // conversion, and that the current "%" character is the start
1145 // of a new conversion.
1146 if (StrIdx - LastConversionIdx == 1)
1147 CurrentState = state_OrdChr;
1148 else {
1149 // Issue a warning: invalid format conversion.
Chris Lattner60800082009-02-18 17:49:48 +00001150 SourceLocation Loc =
1151 getLocationOfStringLiteralByte(FExpr, LastConversionIdx);
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001152
Chris Lattnerd3a94e22008-11-20 06:06:08 +00001153 Diag(Loc, diag::warn_printf_invalid_conversion)
1154 << std::string(Str+LastConversionIdx, Str+StrIdx)
1155 << OrigFormatExpr->getSourceRange();
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001156
1157 // This conversion is broken. Advance to the next format
1158 // conversion.
1159 LastConversionIdx = StrIdx;
1160 ++numConversions;
Ted Kremenek71895b92007-08-14 17:39:48 +00001161 }
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001162 break;
Ted Kremenek71895b92007-08-14 17:39:48 +00001163
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001164 default:
1165 // This case catches all other characters: flags, widths, etc.
1166 // We should eventually process those as well.
1167 break;
Ted Kremenek71895b92007-08-14 17:39:48 +00001168 }
1169 }
1170
1171 if (CurrentState == state_Conversion) {
1172 // Issue a warning: invalid format conversion.
Chris Lattner60800082009-02-18 17:49:48 +00001173 SourceLocation Loc =
1174 getLocationOfStringLiteralByte(FExpr, LastConversionIdx);
Ted Kremenek71895b92007-08-14 17:39:48 +00001175
Chris Lattnerd3a94e22008-11-20 06:06:08 +00001176 Diag(Loc, diag::warn_printf_invalid_conversion)
1177 << std::string(Str+LastConversionIdx,
1178 Str+std::min(LastConversionIdx+2, StrLen))
1179 << OrigFormatExpr->getSourceRange();
Ted Kremenek71895b92007-08-14 17:39:48 +00001180 return;
1181 }
1182
1183 if (!HasVAListArg) {
1184 // CHECK: Does the number of format conversions exceed the number
1185 // of data arguments?
1186 if (numConversions > numDataArgs) {
Chris Lattner60800082009-02-18 17:49:48 +00001187 SourceLocation Loc =
1188 getLocationOfStringLiteralByte(FExpr, LastConversionIdx);
Ted Kremenek71895b92007-08-14 17:39:48 +00001189
Chris Lattnerdcd5ef12008-11-19 05:27:50 +00001190 Diag(Loc, diag::warn_printf_insufficient_data_args)
1191 << OrigFormatExpr->getSourceRange();
Ted Kremenek71895b92007-08-14 17:39:48 +00001192 }
1193 // CHECK: Does the number of data arguments exceed the number of
1194 // format conversions in the format string?
1195 else if (numConversions < numDataArgs)
Chris Lattner925e60d2007-12-28 05:29:59 +00001196 Diag(TheCall->getArg(format_idx+numConversions+1)->getLocStart(),
Chris Lattnerdcd5ef12008-11-19 05:27:50 +00001197 diag::warn_printf_too_many_data_args)
1198 << OrigFormatExpr->getSourceRange();
Ted Kremenek71895b92007-08-14 17:39:48 +00001199 }
1200}
Ted Kremenek06de2762007-08-17 16:46:58 +00001201
1202//===--- CHECK: Return Address of Stack Variable --------------------------===//
1203
1204static DeclRefExpr* EvalVal(Expr *E);
1205static DeclRefExpr* EvalAddr(Expr* E);
1206
1207/// CheckReturnStackAddr - Check if a return statement returns the address
1208/// of a stack variable.
1209void
1210Sema::CheckReturnStackAddr(Expr *RetValExp, QualType lhsType,
1211 SourceLocation ReturnLoc) {
Chris Lattner56f34942008-02-13 01:02:39 +00001212
Ted Kremenek06de2762007-08-17 16:46:58 +00001213 // Perform checking for returned stack addresses.
Steve Naroffdd972f22008-09-05 22:11:13 +00001214 if (lhsType->isPointerType() || lhsType->isBlockPointerType()) {
Ted Kremenek06de2762007-08-17 16:46:58 +00001215 if (DeclRefExpr *DR = EvalAddr(RetValExp))
Chris Lattner3c73c412008-11-19 08:23:25 +00001216 Diag(DR->getLocStart(), diag::warn_ret_stack_addr)
Chris Lattner08631c52008-11-23 21:45:46 +00001217 << DR->getDecl()->getDeclName() << RetValExp->getSourceRange();
Steve Naroffc50a4a52008-09-16 22:25:10 +00001218
1219 // Skip over implicit cast expressions when checking for block expressions.
1220 if (ImplicitCastExpr *IcExpr =
1221 dyn_cast_or_null<ImplicitCastExpr>(RetValExp))
1222 RetValExp = IcExpr->getSubExpr();
1223
Steve Naroff61f40a22008-09-10 19:17:48 +00001224 if (BlockExpr *C = dyn_cast_or_null<BlockExpr>(RetValExp))
Mike Stump397195b2009-04-17 00:09:41 +00001225 if (C->hasBlockDeclRefExprs())
1226 Diag(C->getLocStart(), diag::err_ret_local_block)
1227 << C->getSourceRange();
Mike Stumpac5fc7c2009-08-04 21:02:39 +00001228 } else if (lhsType->isReferenceType()) {
1229 // Perform checking for stack values returned by reference.
Douglas Gregor49badde2008-10-27 19:41:14 +00001230 // Check for a reference to the stack
1231 if (DeclRefExpr *DR = EvalVal(RetValExp))
Chris Lattnerdcd5ef12008-11-19 05:27:50 +00001232 Diag(DR->getLocStart(), diag::warn_ret_stack_ref)
Chris Lattner08631c52008-11-23 21:45:46 +00001233 << DR->getDecl()->getDeclName() << RetValExp->getSourceRange();
Ted Kremenek06de2762007-08-17 16:46:58 +00001234 }
1235}
1236
1237/// EvalAddr - EvalAddr and EvalVal are mutually recursive functions that
1238/// check if the expression in a return statement evaluates to an address
1239/// to a location on the stack. The recursion is used to traverse the
1240/// AST of the return expression, with recursion backtracking when we
1241/// encounter a subexpression that (1) clearly does not lead to the address
1242/// of a stack variable or (2) is something we cannot determine leads to
1243/// the address of a stack variable based on such local checking.
1244///
Ted Kremeneke8c600f2007-08-28 17:02:55 +00001245/// EvalAddr processes expressions that are pointers that are used as
1246/// references (and not L-values). EvalVal handles all other values.
Ted Kremenek06de2762007-08-17 16:46:58 +00001247/// At the base case of the recursion is a check for a DeclRefExpr* in
1248/// the refers to a stack variable.
1249///
1250/// This implementation handles:
1251///
1252/// * pointer-to-pointer casts
1253/// * implicit conversions from array references to pointers
1254/// * taking the address of fields
1255/// * arbitrary interplay between "&" and "*" operators
1256/// * pointer arithmetic from an address of a stack variable
1257/// * taking the address of an array element where the array is on the stack
1258static DeclRefExpr* EvalAddr(Expr *E) {
Ted Kremenek06de2762007-08-17 16:46:58 +00001259 // We should only be called for evaluating pointer expressions.
Steve Naroffdd972f22008-09-05 22:11:13 +00001260 assert((E->getType()->isPointerType() ||
1261 E->getType()->isBlockPointerType() ||
Ted Kremeneka526c5c2008-01-07 19:49:32 +00001262 E->getType()->isObjCQualifiedIdType()) &&
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001263 "EvalAddr only works on pointers");
Ted Kremenek06de2762007-08-17 16:46:58 +00001264
1265 // Our "symbolic interpreter" is just a dispatch off the currently
1266 // viewed AST node. We then recursively traverse the AST by calling
1267 // EvalAddr and EvalVal appropriately.
1268 switch (E->getStmtClass()) {
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001269 case Stmt::ParenExprClass:
1270 // Ignore parentheses.
