hliao added a comment. In D62603#1521503 <https://reviews.llvm.org/D62603#1521503>, @tra wrote:
> In D62603#1521484 <https://reviews.llvm.org/D62603#1521484>, @hliao wrote: > > > thanks, but that `static __device__` variable won't have shadow in host > > anymore. > > > Why not? Your change only changes whether `externally_initialized` is applied > to the variable during device-side compilation. It does not change what > happens on the host side. > AFAICT, it will still be generated on the host side and the host side should > still be able to take its address. > NVCC also allows that: https://godbolt.org/z/t78RvM In D62603#1521507 <https://reviews.llvm.org/D62603#1521507>, @tra wrote: > Note for the future -- it would be great if we could finish discussing the > patch before landing it. > I would still like to see the host-side test. Sorry, will follow that rule. Yes, that patch only changes the device-side. But, for host-side, even that variable is declared as `static` as well, but there's no reference to it. clang just skip emitting it. Repository: rC Clang CHANGES SINCE LAST ACTION https://reviews.llvm.org/D62603/new/ https://reviews.llvm.org/D62603 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits