On Thu, 2 Jul 2026 15:41:38 GMT, Christian Hagedorn <[email protected]> wrote:
>> Ah, I see the reason here, inside `LibraryCallKit::load_klass_from_mirror`, >> we null-check the result we get from the load. This means that if we have a >> constant `int.class`, the load will be a constant `null`, and the return >> value is `top`. But this also means that `klass_node` has already been >> null-checked, so the null-check we do right below seems redundant. > > That's a good observation and also seems to be redundant in mainline. We > could follow-up with an RFE there, what do you think? Yes, that's good to me. I was confused because I thought the function results in different runtime value when the input is a constant vs when it is not (`top` vs `null`). ------------- PR Review Comment: https://git.openjdk.org/valhalla/pull/2569#discussion_r3514491305
