zhuhan0 added a comment. In D141824#4250011 <https://reviews.llvm.org/D141824#4250011>, @sgraenitz wrote:
> In D141824#4247891 <https://reviews.llvm.org/D141824#4247891>, @zhuhan0 wrote: > >> This seems to be the first clang test that involves actual linking. > > Interesting, I think we all didn't have this on our radars. > >> If the intention is to test loading a dynamic library, could you commit the >> library in the Inputs directory instead of building it from source? cc >> @ayermolo > > Is that something other tests do? Otherwise, I am not sure it's a practical > solution. Yes. You can search for .so files under various /test directories. See this <https://github.com/llvm/llvm-project/blob/main/clang/test/Driver/hip-link-shared-library.hip> and this <https://github.com/llvm/llvm-project/blob/main/lld/test/ELF/corrupted-version-reference.s> as some examples. Admittedly you lose some readability if you replace the cpp source input with a compiled library, but you can simply add the source cpp as comments in the test file. After all if the intention is to test loading a shared library, I think it's better to reduce it to just doing that, so you avoid any flakiness that might come from the compilation. > We could also consider another requirement like `host-supports-linking` for > the test. What do you think? I don't think this will work. Our platform supports linking. It's just the C startup/finish libraries are at a different path. It might be overly complicated to plumb that down to the test through cmake and llvm-lit. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D141824/new/ https://reviews.llvm.org/D141824 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits