Yes, indeed. The code in clang/lib/Driver/ToolChains/Arch/X86.cpp makes that –march is always parsed, leaving out /arch unused, no matter in which order they appear.
De : Nico Weber <[email protected]> Envoyé : 10 septembre 2018 12:14 À : [email protected]; Senthil Kumar Selvaraj via Phabricator <[email protected]> Cc : Alexandre Ganea <[email protected]>; cfe-commits <[email protected]> Objet : Re: [PATCH] D51806: [clang-cl] Enable -march On Mon, Sep 10, 2018 at 10:33 AM Hans Wennborg via Phabricator via cfe-commits <[email protected]<mailto:[email protected]>> wrote: hans added a comment. In https://reviews.llvm.org/D51806#1228893, @aganea wrote: > @hans Just an after thought: maybe we should prevent usage of `-march=` and > `/arch:` at the same time. What do you think? I can add another patch for > that purpose. Hmm, yes, at least we should warn or do something smart. Currently it doesn't look like they'd interact well together in x86::getX86TargetCPU Wouldn't you get an "unused arg" for /arch if you use -march and /arch at the same time? Repository: rC Clang https://reviews.llvm.org/D51806 _______________________________________________ cfe-commits mailing list [email protected]<mailto:[email protected]> http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
_______________________________________________ cfe-commits mailing list [email protected] http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
