[PATCH] D112492: [CUDA][HIP] Allow comdat for kernels

Wed Nov 10 10:25:20 PST 2021

rnk added a comment.

I think the key is the self-reference in the LEA instruction:

>   ; foo<int>
>   .seh_proc "??$foo at H@@YAXH at Z"
>   ...
>           leaq    "??$foo at H@@YAXH at Z"(%rip), %rcx
>   ...
>   ; foo<float>
>   .seh_proc "??$foo at M@@YAXM at Z"
>   ...
>           leaq    "??$foo at M@@YAXM at Z"(%rip), %rcx
>
> I think they are not folded because link.exe is smart enough to treat them as not identical comdat functions. I think we may stop worrying about the ICF foading kernel stubs.

It sounds like the behavior may have changed since D63277 <https://reviews.llvm.org/D63277> was landed, and maybe we don't need it anymore.

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D112492/new/

https://reviews.llvm.org/D112492