1271 return EvalAddr(cast<ParenExpr>(E)->getSubExpr());
Ted Kremenek06de2762007-08-17 16:46:58 +00001272
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001273 case Stmt::UnaryOperatorClass: {
1274 // The only unary operator that make sense to handle here
1275 // is AddrOf. All others don't make sense as pointers.
1276 UnaryOperator *U = cast<UnaryOperator>(E);
Ted Kremenek06de2762007-08-17 16:46:58 +00001277
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001278 if (U->getOpcode() == UnaryOperator::AddrOf)
1279 return EvalVal(U->getSubExpr());
1280 else
Ted Kremenek06de2762007-08-17 16:46:58 +00001281 return NULL;
1282 }
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001283
1284 case Stmt::BinaryOperatorClass: {
1285 // Handle pointer arithmetic. All other binary operators are not valid
1286 // in this context.
1287 BinaryOperator *B = cast<BinaryOperator>(E);
1288 BinaryOperator::Opcode op = B->getOpcode();
1289
1290 if (op != BinaryOperator::Add && op != BinaryOperator::Sub)
1291 return NULL;
1292
1293 Expr *Base = B->getLHS();
1294
1295 // Determine which argument is the real pointer base. It could be
1296 // the RHS argument instead of the LHS.
1297 if (!Base->getType()->isPointerType()) Base = B->getRHS();
1298
1299 assert (Base->getType()->isPointerType());
1300 return EvalAddr(Base);
1301 }
Steve Naroff61f40a22008-09-10 19:17:48 +00001302
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001303 // For conditional operators we need to see if either the LHS or RHS are
1304 // valid DeclRefExpr*s. If one of them is valid, we return it.
1305 case Stmt::ConditionalOperatorClass: {
1306 ConditionalOperator *C = cast<ConditionalOperator>(E);
1307
1308 // Handle the GNU extension for missing LHS.
1309 if (Expr *lhsExpr = C->getLHS())
1310 if (DeclRefExpr* LHS = EvalAddr(lhsExpr))
1311 return LHS;
1312
1313 return EvalAddr(C->getRHS());
1314 }
1315
Ted Kremenek54b52742008-08-07 00:49:01 +00001316 // For casts, we need to handle conversions from arrays to
1317 // pointer values, and pointer-to-pointer conversions.
Douglas Gregor49badde2008-10-27 19:41:14 +00001318 case Stmt::ImplicitCastExprClass:
Douglas Gregor6eec8e82008-10-28 15:36:24 +00001319 case Stmt::CStyleCastExprClass:
Douglas Gregor49badde2008-10-27 19:41:14 +00001320 case Stmt::CXXFunctionalCastExprClass: {
Argyrios Kyrtzidis0835a3c2008-08-18 23:01:59 +00001321 Expr* SubExpr = cast<CastExpr>(E)->getSubExpr();
Ted Kremenek54b52742008-08-07 00:49:01 +00001322 QualType T = SubExpr->getType();
1323
Steve Naroffdd972f22008-09-05 22:11:13 +00001324 if (SubExpr->getType()->isPointerType() ||
1325 SubExpr->getType()->isBlockPointerType() ||
1326 SubExpr->getType()->isObjCQualifiedIdType())
Ted Kremenek54b52742008-08-07 00:49:01 +00001327 return EvalAddr(SubExpr);
1328 else if (T->isArrayType())
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001329 return EvalVal(SubExpr);
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001330 else
Ted Kremenek54b52742008-08-07 00:49:01 +00001331 return 0;
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001332 }
1333
1334 // C++ casts. For dynamic casts, static casts, and const casts, we
1335 // are always converting from a pointer-to-pointer, so we just blow
Douglas Gregor49badde2008-10-27 19:41:14 +00001336 // through the cast. In the case the dynamic cast doesn't fail (and
1337 // return NULL), we take the conservative route and report cases
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001338 // where we return the address of a stack variable. For Reinterpre
Douglas Gregor49badde2008-10-27 19:41:14 +00001339 // FIXME: The comment about is wrong; we're not always converting
1340 // from pointer to pointer. I'm guessing that this code should also
1341 // handle references to objects.
1342 case Stmt::CXXStaticCastExprClass:
1343 case Stmt::CXXDynamicCastExprClass:
1344 case Stmt::CXXConstCastExprClass:
1345 case Stmt::CXXReinterpretCastExprClass: {
1346 Expr *S = cast<CXXNamedCastExpr>(E)->getSubExpr();
Steve Naroffdd972f22008-09-05 22:11:13 +00001347 if (S->getType()->isPointerType() || S->getType()->isBlockPointerType())
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001348 return EvalAddr(S);
1349 else
1350 return NULL;
Chris Lattnerfae3f1f2007-12-28 05:31:15 +00001351 }
1352
1353 // Everything else: we simply don't reason about them.
1354 default:
1355 return NULL;
1356 }
Ted Kremenek06de2762007-08-17 16:46:58 +00001357}
1358
1359
1360/// EvalVal - This function is complements EvalAddr in the mutual recursion.
1361/// See the comments for EvalAddr for more details.
1362static DeclRefExpr* EvalVal(Expr *E) {
1363
Ted Kremeneke8c600f2007-08-28 17:02:55 +00001364 // We should only be called for evaluating non-pointer expressions, or
1365 // expressions with a pointer type that are not used as references but instead
1366 // are l-values (e.g., DeclRefExpr with a pointer type).
1367
Ted Kremenek06de2762007-08-17 16:46:58 +00001368 // Our "symbolic interpreter" is just a dispatch off the currently
1369 // viewed AST node. We then recursively traverse the AST by calling
1370 // EvalAddr and EvalVal appropriately.
1371 switch (E->getStmtClass()) {
Douglas Gregor1a49af92009-01-06 05:10:23 +00001372 case Stmt::DeclRefExprClass:
1373 case Stmt::QualifiedDeclRefExprClass: {
Ted Kremenek06de2762007-08-17 16:46:58 +00001374 // DeclRefExpr: the base case. When we hit a DeclRefExpr we are looking
1375 // at code that refers to a variable's name. We check if it has local
1376 // storage within the function, and if so, return the expression.
1377 DeclRefExpr *DR = cast<DeclRefExpr>(E);
1378
1379 if (VarDecl *V = dyn_cast<VarDecl>(DR->getDecl()))
Douglas Gregor27c8dc02008-10-29 00:13:59 +00001380 if(V->hasLocalStorage() && !V->getType()->isReferenceType()) return DR;
Ted Kremenek06de2762007-08-17 16:46:58 +00001381
1382 return NULL;
1383 }
1384
1385 case Stmt::ParenExprClass:
1386 // Ignore parentheses.
1387 return EvalVal(cast<ParenExpr>(E)->getSubExpr());
1388
1389 case Stmt::UnaryOperatorClass: {
1390 // The only unary operator that make sense to handle here
1391 // is Deref. All others don't resolve to a "name." This includes
1392 // handling all sorts of rvalues passed to a unary operator.
1393 UnaryOperator *U = cast<UnaryOperator>(E);
1394
1395 if (U->getOpcode() == UnaryOperator::Deref)
1396 return EvalAddr(U->getSubExpr());
1397
1398 return NULL;
1399 }
1400
1401 case Stmt::ArraySubscriptExprClass: {
1402 // Array subscripts are potential references to data on the stack. We
1403 // retrieve the DeclRefExpr* for the array variable if it indeed
1404 // has local storage.
Ted Kremenek23245122007-08-20 16:18:38 +00001405 return EvalAddr(cast<ArraySubscriptExpr>(E)->getBase());
Ted Kremenek06de2762007-08-17 16:46:58 +00001406 }
1407
1408 case Stmt::ConditionalOperatorClass: {
1409 // For conditional operators we need to see if either the LHS or RHS are
1410 // non-NULL DeclRefExpr's. If one is non-NULL, we return it.
1411 ConditionalOperator *C = cast<ConditionalOperator>(E);
1412
Anders Carlsson39073232007-11-30 19:04:31 +00001413 // Handle the GNU extension for missing LHS.
1414 if (Expr *lhsExpr = C->getLHS())
1415 if (DeclRefExpr *LHS = EvalVal(lhsExpr))
1416 return LHS;
1417
1418 return EvalVal(C->getRHS());
Ted Kremenek06de2762007-08-17 16:46:58 +00001419 }
1420
1421 // Accesses to members are potential references to data on the stack.
