blob: cb22f03a31fd22fbeaf4c4704f6c0731cc2bc1c9 [file] [log] [blame]
Chris Lattner59907c42007-08-10 20:18:51 +00001//===--- SemaChecking.cpp - Extra Semantic Checking -----------------------===//
2//
3// The LLVM Compiler Infrastructure
4//
5// This file was developed by Ted Kremenek and is distributed under
6// the University of Illinois Open Source License. See LICENSE.TXT for details.
7//
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"
17#include "clang/AST/Decl.h"
18#include "clang/AST/Expr.h"
Ted Kremenek23245122007-08-20 16:18:38 +000019#include "clang/AST/ExprCXX.h"
Chris Lattner59907c42007-08-10 20:18:51 +000020#include "clang/Lex/Preprocessor.h"
21#include "clang/Lex/LiteralSupport.h"
22#include "clang/Basic/SourceManager.h"
23#include "clang/Basic/Diagnostic.h"
24#include "clang/Basic/LangOptions.h"
25#include "clang/Basic/TargetInfo.h"
26#include "llvm/ADT/SmallString.h"
27#include "llvm/ADT/StringExtras.h"
Ted Kremenek588e5eb2007-11-25 00:58:00 +000028#include "SemaUtil.h"
29
Chris Lattner59907c42007-08-10 20:18:51 +000030using namespace clang;
31
32/// CheckFunctionCall - Check a direct function call for various correctness
33/// and safety properties not strictly enforced by the C type system.
Anders Carlsson71993dd2007-08-17 05:31:46 +000034bool
Ted Kremenek71895b92007-08-14 17:39:48 +000035Sema::CheckFunctionCall(Expr *Fn,
36 SourceLocation LParenLoc, SourceLocation RParenLoc,
37 FunctionDecl *FDecl,
Chris Lattner59907c42007-08-10 20:18:51 +000038 Expr** Args, unsigned NumArgsInCall) {
39
40 // Get the IdentifierInfo* for the called function.
41 IdentifierInfo *FnInfo = FDecl->getIdentifier();
42
Anders Carlsson71993dd2007-08-17 05:31:46 +000043 if (FnInfo->getBuiltinID() ==
44 Builtin::BI__builtin___CFStringMakeConstantString) {
45 assert(NumArgsInCall == 1 &&
Chris Lattner6f4b92c2007-08-30 17:08:17 +000046 "Wrong number of arguments to builtin CFStringMakeConstantString");
Anders Carlsson71993dd2007-08-17 05:31:46 +000047 return CheckBuiltinCFStringArgument(Args[0]);
Anders Carlsson6eda8c92007-10-12 17:48:41 +000048 } else if (FnInfo->getBuiltinID() == Builtin::BI__builtin_va_start) {
49 if (NumArgsInCall > 2) {
50 Diag(Args[2]->getLocStart(),
51 diag::err_typecheck_call_too_many_args, Fn->getSourceRange(),
52 SourceRange(Args[2]->getLocStart(),
53 Args[NumArgsInCall - 1]->getLocEnd()));
54 return true;
55 }
56
57 FunctionTypeProto* proto =
58 cast<FunctionTypeProto>(CurFunctionDecl->getType());
59 if (!proto->isVariadic()) {
60 Diag(Fn->getLocStart(),
61 diag::err_va_start_used_in_non_variadic_function);
62 return true;
63 }
64
65 bool SecondArgIsLastNamedArgument = false;
66 if (DeclRefExpr *DR = dyn_cast<DeclRefExpr>(Args[1])) {
67 if (ParmVarDecl *PV = dyn_cast<ParmVarDecl>(DR->getDecl())) {
68 ParmVarDecl *LastNamedArg =
69 CurFunctionDecl->getParamDecl(CurFunctionDecl->getNumParams() - 1);
70
71 if (PV == LastNamedArg)
72 SecondArgIsLastNamedArgument = true;
73 }
74 }
75
76 if (!SecondArgIsLastNamedArgument)
77 Diag(Args[1]->getLocStart(),
78 diag::warn_second_parameter_of_va_start_not_last_named_argument);
Anders Carlsson71993dd2007-08-17 05:31:46 +000079 }
80
Chris Lattner59907c42007-08-10 20:18:51 +000081 // Search the KnownFunctionIDs for the identifier.
82 unsigned i = 0, e = id_num_known_functions;
Ted Kremenek71895b92007-08-14 17:39:48 +000083 for (; i != e; ++i) { if (KnownFunctionIDs[i] == FnInfo) break; }
Anders Carlsson9cdc4d32007-08-17 15:44:17 +000084 if (i == e) return false;
Chris Lattner59907c42007-08-10 20:18:51 +000085
86 // Printf checking.
87 if (i <= id_vprintf) {
Ted Kremenek71895b92007-08-14 17:39:48 +000088 // Retrieve the index of the format string parameter and determine
89 // if the function is passed a va_arg argument.
Chris Lattner59907c42007-08-10 20:18:51 +000090 unsigned format_idx = 0;
Ted Kremenek71895b92007-08-14 17:39:48 +000091 bool HasVAListArg = false;
92
Chris Lattner59907c42007-08-10 20:18:51 +000093 switch (i) {
94 default: assert(false && "No format string argument index.");
95 case id_printf: format_idx = 0; break;
96 case id_fprintf: format_idx = 1; break;
97 case id_sprintf: format_idx = 1; break;
98 case id_snprintf: format_idx = 2; break;
Ted Kremenek71895b92007-08-14 17:39:48 +000099 case id_asprintf: format_idx = 1; HasVAListArg = true; break;
100 case id_vsnprintf: format_idx = 2; HasVAListArg = true; break;
101 case id_vasprintf: format_idx = 1; HasVAListArg = true; break;
102 case id_vfprintf: format_idx = 1; HasVAListArg = true; break;
103 case id_vsprintf: format_idx = 1; HasVAListArg = true; break;
104 case id_vprintf: format_idx = 0; HasVAListArg = true; break;
105 }
106
107 CheckPrintfArguments(Fn, LParenLoc, RParenLoc, HasVAListArg,
Ted Kremenek06de2762007-08-17 16:46:58 +0000108 FDecl, format_idx, Args, NumArgsInCall);
Chris Lattner59907c42007-08-10 20:18:51 +0000109 }
Anders Carlsson71993dd2007-08-17 05:31:46 +0000110
Anders Carlsson9cdc4d32007-08-17 15:44:17 +0000111 return false;
Anders Carlsson71993dd2007-08-17 05:31:46 +0000112}
113
114/// CheckBuiltinCFStringArgument - Checks that the argument to the builtin
115/// CFString constructor is correct
Chris Lattnercc6f65d2007-08-25 05:30:33 +0000116bool Sema::CheckBuiltinCFStringArgument(Expr* Arg) {
Chris Lattner459e8482007-08-25 05:36:18 +0000117 // FIXME: This should go in a helper.
Chris Lattnercc6f65d2007-08-25 05:30:33 +0000118 while (1) {
119 if (ParenExpr *PE = dyn_cast<ParenExpr>(Arg))
120 Arg = PE->getSubExpr();
121 else if (ImplicitCastExpr *ICE = dyn_cast<ImplicitCastExpr>(Arg))
122 Arg = ICE->getSubExpr();
123 else
124 break;
125 }
Anders Carlsson71993dd2007-08-17 05:31:46 +0000126
127 StringLiteral *Literal = dyn_cast<StringLiteral>(Arg);
128
129 if (!Literal || Literal->isWide()) {
130 Diag(Arg->getLocStart(),
131 diag::err_cfstring_literal_not_string_constant,
132 Arg->getSourceRange());
Anders Carlsson9cdc4d32007-08-17 15:44:17 +0000133 return true;
Anders Carlsson71993dd2007-08-17 05:31:46 +0000134 }
135
136 const char *Data = Literal->getStrData();
137 unsigned Length = Literal->getByteLength();
138
139 for (unsigned i = 0; i < Length; ++i) {
140 if (!isascii(Data[i])) {
141 Diag(PP.AdvanceToTokenCharacter(Arg->getLocStart(), i + 1),
142 diag::warn_cfstring_literal_contains_non_ascii_character,
143 Arg->getSourceRange());
144 break;
145 }
146
147 if (!Data[i]) {
148 Diag(PP.AdvanceToTokenCharacter(Arg->getLocStart(), i + 1),
149 diag::warn_cfstring_literal_contains_nul_character,
150 Arg->getSourceRange());
151 break;
152 }
153 }
154
Anders Carlsson9cdc4d32007-08-17 15:44:17 +0000155 return false;
Chris Lattner59907c42007-08-10 20:18:51 +0000156}
157
158/// CheckPrintfArguments - Check calls to printf (and similar functions) for
Ted Kremenek71895b92007-08-14 17:39:48 +0000159/// correct use of format strings.
160///
161/// HasVAListArg - A predicate indicating whether the printf-like
162/// function is passed an explicit va_arg argument (e.g., vprintf)
163///
164/// format_idx - The index into Args for the format string.
165///
166/// Improper format strings to functions in the printf family can be
167/// the source of bizarre bugs and very serious security holes. A
168/// good source of information is available in the following paper
169/// (which includes additional references):
Chris Lattner59907c42007-08-10 20:18:51 +0000170///
171/// FormatGuard: Automatic Protection From printf Format String
172/// Vulnerabilities, Proceedings of the 10th USENIX Security Symposium, 2001.
Ted Kremenek71895b92007-08-14 17:39:48 +0000173///
174/// Functionality implemented:
175///
176/// We can statically check the following properties for string
177/// literal format strings for non v.*printf functions (where the
178/// arguments are passed directly):
179//
180/// (1) Are the number of format conversions equal to the number of
181/// data arguments?
182///
183/// (2) Does each format conversion correctly match the type of the
184/// corresponding data argument? (TODO)
185///
186/// Moreover, for all printf functions we can:
187///
188/// (3) Check for a missing format string (when not caught by type checking).
189///
190/// (4) Check for no-operation flags; e.g. using "#" with format
191/// conversion 'c' (TODO)
192///
193/// (5) Check the use of '%n', a major source of security holes.
194///
195/// (6) Check for malformed format conversions that don't specify anything.
196///
197/// (7) Check for empty format strings. e.g: printf("");
198///
199/// (8) Check that the format string is a wide literal.
200///
201/// All of these checks can be done by parsing the format string.
202///
203/// For now, we ONLY do (1), (3), (5), (6), (7), and (8).
Chris Lattner59907c42007-08-10 20:18:51 +0000204void
Ted Kremenek71895b92007-08-14 17:39:48 +0000205Sema::CheckPrintfArguments(Expr *Fn,
206 SourceLocation LParenLoc, SourceLocation RParenLoc,
207 bool HasVAListArg, FunctionDecl *FDecl,
Ted Kremenek82077102007-08-10 21:21:05 +0000208 unsigned format_idx, Expr** Args,
209 unsigned NumArgsInCall) {
Ted Kremenek71895b92007-08-14 17:39:48 +0000210 // CHECK: printf-like function is called with no format string.
211 if (format_idx >= NumArgsInCall) {
212 Diag(RParenLoc, diag::warn_printf_missing_format_string,
213 Fn->getSourceRange());
214 return;
215 }
216
Chris Lattner459e8482007-08-25 05:36:18 +0000217 Expr *OrigFormatExpr = Args[format_idx];
218 // FIXME: This should go in a helper.
219 while (1) {
220 if (ParenExpr *PE = dyn_cast<ParenExpr>(OrigFormatExpr))
221 OrigFormatExpr = PE->getSubExpr();
222 else if (ImplicitCastExpr *ICE = dyn_cast<ImplicitCastExpr>(OrigFormatExpr))
223 OrigFormatExpr = ICE->getSubExpr();
224 else
225 break;
226 }
227
Chris Lattner59907c42007-08-10 20:18:51 +0000228 // CHECK: format string is not a string literal.
229 //
Ted Kremenek71895b92007-08-14 17:39:48 +0000230 // Dynamically generated format strings are difficult to
231 // automatically vet at compile time. Requiring that format strings
232 // are string literals: (1) permits the checking of format strings by
233 // the compiler and thereby (2) can practically remove the source of
234 // many format string exploits.
Chris Lattner459e8482007-08-25 05:36:18 +0000235 StringLiteral *FExpr = dyn_cast<StringLiteral>(OrigFormatExpr);
Chris Lattner59907c42007-08-10 20:18:51 +0000236
Ted Kremenek71895b92007-08-14 17:39:48 +0000237 if (FExpr == NULL) {
238 Diag(Args[format_idx]->getLocStart(),
239 diag::warn_printf_not_string_constant, Fn->getSourceRange());
240 return;
241 }
242
243 // CHECK: is the format string a wide literal?
244 if (FExpr->isWide()) {
245 Diag(Args[format_idx]->getLocStart(),
246 diag::warn_printf_format_string_is_wide_literal,
247 Fn->getSourceRange());
248 return;
249 }
250
251 // Str - The format string. NOTE: this is NOT null-terminated!
252 const char * const Str = FExpr->getStrData();
253
254 // CHECK: empty format string?
255 const unsigned StrLen = FExpr->getByteLength();
256
257 if (StrLen == 0) {
258 Diag(Args[format_idx]->getLocStart(),
259 diag::warn_printf_empty_format_string, Fn->getSourceRange());
260 return;
261 }
262
263 // We process the format string using a binary state machine. The
264 // current state is stored in CurrentState.
265 enum {
266 state_OrdChr,
267 state_Conversion
268 } CurrentState = state_OrdChr;
269
270 // numConversions - The number of conversions seen so far. This is
271 // incremented as we traverse the format string.
272 unsigned numConversions = 0;
273
274 // numDataArgs - The number of data arguments after the format
275 // string. This can only be determined for non vprintf-like
276 // functions. For those functions, this value is 1 (the sole
277 // va_arg argument).
278 unsigned numDataArgs = NumArgsInCall-(format_idx+1);
279
280 // Inspect the format string.
281 unsigned StrIdx = 0;
282
283 // LastConversionIdx - Index within the format string where we last saw
284 // a '%' character that starts a new format conversion.
285 unsigned LastConversionIdx = 0;
286
287 for ( ; StrIdx < StrLen ; ++StrIdx ) {
288
289 // Is the number of detected conversion conversions greater than
290 // the number of matching data arguments? If so, stop.
291 if (!HasVAListArg && numConversions > numDataArgs) break;
292
293 // Handle "\0"
294 if(Str[StrIdx] == '\0' ) {
295 // The string returned by getStrData() is not null-terminated,
296 // so the presence of a null character is likely an error.
297
298 SourceLocation Loc =
299 PP.AdvanceToTokenCharacter(Args[format_idx]->getLocStart(),StrIdx+1);
300
301 Diag(Loc, diag::warn_printf_format_string_contains_null_char,
302 Fn->getSourceRange());
303
304 return;
305 }
306
307 // Ordinary characters (not processing a format conversion).
308 if (CurrentState == state_OrdChr) {
309 if (Str[StrIdx] == '%') {
310 CurrentState = state_Conversion;
311 LastConversionIdx = StrIdx;
312 }
313 continue;
314 }
315
316 // Seen '%'. Now processing a format conversion.
317 switch (Str[StrIdx]) {
Ted Kremenek580b6642007-10-12 20:51:52 +0000318 // Handle dynamic precision or width specifier.
319 case '*': {
320 ++numConversions;
321
322 if (!HasVAListArg && numConversions > numDataArgs) {
323
324 SourceLocation Loc =
325 PP.AdvanceToTokenCharacter(Args[format_idx]->getLocStart(),
326 StrIdx+1);
327
328 if (Str[StrIdx-1] == '.')
329 Diag(Loc, diag::warn_printf_asterisk_precision_missing_arg,
330 Fn->getSourceRange());
331 else
332 Diag(Loc, diag::warn_printf_asterisk_width_missing_arg,
333 Fn->getSourceRange());
334
335 // Don't do any more checking. We'll just emit spurious errors.
336 return;
337 }
338
339 // Perform type checking on width/precision specifier.
340 Expr* E = Args[format_idx+numConversions];
341 QualType T = E->getType().getCanonicalType();
342 if (BuiltinType *BT = dyn_cast<BuiltinType>(T))
343 if (BT->getKind() == BuiltinType::Int)
344 break;
345
346 SourceLocation Loc =
347 PP.AdvanceToTokenCharacter(Args[format_idx]->getLocStart(),
348 StrIdx+1);
349
350 if (Str[StrIdx-1] == '.')
351 Diag(Loc, diag::warn_printf_asterisk_precision_wrong_type,
352 T.getAsString(), E->getSourceRange());
353 else
354 Diag(Loc, diag::warn_printf_asterisk_width_wrong_type,
355 T.getAsString(), E->getSourceRange());
356
Ted Kremenekbef679c2007-10-12 00:11:27 +0000357 break;
Ted Kremenek580b6642007-10-12 20:51:52 +0000358 }
Ted Kremenekbef679c2007-10-12 00:11:27 +0000359
Ted Kremenek71895b92007-08-14 17:39:48 +0000360 // Characters which can terminate a format conversion
361 // (e.g. "%d"). Characters that specify length modifiers or
362 // other flags are handled by the default case below.
363 //
Ted Kremenekbef679c2007-10-12 00:11:27 +0000364 // FIXME: additional checks will go into the following cases.
Ted Kremenek71895b92007-08-14 17:39:48 +0000365 case 'i':
366 case 'd':
367 case 'o':
368 case 'u':
369 case 'x':
370 case 'X':
371 case 'D':
372 case 'O':
373 case 'U':
374 case 'e':
375 case 'E':
376 case 'f':
377 case 'F':
378 case 'g':
379 case 'G':
380 case 'a':
381 case 'A':
382 case 'c':
383 case 'C':
384 case 'S':
385 case 's':
Chris Lattner5e9885d2007-08-26 17:39:38 +0000386 case 'p':
Ted Kremenek71895b92007-08-14 17:39:48 +0000387 ++numConversions;
388 CurrentState = state_OrdChr;
389 break;
390
391 // CHECK: Are we using "%n"? Issue a warning.
392 case 'n': {
393 ++numConversions;
394 CurrentState = state_OrdChr;
395 SourceLocation Loc =
396 PP.AdvanceToTokenCharacter(Args[format_idx]->getLocStart(),
397 LastConversionIdx+1);
398
399 Diag(Loc, diag::warn_printf_write_back, Fn->getSourceRange());
400 break;
401 }
402
403 // Handle "%%"
404 case '%':
405 // Sanity check: Was the first "%" character the previous one?
406 // If not, we will assume that we have a malformed format
407 // conversion, and that the current "%" character is the start
408 // of a new conversion.
409 if (StrIdx - LastConversionIdx == 1)
410 CurrentState = state_OrdChr;
411 else {
412 // Issue a warning: invalid format conversion.
413 SourceLocation Loc =
414 PP.AdvanceToTokenCharacter(Args[format_idx]->getLocStart(),
415 LastConversionIdx+1);
416
417 Diag(Loc, diag::warn_printf_invalid_conversion,
Ted Kremenek580b6642007-10-12 20:51:52 +0000418 std::string(Str+LastConversionIdx, Str+StrIdx),
Ted Kremenek71895b92007-08-14 17:39:48 +0000419 Fn->getSourceRange());
420
421 // This conversion is broken. Advance to the next format
422 // conversion.
423 LastConversionIdx = StrIdx;
424 ++numConversions;
425 }
426
427 break;
428
429 default:
430 // This case catches all other characters: flags, widths, etc.
431 // We should eventually process those as well.
432 break;
433 }
434 }
435
436 if (CurrentState == state_Conversion) {
437 // Issue a warning: invalid format conversion.
438 SourceLocation Loc =
439 PP.AdvanceToTokenCharacter(Args[format_idx]->getLocStart(),
440 LastConversionIdx+1);
441
442 Diag(Loc, diag::warn_printf_invalid_conversion,
Chris Lattnera9e2ea12007-08-26 17:38:22 +0000443 std::string(Str+LastConversionIdx,
444 Str+std::min(LastConversionIdx+2, StrLen)),
Ted Kremenek71895b92007-08-14 17:39:48 +0000445 Fn->getSourceRange());
446 return;
447 }
448
449 if (!HasVAListArg) {
450 // CHECK: Does the number of format conversions exceed the number
451 // of data arguments?
452 if (numConversions > numDataArgs) {
453 SourceLocation Loc =
454 PP.AdvanceToTokenCharacter(Args[format_idx]->getLocStart(),
455 LastConversionIdx);
456
457 Diag(Loc, diag::warn_printf_insufficient_data_args,
458 Fn->getSourceRange());
459 }
460 // CHECK: Does the number of data arguments exceed the number of
461 // format conversions in the format string?
462 else if (numConversions < numDataArgs)
463 Diag(Args[format_idx+numConversions+1]->getLocStart(),
464 diag::warn_printf_too_many_data_args, Fn->getSourceRange());
465 }
466}
Ted Kremenek06de2762007-08-17 16:46:58 +0000467
468//===--- CHECK: Return Address of Stack Variable --------------------------===//
469
470static DeclRefExpr* EvalVal(Expr *E);
471static DeclRefExpr* EvalAddr(Expr* E);
472
473/// CheckReturnStackAddr - Check if a return statement returns the address
474/// of a stack variable.
475void
476Sema::CheckReturnStackAddr(Expr *RetValExp, QualType lhsType,
477 SourceLocation ReturnLoc) {
478
479 // Perform checking for returned stack addresses.
480 if (lhsType->isPointerType()) {
481 if (DeclRefExpr *DR = EvalAddr(RetValExp))
482 Diag(DR->getLocStart(), diag::warn_ret_stack_addr,
483 DR->getDecl()->getIdentifier()->getName(),
484 RetValExp->getSourceRange());
485 }
486 // Perform checking for stack values returned by reference.
487 else if (lhsType->isReferenceType()) {
Ted Kremenek96eabe02007-08-27 16:39:17 +0000488 // Check for an implicit cast to a reference.
489 if (ImplicitCastExpr *I = dyn_cast<ImplicitCastExpr>(RetValExp))
490 if (DeclRefExpr *DR = EvalVal(I->getSubExpr()))
491 Diag(DR->getLocStart(), diag::warn_ret_stack_ref,
492 DR->getDecl()->getIdentifier()->getName(),
493 RetValExp->getSourceRange());
Ted Kremenek06de2762007-08-17 16:46:58 +0000494 }
495}
496
497/// EvalAddr - EvalAddr and EvalVal are mutually recursive functions that
498/// check if the expression in a return statement evaluates to an address
499/// to a location on the stack. The recursion is used to traverse the
500/// AST of the return expression, with recursion backtracking when we
501/// encounter a subexpression that (1) clearly does not lead to the address
502/// of a stack variable or (2) is something we cannot determine leads to
503/// the address of a stack variable based on such local checking.
504///
Ted Kremeneke8c600f2007-08-28 17:02:55 +0000505/// EvalAddr processes expressions that are pointers that are used as
506/// references (and not L-values). EvalVal handles all other values.
Ted Kremenek06de2762007-08-17 16:46:58 +0000507/// At the base case of the recursion is a check for a DeclRefExpr* in
508/// the refers to a stack variable.
509///
510/// This implementation handles:
511///
512/// * pointer-to-pointer casts
513/// * implicit conversions from array references to pointers
514/// * taking the address of fields
515/// * arbitrary interplay between "&" and "*" operators
516/// * pointer arithmetic from an address of a stack variable
517/// * taking the address of an array element where the array is on the stack
518static DeclRefExpr* EvalAddr(Expr *E) {
519
520 // We should only be called for evaluating pointer expressions.
521 assert (E->getType()->isPointerType() && "EvalAddr only works on pointers");
522
523 // Our "symbolic interpreter" is just a dispatch off the currently
524 // viewed AST node. We then recursively traverse the AST by calling
525 // EvalAddr and EvalVal appropriately.
526 switch (E->getStmtClass()) {
527
528 case Stmt::ParenExprClass:
529 // Ignore parentheses.
530 return EvalAddr(cast<ParenExpr>(E)->getSubExpr());
531
532 case Stmt::UnaryOperatorClass: {
533 // The only unary operator that make sense to handle here
534 // is AddrOf. All others don't make sense as pointers.
535 UnaryOperator *U = cast<UnaryOperator>(E);
536
537 if (U->getOpcode() == UnaryOperator::AddrOf)
538 return EvalVal(U->getSubExpr());
539 else
540 return NULL;
541 }
542
543 case Stmt::BinaryOperatorClass: {
544 // Handle pointer arithmetic. All other binary operators are not valid
545 // in this context.
546 BinaryOperator *B = cast<BinaryOperator>(E);
547 BinaryOperator::Opcode op = B->getOpcode();
548
549 if (op != BinaryOperator::Add && op != BinaryOperator::Sub)
550 return NULL;
551
552 Expr *Base = B->getLHS();
553
554 // Determine which argument is the real pointer base. It could be
555 // the RHS argument instead of the LHS.
556 if (!Base->getType()->isPointerType()) Base = B->getRHS();
557
558 assert (Base->getType()->isPointerType());
559 return EvalAddr(Base);
560 }
561
562 // For conditional operators we need to see if either the LHS or RHS are
563 // valid DeclRefExpr*s. If one of them is valid, we return it.
564 case Stmt::ConditionalOperatorClass: {
565 ConditionalOperator *C = cast<ConditionalOperator>(E);
566
567 if (DeclRefExpr* LHS = EvalAddr(C->getLHS()))
568 return LHS;
569 else
570 return EvalAddr(C->getRHS());
571 }
572
573 // For implicit casts, we need to handle conversions from arrays to
574 // pointer values, and implicit pointer-to-pointer conversions.
575 case Stmt::ImplicitCastExprClass: {
576 ImplicitCastExpr *IE = cast<ImplicitCastExpr>(E);
577 Expr* SubExpr = IE->getSubExpr();
578
579 if (SubExpr->getType()->isPointerType())
580 return EvalAddr(SubExpr);
581 else
582 return EvalVal(SubExpr);
583 }
584
585 // For casts, we handle pointer-to-pointer conversions (which
586 // is essentially a no-op from our mini-interpreter's standpoint).
587 // For other casts we abort.
588 case Stmt::CastExprClass: {
589 CastExpr *C = cast<CastExpr>(E);
590 Expr *SubExpr = C->getSubExpr();
591
592 if (SubExpr->getType()->isPointerType())
593 return EvalAddr(SubExpr);
594 else
595 return NULL;
596 }
597
Ted Kremenek23245122007-08-20 16:18:38 +0000598 // C++ casts. For dynamic casts, static casts, and const casts, we
599 // are always converting from a pointer-to-pointer, so we just blow
600 // through the cast. In the case the dynamic cast doesn't fail
601 // (and return NULL), we take the conservative route and report cases
602 // where we return the address of a stack variable. For Reinterpre
603 case Stmt::CXXCastExprClass: {
604 CXXCastExpr *C = cast<CXXCastExpr>(E);
605
606 if (C->getOpcode() == CXXCastExpr::ReinterpretCast) {
607 Expr *S = C->getSubExpr();
608 if (S->getType()->isPointerType())
609 return EvalAddr(S);
610 else
611 return NULL;
612 }
613 else
614 return EvalAddr(C->getSubExpr());
615 }
Ted Kremenek06de2762007-08-17 16:46:58 +0000616
617 // Everything else: we simply don't reason about them.
618 default:
619 return NULL;
620 }
621}
622
623
624/// EvalVal - This function is complements EvalAddr in the mutual recursion.
625/// See the comments for EvalAddr for more details.
626static DeclRefExpr* EvalVal(Expr *E) {
627
Ted Kremeneke8c600f2007-08-28 17:02:55 +0000628 // We should only be called for evaluating non-pointer expressions, or
629 // expressions with a pointer type that are not used as references but instead
630 // are l-values (e.g., DeclRefExpr with a pointer type).
631
Ted Kremenek06de2762007-08-17 16:46:58 +0000632 // Our "symbolic interpreter" is just a dispatch off the currently
633 // viewed AST node. We then recursively traverse the AST by calling
634 // EvalAddr and EvalVal appropriately.
635 switch (E->getStmtClass()) {
636
637 case Stmt::DeclRefExprClass: {
638 // DeclRefExpr: the base case. When we hit a DeclRefExpr we are looking
639 // at code that refers to a variable's name. We check if it has local
640 // storage within the function, and if so, return the expression.
641 DeclRefExpr *DR = cast<DeclRefExpr>(E);
642
643 if (VarDecl *V = dyn_cast<VarDecl>(DR->getDecl()))
644 if(V->hasLocalStorage()) return DR;
645
646 return NULL;
647 }
648
649 case Stmt::ParenExprClass:
650 // Ignore parentheses.
651 return EvalVal(cast<ParenExpr>(E)->getSubExpr());
652
653 case Stmt::UnaryOperatorClass: {
654 // The only unary operator that make sense to handle here
655 // is Deref. All others don't resolve to a "name." This includes
656 // handling all sorts of rvalues passed to a unary operator.
657 UnaryOperator *U = cast<UnaryOperator>(E);
658
659 if (U->getOpcode() == UnaryOperator::Deref)
660 return EvalAddr(U->getSubExpr());
661
662 return NULL;
663 }
664
665 case Stmt::ArraySubscriptExprClass: {
666 // Array subscripts are potential references to data on the stack. We
667 // retrieve the DeclRefExpr* for the array variable if it indeed
668 // has local storage.
Ted Kremenek23245122007-08-20 16:18:38 +0000669 return EvalAddr(cast<ArraySubscriptExpr>(E)->getBase());
Ted Kremenek06de2762007-08-17 16:46:58 +0000670 }
671
672 case Stmt::ConditionalOperatorClass: {
673 // For conditional operators we need to see if either the LHS or RHS are
674 // non-NULL DeclRefExpr's. If one is non-NULL, we return it.
675 ConditionalOperator *C = cast<ConditionalOperator>(E);
676
677 if (DeclRefExpr *LHS = EvalVal(C->getLHS()))
678 return LHS;
679 else
680 return EvalVal(C->getRHS());
681 }
682
683 // Accesses to members are potential references to data on the stack.
684 case Stmt::MemberExprClass: {
685 MemberExpr *M = cast<MemberExpr>(E);
686
687 // Check for indirect access. We only want direct field accesses.
688 if (!M->isArrow())
689 return EvalVal(M->getBase());
690 else
691 return NULL;
692 }
693
694 // Everything else: we simply don't reason about them.
695 default:
696 return NULL;
697 }
698}
Ted Kremenek588e5eb2007-11-25 00:58:00 +0000699
700//===--- CHECK: Floating-Point comparisons (-Wfloat-equal) ---------------===//
701
702/// Check for comparisons of floating point operands using != and ==.
703/// Issue a warning if these are no self-comparisons, as they are not likely
704/// to do what the programmer intended.
705void Sema::CheckFloatComparison(SourceLocation loc, Expr* lex, Expr *rex) {
706 bool EmitWarning = true;
707
708 Expr* LeftExprSansParen = IgnoreParen(lex);
709 Expr* RightExprSansParen = IgnoreParen(rex);
710
711 // Special case: check for x == x (which is OK).
712 // Do not emit warnings for such cases.
713 if (DeclRefExpr* DRL = dyn_cast<DeclRefExpr>(LeftExprSansParen))
714 if (DeclRefExpr* DRR = dyn_cast<DeclRefExpr>(RightExprSansParen))
715 if (DRL->getDecl() == DRR->getDecl())
716 EmitWarning = false;
717
718 // Check for comparisons with builtin types.
719 if (EmitWarning)
720 if (CallExpr* CL = dyn_cast<CallExpr>(LeftExprSansParen))
721 if (isCallBuiltin(CL))
722 EmitWarning = false;
723
724 if (EmitWarning)
725 if (CallExpr* CR = dyn_cast<CallExpr>(RightExprSansParen))
726 if (isCallBuiltin(CR))
727 EmitWarning = false;
728
729 // Emit the diagnostic.
730 if (EmitWarning)
731 Diag(loc, diag::warn_floatingpoint_eq,
732 lex->getSourceRange(),rex->getSourceRange());
733}