As noted in PR45061, a two-stage ThinLTO build fails the clang/test/Driver/hurd.c test because of a static_cast here which wrongly converts an instance of clang::driver::toolchains::Hurd into that of clang::driver::toolchains::Linux. ThinLTO will later devirtualize the ToolChain.getDynamicLinker(Args); call and use Linux::getDynamicLinker() instead, causing the test to generate a wrong -dynamic-linker linker flag (/lib/ld-linux.so.2 instead of /lib/ld.so)
Ideally, and because of the static_cast mentioned above, maybe gnutools::Linker should take a const Generic_ELF & instance, not a const ToolChain &. But that causes other issues and could be done later.