1422 case Stmt::MemberExprClass: {
1423 MemberExpr *M = cast<MemberExpr>(E);
1424
1425 // Check for indirect access. We only want direct field accesses.
1426 if (!M->isArrow())
1427 return EvalVal(M->getBase());
1428 else
1429 return NULL;
1430 }
1431
1432 // Everything else: we simply don't reason about them.
1433 default:
1434 return NULL;
1435 }
1436}
Ted Kremenek588e5eb2007-11-25 00:58:00 +00001437
1438//===--- CHECK: Floating-Point comparisons (-Wfloat-equal) ---------------===//
1439
1440/// Check for comparisons of floating point operands using != and ==.
1441/// Issue a warning if these are no self-comparisons, as they are not likely
1442/// to do what the programmer intended.
1443void Sema::CheckFloatComparison(SourceLocation loc, Expr* lex, Expr *rex) {
1444 bool EmitWarning = true;
1445
Ted Kremenek4e99a5f2008-01-17 16:57:34 +00001446 Expr* LeftExprSansParen = lex->IgnoreParens();
Ted Kremenek32e97b62008-01-17 17:55:13 +00001447 Expr* RightExprSansParen = rex->IgnoreParens();
Ted Kremenek588e5eb2007-11-25 00:58:00 +00001448
1449 // Special case: check for x == x (which is OK).
1450 // Do not emit warnings for such cases.
1451 if (DeclRefExpr* DRL = dyn_cast<DeclRefExpr>(LeftExprSansParen))
1452 if (DeclRefExpr* DRR = dyn_cast<DeclRefExpr>(RightExprSansParen))
1453 if (DRL->getDecl() == DRR->getDecl())
1454 EmitWarning = false;
1455
Ted Kremenek1b500bb2007-11-29 00:59:04 +00001456
1457 // Special case: check for comparisons against literals that can be exactly
1458 // represented by APFloat. In such cases, do not emit a warning. This
1459 // is a heuristic: often comparison against such literals are used to
1460 // detect if a value in a variable has not changed. This clearly can
1461 // lead to false negatives.
1462 if (EmitWarning) {
1463 if (FloatingLiteral* FLL = dyn_cast<FloatingLiteral>(LeftExprSansParen)) {
1464 if (FLL->isExact())
1465 EmitWarning = false;
Mike Stumpac5fc7c2009-08-04 21:02:39 +00001466 } else
Ted Kremenek1b500bb2007-11-29 00:59:04 +00001467 if (FloatingLiteral* FLR = dyn_cast<FloatingLiteral>(RightExprSansParen)){
1468 if (FLR->isExact())
1469 EmitWarning = false;
1470 }
1471 }
1472
Ted Kremenek588e5eb2007-11-25 00:58:00 +00001473 // Check for comparisons with builtin types.
Sebastian Redl0eb23302009-01-19 00:08:26 +00001474 if (EmitWarning)
Ted Kremenek588e5eb2007-11-25 00:58:00 +00001475 if (CallExpr* CL = dyn_cast<CallExpr>(LeftExprSansParen))
Douglas Gregor3c385e52009-02-14 18:57:46 +00001476 if (CL->isBuiltinCall(Context))
Ted Kremenek588e5eb2007-11-25 00:58:00 +00001477 EmitWarning = false;
1478
Sebastian Redl0eb23302009-01-19 00:08:26 +00001479 if (EmitWarning)
Ted Kremenek588e5eb2007-11-25 00:58:00 +00001480 if (CallExpr* CR = dyn_cast<CallExpr>(RightExprSansParen))
Douglas Gregor3c385e52009-02-14 18:57:46 +00001481 if (CR->isBuiltinCall(Context))
Ted Kremenek588e5eb2007-11-25 00:58:00 +00001482 EmitWarning = false;
1483
1484 // Emit the diagnostic.
1485 if (EmitWarning)
Chris Lattnerfa25bbb2008-11-19 05:08:23 +00001486 Diag(loc, diag::warn_floatingpoint_eq)
1487 << lex->getSourceRange() << rex->getSourceRange();
Ted Kremenek588e5eb2007-11-25 00:58:00 +00001488}