benshi001 added a comment. In D122524#3411879 <https://reviews.llvm.org/D122524#3411879>, @MaskRay wrote:
> In D122524#3411872 <https://reviews.llvm.org/D122524#3411872>, @benshi001 > wrote: > >> In D122524#3410542 <https://reviews.llvm.org/D122524#3410542>, @MaskRay >> wrote: >> >>> I think it is excessive to add so many RUN lines. I do not understand much >>> about AVR -mcpu. That said, I created D122553 >>> <https://reviews.llvm.org/D122553> for what I think should be done for the >>> `-c/-S/-fsyntax-only` condition. More tests would just be excessive. >>> >>> You may adjust the patch to do the rest cleanups/fixes. >> >> I have made some changes based on your https://reviews.llvm.org/D122553, it >> seems you have reverted. I suggest you recommit, since I have fixed the >> failures of lacking avr-gcc. > > I did not specify --sysroot so the new tests failed on systems without > avr-gcc. Relanded. I have made some changes based on your code: 1. We should always warn if there is no mmcu specified, since its info (instruction set version) is used by the llvm backend. 2. We need no warn no avr-gcc&avr-libc if `-S`/`-c` is specified, since the compiling stage need neither avr-gcc nor avr-libc. Thanks for your way to avoid checking so many `-c`/`-S`/`-fsyntax-only`/... conditions. CHANGES SINCE LAST ACTION https://reviews.llvm.org/D122524/new/ https://reviews.llvm.org/D122524 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
