Although we currently have explicit lvalue-to-rvalue conversions, they're
not actually frequently used, because ImpCastExprToType only creates a node
if the types differ. So explicitly create an ICE in the lvalue-to-rvalue
conversion code in DefaultFunctionArrayLvalueConversion() as well as several
other new places, and consistently deal with the consequences throughout the
compiler.
In addition, introduce a new cast kind for loading an ObjCProperty l-value,
and make sure we emit those nodes whenever an ObjCProperty l-value appears
that's not on the LHS of an assignment operator.
This breaks a couple of rewriter tests, which I've x-failed until future
development occurs on the rewriter.
Ted Kremenek kindly contributed the analyzer workarounds in this patch.
llvm-svn: 120890
diff --git a/clang/lib/AST/ExprClassification.cpp b/clang/lib/AST/ExprClassification.cpp
index e55545e..1afc760 100644
--- a/clang/lib/AST/ExprClassification.cpp
+++ b/clang/lib/AST/ExprClassification.cpp
@@ -413,8 +413,10 @@
assert(Ctx.getLangOptions().CPlusPlus &&
"This is only relevant for C++.");
// C++ [expr.ass]p1: All [...] return an lvalue referring to the left operand.
+ // Except we override this for writes to ObjC properties.
if (E->isAssignmentOp())
- return Cl::CL_LValue;
+ return (E->getLHS()->getObjectKind() == OK_ObjCProperty
+ ? Cl::CL_PRValue : Cl::CL_LValue);
// C++ [expr.comma]p1: the result is of the same value category as its right
// operand, [...].
@@ -468,9 +470,10 @@
if (Kind == Cl::CL_PRValue) {
// For the sake of better diagnostics, we want to specifically recognize
// use of the GCC cast-as-lvalue extension.
- if (const CStyleCastExpr *CE = dyn_cast<CStyleCastExpr>(E->IgnoreParens())){
- if (CE->getSubExpr()->Classify(Ctx).isLValue()) {
- Loc = CE->getLParenLoc();
+ if (const ExplicitCastExpr *CE =
+ dyn_cast<ExplicitCastExpr>(E->IgnoreParens())) {
+ if (CE->getSubExpr()->IgnoreParenImpCasts()->isLValue()) {
+ Loc = CE->getExprLoc();
return Cl::CM_LValueCast;
}
}