rjmccall added inline comments.
================ Comment at: clang/lib/CodeGen/CGCall.cpp:4091 + V = Builder.CreatePointerBitCastOrAddrSpaceCast( + V, IRFuncTy->getParamType(FirstIRArg)); + else ---------------- Anastasia wrote: > svenvh wrote: > > @Anastasia pointed out that the AST might not be correct to begin with, > > which could cause us to fail here. > > > > ``` > > | `-MemberExpr 0x5555558af2c0 <col:3, col:5> '<bound member function > > type>' .getRef 0x5555558a4008 > > | `-ImplicitCastExpr 0x5555558af310 <col:3> '__generic B2' lvalue > > <UncheckedDerivedToBase (B2)> > > | `-DeclRefExpr 0x5555558af2a0 <col:3> 'Derived' lvalue Var > > 0x5555558ab290 'd' 'Derived' > > ``` > > > > The question is whether it's fine for the `UncheckedDerivedToBase > > ImplicitCastExpr` to cast from `private` to `__generic`; or do we need > > another `AddrSpaceConversion ImplicitCastExpr` node perhaps? > Yes, I am not sure AST is correct wrt addr spaces. > > Also we should at least use `performAddrSpaceCast` that isn't easy here > because the source addr spaces are not available at this point. > > But overall I would like @rjmccall to have a look. I agree that the AST should include an address-space conversion, not just a derived-to-base conversion. Repository: rC Clang CHANGES SINCE LAST ACTION https://reviews.llvm.org/D69810/new/ https://reviews.llvm.org/D69810 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
