ro added a comment. In D130688#3686512 <https://reviews.llvm.org/D130688#3686512>, @MaskRay wrote:
> Making the behavior different for different Linux distributions is not great, > but if it matches the practice, I think it is fine. It's the best we can do in `clang`, I fear, due to the compiler's desire to decide as much as possible at runtime. GCC is way more static here, and whoever builds it can decide on the default CPU with `--with-cpu=v9` e.g. at configure time. > I don't know Sparc enough to suggest anything for 32-bit, though... An alternative would be to just check the host CPU with `getHostCPUName` as updated by D130272 <https://reviews.llvm.org/D130272>. That CPU name then needs to be checked if it supports the V9 ISA. That could be done using `getCPUGeneration(CPU) == CG_V9` in `clang/lib/Basic/Targets/Sparc.h`. However, `SparcTargetInfo` (or `TargetInfo` in general) is not currently used inside the driver code. Of course, this only works in the native case, but that's true for distro checks, too. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D130688/new/ https://reviews.llvm.org/D130688 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits