Is dso_local set in LLVM IR output?

[Ray Song]

(I've posted very similar questions to 

Julia: https://discourse.julialang.org/t/is-dso-local-set-in-llvm-ir-output/52707

Rust: https://internals.rust-lang.org/t/is-dso-local-set-in-llvm-ir-output/13725

)

https://github.com/llvm/llvm-project/commit/2047c10c22b071cccc57a7e2779d6603512e9113 was to clean up `llvm::TargetMachine::shouldAssumeDSOLocal`. For Clang, that code is unneeded because Clang sets dso_local https://github.com/llvm/llvm-project/blob/main/clang/lib/CodeGen/CodeGenModule.cpp#L910

I am thinking whether https://github.com/llvm/llvm-project/commit/2047c10c22b071cccc57a7e2779d6603512e9113 should be reverted (or be replaced with `if (RM == Reloc::Static && !GV->isDeclarationForLinker()) return true`) as a workaround before IR producers set dso_local properly.

**But for Numba, it'd be nice to explicitly set dso_local like Clang.**

[Valentin Haenel]

Hi Ray,

thank you for raising this issue. I believe this mailinglist is deprecated and I would suggest raising this issue for discussion on Discourse:

https://numba.discourse.group/

Best,

V-

--
You received this message because you are subscribed to the Google Groups "Numba Public Discussion - Public" group.
To unsubscribe from this group and stop receiving emails from it, send an email to numba-users...@continuum.io.
To view this discussion on the web visit https://groups.google.com/a/continuum.io/d/msgid/numba-users/330db794-d768-49ae-948f-0a778fc0cce5n%40continuum.io.