Simple test:
int a() {
int i; short j;
float* k = 1 ? &i : &j;
}
This crashes in codegen because clang warns, and then neglects to
actually make the types match.
Patch attached. As noted, I deciced to coerce both pointers to void*
because that's what gcc does.
-Eli
Index: Sema/SemaExpr.cpp
===================================================================
--- Sema/SemaExpr.cpp (revision 46389)
+++ Sema/SemaExpr.cpp (working copy)
@@ -823,7 +833,13 @@
Diag(questionLoc, diag::ext_typecheck_cond_incompatible_pointers,
lexT.getAsString(), rexT.getAsString(),
lex->getSourceRange(), rex->getSourceRange());
- return lexT; // FIXME: this is an _ext - is this return o.k?
+ // In this situation, we assume void* type. No especially good
+ // reason, but this is what gcc does, and we do have to pick
+ // to get a consistent AST.
+ QualType voidPtrTy = Context.getPointerType(Context.VoidTy);
+ ImpCastExprToType(lex, voidPtrTy);
+ ImpCastExprToType(rex, voidPtrTy);
+ return voidPtrTy;
}
// The pointer types are compatible.
// C99 6.5.15p6: If both operands are pointers to compatible types *or*
to
_______________________________________________
cfe-dev mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev
