dsanders added subscribers: rengolin, joerg.
dsanders added a comment.

+Renato and Joerg

I was going to say I think it's ok and the optimizer should be smart enough to 
factor out the common IsMips check but I've just realized there may be a better 
way. The current code is using an else after an (implicit) continue. If we made 
that continue explicit, we could make this code a bit neater and have a place 
to add target specific options.

I'm thinking something like:

  for (...) {
    ...
  
    auto Arch = C.getDefaultToolChain().getArch();
  
    if (C.getDefaultToolChain().getArch() == llvm::Triple::mips ||
        C.getDefaultToolChain().getArch() == llvm::Triple::mipsel ||
        C.getDefaultToolChain().getArch() == llvm::Triple::mips64 ||
        C.getDefaultToolChain().getArch() == llvm::Triple::mips64el)
      if (mips::CollectArgsForIntegratedAssembler(...)
        continue;
  
    if (Value == "-force_cpusubtype_ALL")
      continue;
  
    ...
  
    D.Diag(diag::err_drv_unsupported_option_argument)
        << A->getOption().getName() << Value;
  }

Thoughts?


http://reviews.llvm.org/D13100



_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to