From 222e2a7620e6520ffaf4fc4e69d79c18da31542e Mon Sep 17 00:00:00 2001 From: "Zancanaro; Carlo" Date: Mon, 24 Sep 2012 09:58:17 +1000 Subject: Add the clang library to the repo (with some of my changes, too). --- clang/test/Analysis/misc-ps-eager-assume.m | 146 +++++++++++++++++++++++++++++ 1 file changed, 146 insertions(+) create mode 100644 clang/test/Analysis/misc-ps-eager-assume.m (limited to 'clang/test/Analysis/misc-ps-eager-assume.m') diff --git a/clang/test/Analysis/misc-ps-eager-assume.m b/clang/test/Analysis/misc-ps-eager-assume.m new file mode 100644 index 0000000..0aff8e4 --- /dev/null +++ b/clang/test/Analysis/misc-ps-eager-assume.m @@ -0,0 +1,146 @@ +// RUN: %clang_cc1 -analyze -analyzer-checker=core,experimental.core -analyzer-store=region -analyzer-constraints=range -verify -fblocks %s -analyzer-eagerly-assume + +// Delta-reduced header stuff (needed for test cases). +typedef signed char BOOL; +typedef unsigned int NSUInteger; +typedef struct _NSZone NSZone; +@class NSInvocation, NSMethodSignature, NSCoder, NSString, NSEnumerator; +@protocol NSObject - (BOOL)isEqual:(id)object; +- (oneway void)release; +@end @protocol NSCopying - (id)copyWithZone:(NSZone *)zone; +@end @protocol NSMutableCopying - (id)mutableCopyWithZone:(NSZone *)zone; +@end @protocol NSCoding - (void)encodeWithCoder:(NSCoder *)aCoder; +@end @interface NSObject {} ++ (id)alloc; +- (id)init; +@end typedef struct {} +NSFastEnumerationState; +@protocol NSFastEnumeration - (NSUInteger)countByEnumeratingWithState:(NSFastEnumerationState *)state objects:(id *)stackbuf count:(NSUInteger)len; +@end @interface NSArray : NSObject - (NSUInteger)count; +@end @interface NSMutableArray : NSArray - (void)addObject:(id)anObject; +- (BOOL)isEqualToString:(NSString *)aString; +@end @interface NSAutoreleasePool : NSObject {} +- (void)drain; +- (id)init; +@end + +// This test case tests that (x != 0) is eagerly evaluated before stored to +// 'y'. This test case complements recoverCastedSymbol (see below) because +// the symbolic expression is stored to 'y' (which is a short instead of an +// int). recoverCastedSymbol() only recovers path-sensitivity when the +// symbolic expression is literally the branch condition. +// +void handle_assign_of_condition(int x) { + // The cast to 'short' causes us to lose symbolic constraint. + short y = (x != 0); + char *p = 0; + if (y) { + // This should be infeasible. + if (!(x != 0)) { + *p = 1; // no-warning + } + } +} + +// From +// +// In this test case, 'needsAnArray' is a signed char. The analyzer tracks +// a symbolic value for this variable, but in the branch condition it is +// promoted to 'int'. Currently the analyzer doesn't reason well about +// promotions of symbolic values, so this test case tests the logic in +// 'recoverCastedSymbol()' (GRExprEngine.cpp) to test that we recover +// path-sensitivity and use the symbol for 'needsAnArray' in the branch +// condition. +// +void handle_symbolic_cast_in_condition(void) { + NSAutoreleasePool* pool = [[NSAutoreleasePool alloc] init]; + + BOOL needsAnArray = [@"aString" isEqualToString:@"anotherString"]; + NSMutableArray* array = needsAnArray ? [[NSMutableArray alloc] init] : 0; + if(needsAnArray) + [array release]; + + [pool drain]; +} + +// From PR 3836 (http://llvm.org/bugs/show_bug.cgi?id=3836) +// +// In this test case, the double '!' works fine with our symbolic constraints, +// but we don't support comparing SymConstraint != SymConstraint. By eagerly +// assuming the truth of !!a or !!b, we can compare these values directly. +// +void pr3836(int *a, int *b) { + if (!!a != !!b) /* one of them is NULL */ + return; + if (!a && !b) /* both are NULL */ + return; + + *a = 1; // no-warning + *b = 1; // no-warning +} + + +//===---------------------------------------------------------------------===// +// +// This false positive occurred because the symbolic constraint on a short was +// not maintained via sign extension. The analyzer doesn't properly handle +// the sign extension, but now tracks the constraint. This particular +// case relies on -analyzer-eagerly-assume because of the expression +// 'Flag1 != Count > 0'. +//===---------------------------------------------------------------------===// + +void rdar7342806_aux(short x); + +void rdar7342806() { + extern short Count; + extern short Flag1; + + short *Pointer = 0; + short Flag2 = !!Pointer; // Flag2 is false (0). + short Ok = 1; + short Which; + + if( Flag1 != Count > 0 ) + // Static analyzer skips this so either + // Flag1 is true and Count > 0 + // or + // Flag1 is false and Count <= 0 + Ok = 0; + + if( Flag1 != Flag2 ) + // Analyzer skips this so Flag1 and Flag2 have the + // same value, both are false because Flag2 is false. And + // from that we know Count must be <= 0. + Ok = 0; + + for( Which = 0; + Which < Count && Ok; + Which++ ) + // This statement can only execute if Count > 0 which can only + // happen when Flag1 and Flag2 are both true and Flag2 will only + // be true when Pointer is not NULL. + rdar7342806_aux(*Pointer); // no-warning +} + +//===---------------------------------------------------------------------===// +// PR 5627 - http://llvm.org/bugs/show_bug.cgi?id=5627 +// This test case depends on using -analyzer-eagerly-assume and +// -analyzer-store=region. The '-analyzer-eagerly-assume' causes the path +// to bifurcate when evaluating the function call argument, and a state +// caching bug in GRExprEngine::CheckerVisit (and friends) caused the store +// to 'p' to not be evaluated along one path, but then an autotransition caused +// the path to keep on propagating with 'p' still set to an undefined value. +// We would then get a bogus report of returning uninitialized memory. +// Note: CheckerVisit mistakenly cleared an existing node, and the cleared +// node was resurrected by GRStmtNodeBuilder::~GRStmtNodeBuilder(), where +// 'p' was not assigned. +//===---------------------------------------------------------------------===// + +float *pr5627_f(int y); + +float *pr5627_g(int x) { + float *p; + p = pr5627_f(!x); + return p; // no-warning +} + -- cgit v1.2.3