Ben Shi via llvm-dev
2021-Jan-23 15:36 UTC
[llvm-dev] A question about failures on build machines
Hello, I have made a patch to the avr backend and it passed code review. But the patch failed on two build machines. The reason is that my patch is about improving gnu avr-ld command line options but the avr-ld is not installed on the two machines. The detailed info locates at https://reviews.llvm.org/harbormaster/unit/view/245093/ https://reviews.llvm.org/harbormaster/unit/view/245092/ So can I commit my patch? Or is it possible to install correspond dependancy on the build machines? Thank you. Ben -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20210123/bca1a21f/attachment.html>
Fāng-ruì Sòng via llvm-dev
2021-Jan-23 18:11 UTC
[llvm-dev] A question about failures on build machines
On Sat, Jan 23, 2021 at 7:37 AM Ben Shi via llvm-dev <llvm-dev at lists.llvm.org> wrote:> > Hello, > > I have made a patch to the avr backend and it passed code review. But the patch failed on two build machines. > > The reason is that my patch is about improving gnu avr-ld command line options but the avr-ld is not installed on the two machines. > > The detailed info locates at > https://reviews.llvm.org/harbormaster/unit/view/245093/ > https://reviews.llvm.org/harbormaster/unit/view/245092/ > > So can I commit my patch? Or is it possible to install correspond dependancy on the build machines? > > Thank you. > > BenNo. Clang may check whether the linker exists. By default it may detect some system directories. Driver tests usually specify --sysroot to point to a mock sysroot tree (test/Driver/Inputs/....). You can fix the failures without installing avr-ld.