[PATCH] D73985: [bpf] zero extension is required in BPF implementaiton so remove <<=32 >>=32

[Yonghong Song via Phabricator via llvm-commits]

yonghong-song added a comment.

In D73985#2054790 <https://reviews.llvm.org/D73985#2054790>, @jrfastab wrote:

> @yonghong-song I can merge the fix with this patch, but why do we eliminate MOV_rr? I'm trying to see where/why this case would happen I'm not seeing other backends with similar logic. Could we just remove the entire block for all cases?


No particular reason. Just want to be cautiously. I agree that MOV_rr should be really unlikely generated by the compiler, but with 32bit subregister, not 100% sure. Could you check with kernel selftest bpf programs? If any program hits MOV_rr, I suggest to keep it. Otherwise, we can remove it.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D73985/new/

https://reviews.llvm.org/D73985