[clang] [Driver] Add support for GCC installation detection in Baremetal toolchain (PR #121829)

Petr Hosek via cfe-commits cfe-commits at lists.llvm.org
Wed Jun 18 11:48:27 PDT 2025


petrhosek wrote:

> > > Do the fuchsia builders actually produce output on a failure? We seem to just get exit code 1, but no indication of what filecheck had problems with.
> > 
> > 
> > They do; here's a link! Let me know if you have an problems with the Web UI; I can paste things here if so. https://luci-milo.appspot.com/ui/p/fuchsia/builders/toolchain.ci/clang-linux-x64/b8711783358287375697/overview
> 
> The issue doesn't just exist with expected string not found in standard input. Fuchsia sets CLANG_DEFAULT_UNWINDLIB to libunwind. As a result, when rtlib is set to libgcc and unwindlib is not explicitly specified, tests using Fuchsia as the default platform will error out with message `clang: error: --rtlib=libgcc requires --unwindlib=libgcc`. This would either require tests to be xfailed or marked unsupported for fuchsia. I will include the fix for this in the reland patch itself.

This issue is not specific to Fuchsia as a target, these builders build the Fuchsia Clang toolchain which is simply a distribution of Clang akin to other distributions such as Arm LLVM Toolchain. Using `XFAIL` isn't a valid solution since these tests aren't failing on Fuchsia. Rather, the tests needs to be written in a way that correctly handles different settings of `CLANG_DEFAULT_RTLIB` and `CLANG_DEFAULT_UNWINDLIB` which we've already done in different tests, typically by explicitly specifying `--rtlib=platform` and `--unwindlib=platform`.

https://github.com/llvm/llvm-project/pull/121829


More information about the cfe-commits mailing list